{'_name': None, 'common': {'_name': None, 'no_progress_bar': False, 'log_interval': 100, 'log_format': 'simple', 'log_file': 'chkpt/ja-en.do02.ado00/train.log', 'aim_repo': None, 'aim_run_hash': None, 'tensorboard_logdir': None, 'wandb_project': 'wmt23', 'azureml_logging': False, 'seed': 0, 'cpu': False, 'tpu': False, 'bf16': False, 'memory_efficient_bf16': False, 'fp16': True, 'memory_efficient_fp16': False, 'fp16_no_flatten_grads': False, 'fp16_init_scale': 8, 'fp16_scale_window': None, 'fp16_scale_tolerance': 0.0, 'on_cpu_convert_precision': False, 'min_loss_scale': 0.0001, 'threshold_loss_scale': None, 'amp': False, 'amp_batch_retries': 2, 'amp_init_scale': 128, 'amp_scale_window': None, 'user_dir': None, 'empty_cache_freq': 0, 'all_gather_list_size': 16384, 'model_parallel_size': 1, 'quantization_config_path': None, 'profile': False, 'reset_logging': False, 'suppress_crashes': False, 'use_plasma_view': False, 'plasma_path': '/tmp/plasma'}, 'common_eval': {'_name': None, 'path': None, 'post_process': None, 'quiet': False, 'model_overrides': '{}', 'results_path': None}, 'distributed_training': {'_name': None, 'distributed_world_size': 8, 'distributed_num_procs': 8, 'distributed_rank': 0, 'distributed_backend': 'nccl', 'distributed_init_method': 'tcp://localhost:54807', 'distributed_port': 54807, 'device_id': 0, 'distributed_no_spawn': False, 'ddp_backend': 'pytorch_ddp', 'ddp_comm_hook': 'none', 'bucket_cap_mb': 25, 'fix_batches_to_gpus': False, 'find_unused_parameters': False, 'gradient_as_bucket_view': False, 'fast_stat_sync': False, 'heartbeat_timeout': -1, 'broadcast_buffers': False, 'slowmo_momentum': None, 'slowmo_base_algorithm': 'localsgd', 'localsgd_frequency': 3, 'nprocs_per_node': 8, 'pipeline_model_parallel': False, 'pipeline_balance': None, 'pipeline_devices': None, 'pipeline_chunks': 0, 'pipeline_encoder_balance': None, 'pipeline_encoder_devices': None, 'pipeline_decoder_balance': None, 'pipeline_decoder_devices': None, 'pipeline_checkpoint': 'never', 'zero_sharding': 'none', 'fp16': True, 'memory_efficient_fp16': False, 'tpu': False, 'no_reshard_after_forward': False, 'fp32_reduce_scatter': False, 'cpu_offload': False, 'use_sharded_state': False, 'not_fsdp_flatten_parameters': False}, 'dataset': {'_name': None, 'num_workers': 0, 'skip_invalid_size_inputs_valid_test': False, 'max_tokens': 16384, 'batch_size': None, 'required_batch_size_multiple': 8, 'required_seq_len_multiple': 1, 'dataset_impl': None, 'data_buffer_size': 10, 'train_subset': 'train', 'valid_subset': 'valid', 'combine_valid_subsets': None, 'ignore_unused_valid_subsets': False, 'validate_interval': 100000, 'validate_interval_updates': 0, 'validate_after_updates': 0, 'fixed_validation_seed': None, 'disable_validation': False, 'max_tokens_valid': 16384, 'batch_size_valid': None, 'max_valid_steps': None, 'curriculum': 0, 'gen_subset': 'test', 'num_shards': 1, 'shard_id': 0, 'grouped_shuffling': False, 'update_epoch_batch_itr': False, 'update_ordered_indices_seed': False}, 'optimization': {'_name': None, 'max_epoch': 0, 'max_update': 60000, 'stop_time_hours': 0.0, 'clip_norm': 1.0, 'sentence_avg': False, 'update_freq': [4], 'lr': [0.001], 'stop_min_lr': -1.0, 'use_bmuf': False, 'skip_remainder_batch': False, 'debug_param_names': False}, 'checkpoint': {'_name': None, 'save_dir': 'chkpt/ja-en.do02.ado00', 'restore_file': 'checkpoint_last.pt', 'continue_once': None, 'finetune_from_model': None, 'reset_dataloader': False, 'reset_lr_scheduler': False, 'reset_meters': False, 'reset_optimizer': False, 'optimizer_overrides': '{}', 'save_interval': 100000, 'save_interval_updates': 1000, 'keep_interval_updates': 10, 'keep_interval_updates_pattern': -1, 'keep_last_epochs': -1, 'keep_best_checkpoints': -1, 'no_save': False, 'no_epoch_checkpoints': True, 'no_last_checkpoints': False, 'no_save_optimizer_state': False, 'best_checkpoint_metric': 'loss', 'maximize_best_checkpoint_metric': False, 'patience': -1, 'checkpoint_suffix': '', 'checkpoint_shard_count': 1, 'load_checkpoint_on_all_dp_ranks': False, 'write_checkpoints_asynchronously': False, 'model_parallel_size': 1}, 'bmuf': {'_name': None, 'block_lr': 1.0, 'block_momentum': 0.875, 'global_sync_iter': 50, 'warmup_iterations': 500, 'use_nbm': False, 'average_sync': False, 'distributed_world_size': 8}, 'generation': {'_name': None, 'beam': 5, 'beam_mt': 0, 'nbest': 1, 'max_len_a': 0.0, 'max_len_b': 200, 'max_len_a_mt': 0.0, 'max_len_b_mt': 200, 'min_len': 1, 'match_source_len': False, 'unnormalized': False, 'no_early_stop': False, 'no_beamable_mm': False, 'lenpen': 1.0, 'lenpen_mt': 1.0, 'unkpen': 0.0, 'replace_unk': None, 'sacrebleu': False, 'score_reference': False, 'prefix_size': 0, 'no_repeat_ngram_size': 0, 'sampling': False, 'sampling_topk': -1, 'sampling_topp': -1.0, 'constraints': None, 'temperature': 1.0, 'diverse_beam_groups': -1, 'diverse_beam_strength': 0.5, 'diversity_rate': -1.0, 'print_alignment': None, 'print_step': False, 'lm_path': None, 'lm_weight': 0.0, 'iter_decode_eos_penalty': 0.0, 'iter_decode_max_iter': 10, 'iter_decode_force_max_iter': False, 'iter_decode_with_beam': 1, 'iter_decode_with_external_reranker': False, 'retain_iter_history': False, 'retain_dropout': False, 'retain_dropout_modules': None, 'decoding_format': None, 'no_seed_provided': False, 'eos_token': None}, 'eval_lm': {'_name': None, 'output_word_probs': False, 'output_word_stats': False, 'context_window': 0, 'softmax_batch': 9223372036854775807}, 'interactive': {'_name': None, 'buffer_size': 0, 'input': '-'}, 'model': Namespace(no_progress_bar=False, log_interval=100, log_format='simple', log_file='chkpt/ja-en.do02.ado00/train.log', aim_repo=None, aim_run_hash=None, tensorboard_logdir=None, wandb_project='wmt23', azureml_logging=False, seed=0, cpu=False, tpu=False, bf16=False, memory_efficient_bf16=False, fp16=True, memory_efficient_fp16=False, fp16_no_flatten_grads=False, fp16_init_scale=8, fp16_scale_window=None, fp16_scale_tolerance=0.0, on_cpu_convert_precision=False, min_loss_scale=0.0001, threshold_loss_scale=None, amp=False, amp_batch_retries=2, amp_init_scale=128, amp_scale_window=None, user_dir=None, empty_cache_freq=0, all_gather_list_size=16384, model_parallel_size=1, quantization_config_path=None, profile=False, reset_logging=False, suppress_crashes=False, use_plasma_view=False, plasma_path='/tmp/plasma', criterion='label_smoothed_cross_entropy', tokenizer=None, bpe=None, optimizer='adam', lr_scheduler='inverse_sqrt', scoring='bleu', task='translation', num_workers=0, skip_invalid_size_inputs_valid_test=False, max_tokens=16384, batch_size=None, required_batch_size_multiple=8, required_seq_len_multiple=1, dataset_impl=None, data_buffer_size=10, train_subset='train', valid_subset='valid', combine_valid_subsets=None, ignore_unused_valid_subsets=False, validate_interval=100000, validate_interval_updates=0, validate_after_updates=0, fixed_validation_seed=None, disable_validation=False, max_tokens_valid=16384, batch_size_valid=None, max_valid_steps=None, curriculum=0, gen_subset='test', num_shards=1, shard_id=0, grouped_shuffling=False, update_epoch_batch_itr=False, update_ordered_indices_seed=False, distributed_world_size=8, distributed_num_procs=8, distributed_rank=0, distributed_backend='nccl', distributed_init_method=None, distributed_port=-1, device_id=0, distributed_no_spawn=False, ddp_backend='pytorch_ddp', ddp_comm_hook='none', bucket_cap_mb=25, fix_batches_to_gpus=False, find_unused_parameters=False, gradient_as_bucket_view=False, fast_stat_sync=False, heartbeat_timeout=-1, broadcast_buffers=False, slowmo_momentum=None, slowmo_base_algorithm='localsgd', localsgd_frequency=3, nprocs_per_node=8, pipeline_model_parallel=False, pipeline_balance=None, pipeline_devices=None, pipeline_chunks=0, pipeline_encoder_balance=None, pipeline_encoder_devices=None, pipeline_decoder_balance=None, pipeline_decoder_devices=None, pipeline_checkpoint='never', zero_sharding='none', no_reshard_after_forward=False, fp32_reduce_scatter=False, cpu_offload=False, use_sharded_state=False, not_fsdp_flatten_parameters=False, arch='transformer_vaswani_wmt_en_de_big', max_epoch=0, max_update=60000, stop_time_hours=0, clip_norm=1.0, sentence_avg=False, update_freq=[4], lr=[0.001], stop_min_lr=-1.0, use_bmuf=False, skip_remainder_batch=False, debug_param_names=False, save_dir='chkpt/ja-en.do02.ado00', restore_file='checkpoint_last.pt', continue_once=None, finetune_from_model=None, reset_dataloader=False, reset_lr_scheduler=False, reset_meters=False, reset_optimizer=False, optimizer_overrides='{}', save_interval=100000, save_interval_updates=1000, keep_interval_updates=10, keep_interval_updates_pattern=-1, keep_last_epochs=-1, keep_best_checkpoints=-1, no_save=False, no_epoch_checkpoints=True, no_last_checkpoints=False, no_save_optimizer_state=False, best_checkpoint_metric='loss', maximize_best_checkpoint_metric=False, patience=-1, checkpoint_suffix='', checkpoint_shard_count=1, load_checkpoint_on_all_dp_ranks=False, write_checkpoints_asynchronously=False, store_ema=False, ema_decay=0.9999, ema_start_update=0, ema_seed_model=None, ema_update_freq=1, ema_fp32=False, data='binarized/ja-en/', source_lang=None, target_lang=None, load_alignments=False, left_pad_source=True, left_pad_target=False, upsample_primary=-1, truncate_source=False, num_batch_buckets=0, eval_bleu=False, eval_bleu_args='{}', eval_bleu_detok='space', eval_bleu_detok_args='{}', eval_tokenized_bleu=False, eval_bleu_remove_bpe=None, eval_bleu_print_samples=False, label_smoothing=0.1, report_accuracy=False, ignore_prefix_size=0, adam_betas='(0.9, 0.98)', adam_eps=1e-08, weight_decay=0.0, use_old_adam=False, fp16_adam_stats=False, warmup_updates=4000, warmup_init_lr=-1, pad=1, eos=2, unk=3, encoder_ffn_embed_dim=8192, decoder_ffn_embed_dim=8192, dropout=0.2, attention_dropout=0.0, share_decoder_input_output_embed=True, no_seed_provided=False, encoder_embed_dim=1024, encoder_attention_heads=16, encoder_normalize_before=False, decoder_embed_dim=1024, decoder_attention_heads=16, encoder_embed_path=None, encoder_layers=6, encoder_learned_pos=False, decoder_embed_path=None, decoder_layers=6, decoder_normalize_before=False, decoder_learned_pos=False, activation_dropout=0.0, activation_fn='relu', adaptive_softmax_cutoff=None, adaptive_softmax_dropout=0, share_all_embeddings=False, merge_src_tgt_embed=False, no_token_positional_embeddings=False, adaptive_input=False, no_cross_attention=False, cross_self_attention=False, decoder_output_dim=1024, decoder_input_dim=1024, no_scale_embedding=False, layernorm_embedding=False, tie_adaptive_weights=False, checkpoint_activations=False, offload_activations=False, encoder_layers_to_keep=None, decoder_layers_to_keep=None, encoder_layerdrop=0, decoder_layerdrop=0, quant_noise_pq=0, quant_noise_pq_block_size=8, quant_noise_scalar=0, _name='transformer_vaswani_wmt_en_de_big'), 'task': {'_name': 'translation', 'data': 'binarized/ja-en/', 'source_lang': None, 'target_lang': None, 'load_alignments': False, 'left_pad_source': True, 'left_pad_target': False, 'max_source_positions': 1024, 'max_target_positions': 1024, 'upsample_primary': -1, 'truncate_source': False, 'num_batch_buckets': 0, 'train_subset': 'train', 'dataset_impl': None, 'required_seq_len_multiple': 1, 'eval_bleu': False, 'eval_bleu_args': '{}', 'eval_bleu_detok': 'space', 'eval_bleu_detok_args': '{}', 'eval_tokenized_bleu': False, 'eval_bleu_remove_bpe': None, 'eval_bleu_print_samples': False}, 'criterion': {'_name': 'label_smoothed_cross_entropy', 'label_smoothing': 0.1, 'report_accuracy': False, 'ignore_prefix_size': 0, 'sentence_avg': False}, 'optimizer': {'_name': 'adam', 'adam_betas': '(0.9, 0.98)', 'adam_eps': 1e-08, 'weight_decay': 0.0, 'use_old_adam': False, 'fp16_adam_stats': False, 'tpu': False, 'lr': [0.001]}, 'lr_scheduler': {'_name': 'inverse_sqrt', 'warmup_updates': 4000, 'warmup_init_lr': -1.0, 'lr': [0.001]}, 'scoring': {'_name': 'bleu', 'pad': 1, 'eos': 2, 'unk': 3}, 'bpe': None, 'tokenizer': None, 'ema': {'_name': None, 'store_ema': False, 'ema_decay': 0.9999, 'ema_start_update': 0, 'ema_seed_model': None, 'ema_update_freq': 1, 'ema_fp32': False}} TransformerModel( (encoder): TransformerEncoderBase( (dropout_module): FairseqDropout() (embed_tokens): Embedding(32000, 1024, padding_idx=1) (embed_positions): SinusoidalPositionalEmbedding() (layers): ModuleList( (0-5): 6 x TransformerEncoderLayerBase( (self_attn): MultiheadAttention( (dropout_module): FairseqDropout() (k_proj): Linear(in_features=1024, out_features=1024, bias=True) (v_proj): Linear(in_features=1024, out_features=1024, bias=True) (q_proj): Linear(in_features=1024, out_features=1024, bias=True) (out_proj): Linear(in_features=1024, out_features=1024, bias=True) ) (self_attn_layer_norm): FusedLayerNorm(torch.Size([1024]), eps=1e-05, elementwise_affine=True) (dropout_module): FairseqDropout() (activation_dropout_module): FairseqDropout() (fc1): Linear(in_features=1024, out_features=8192, bias=True) (fc2): Linear(in_features=8192, out_features=1024, bias=True) (final_layer_norm): FusedLayerNorm(torch.Size([1024]), eps=1e-05, elementwise_affine=True) ) ) ) (decoder): TransformerDecoderBase( (dropout_module): FairseqDropout() (embed_tokens): Embedding(16000, 1024, padding_idx=1) (embed_positions): SinusoidalPositionalEmbedding() (layers): ModuleList( (0-5): 6 x TransformerDecoderLayerBase( (dropout_module): FairseqDropout() (self_attn): MultiheadAttention( (dropout_module): FairseqDropout() (k_proj): Linear(in_features=1024, out_features=1024, bias=True) (v_proj): Linear(in_features=1024, out_features=1024, bias=True) (q_proj): Linear(in_features=1024, out_features=1024, bias=True) (out_proj): Linear(in_features=1024, out_features=1024, bias=True) ) (activation_dropout_module): FairseqDropout() (self_attn_layer_norm): FusedLayerNorm(torch.Size([1024]), eps=1e-05, elementwise_affine=True) (encoder_attn): MultiheadAttention( (dropout_module): FairseqDropout() (k_proj): Linear(in_features=1024, out_features=1024, bias=True) (v_proj): Linear(in_features=1024, out_features=1024, bias=True) (q_proj): Linear(in_features=1024, out_features=1024, bias=True) (out_proj): Linear(in_features=1024, out_features=1024, bias=True) ) (encoder_attn_layer_norm): FusedLayerNorm(torch.Size([1024]), eps=1e-05, elementwise_affine=True) (fc1): Linear(in_features=1024, out_features=8192, bias=True) (fc2): Linear(in_features=8192, out_features=1024, bias=True) (final_layer_norm): FusedLayerNorm(torch.Size([1024]), eps=1e-05, elementwise_affine=True) ) ) (output_projection): Linear(in_features=1024, out_features=16000, bias=False) ) ) task: TranslationTask model: TransformerModel criterion: LabelSmoothedCrossEntropyCriterion num. shared model params: 326,221,824 (num. trained: 326,221,824) num. expert model params: 0 (num. trained: 0) training on 8 devices (GPUs/TPUs) max tokens per device = 16384 and max sentences per device = None begin dry-run validation on "valid" subset Start iterating over samples epoch 001: 101 / 1689 loss=12.122, nll_loss=11.813, ppl=3596.84, wps=552847, ups=1.12, wpb=495063, bsz=16556.9, num_updates=100, lr=2.5e-05, gnorm=2.582, clip=76, loss_scale=4, train_wall=94, gb_free=21.5, wall=112 epoch 001: 201 / 1689 loss=10.54, nll_loss=9.999, ppl=1023.12, wps=549069, ups=1.11, wpb=494772, bsz=16958.6, num_updates=200, lr=5e-05, gnorm=1.812, clip=97, loss_scale=4, train_wall=89, gb_free=21.6, wall=202 epoch 001: 301 / 1689 loss=9.851, nll_loss=9.181, ppl=580.39, wps=554831, ups=1.12, wpb=496328, bsz=16644.9, num_updates=300, lr=7.5e-05, gnorm=2.011, clip=100, loss_scale=4, train_wall=89, gb_free=21.8, wall=292 epoch 001: 401 / 1689 loss=9.247, nll_loss=8.468, ppl=354.17, wps=550349, ups=1.11, wpb=495021, bsz=16565.7, num_updates=400, lr=0.0001, gnorm=1.85, clip=100, loss_scale=4, train_wall=89, gb_free=22, wall=382 epoch 001: 501 / 1689 loss=8.807, nll_loss=7.945, ppl=246.44, wps=549334, ups=1.11, wpb=495038, bsz=16610.6, num_updates=500, lr=0.000125, gnorm=1.706, clip=99, loss_scale=4, train_wall=89, gb_free=21.9, wall=472 epoch 001: 601 / 1689 loss=8.437, nll_loss=7.51, ppl=182.26, wps=548131, ups=1.11, wpb=495492, bsz=16496.7, num_updates=600, lr=0.00015, gnorm=1.507, clip=99, loss_scale=8, train_wall=89, gb_free=19.9, wall=562 epoch 001: 701 / 1689 loss=8.057, nll_loss=7.068, ppl=134.21, wps=546618, ups=1.1, wpb=494852, bsz=16337.6, num_updates=700, lr=0.000175, gnorm=1.413, clip=97, loss_scale=8, train_wall=90, gb_free=22.3, wall=653 epoch 001: 801 / 1689 loss=7.689, nll_loss=6.64, ppl=99.74, wps=552167, ups=1.11, wpb=496124, bsz=16372.8, num_updates=800, lr=0.0002, gnorm=1.302, clip=95, loss_scale=8, train_wall=88, gb_free=21.9, wall=743 epoch 001: 901 / 1689 loss=7.326, nll_loss=6.221, ppl=74.61, wps=548640, ups=1.1, wpb=496813, bsz=16599.5, num_updates=900, lr=0.000225, gnorm=1.177, clip=88, loss_scale=8, train_wall=89, gb_free=21.9, wall=833 epoch 001: 1002 / 1689 loss=6.999, nll_loss=5.843, ppl=57.4, wps=539215, ups=1.09, wpb=496202, bsz=16673.8, num_updates=1000, lr=0.00025, gnorm=1.087, clip=65, loss_scale=4, train_wall=90, gb_free=21.8, wall=925 begin validation on "valid" subset epoch 001 | valid on 'valid' subset | loss 6.755 | nll_loss 5.51 | ppl 45.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 1000 epoch 001: 1102 / 1689 loss=6.669, nll_loss=5.464, ppl=44.13, wps=461275, ups=0.93, wpb=494438, bsz=16503.4, num_updates=1100, lr=0.000275, gnorm=1.051, clip=61, loss_scale=4, train_wall=89, gb_free=20.6, wall=1032 epoch 001: 1202 / 1689 loss=6.334, nll_loss=5.08, ppl=33.82, wps=548509, ups=1.11, wpb=494853, bsz=16237.2, num_updates=1200, lr=0.0003, gnorm=0.975, clip=40, loss_scale=4, train_wall=88, gb_free=22.1, wall=1123 epoch 001: 1302 / 1689 loss=6.016, nll_loss=4.716, ppl=26.28, wps=543986, ups=1.1, wpb=493588, bsz=16373.8, num_updates=1300, lr=0.000325, gnorm=0.933, clip=35, loss_scale=4, train_wall=89, gb_free=21.7, wall=1213 epoch 001: 1402 / 1689 loss=5.725, nll_loss=4.385, ppl=20.89, wps=554227, ups=1.11, wpb=497384, bsz=16279.9, num_updates=1400, lr=0.00035, gnorm=0.833, clip=20, loss_scale=4, train_wall=88, gb_free=22.2, wall=1303 epoch 001: 1502 / 1689 loss=5.479, nll_loss=4.108, ppl=17.24, wps=547918, ups=1.11, wpb=494548, bsz=16575.5, num_updates=1500, lr=0.000375, gnorm=0.734, clip=7, loss_scale=4, train_wall=89, gb_free=22.1, wall=1393 epoch 001: 1603 / 1689 loss=5.304, nll_loss=3.912, ppl=15.05, wps=544766, ups=1.1, wpb=495877, bsz=16534.4, num_updates=1600, lr=0.0004, gnorm=0.698, clip=9, loss_scale=4, train_wall=89, gb_free=21.8, wall=1484 end of epoch 1 (average epoch stats below) epoch 001 | loss 7.655 | nll_loss 6.62 | ppl 98.34 | wps 542518 | ups 1.1 | wpb 495135 | bsz 16503.6 | num_updates 1686 | lr 0.0004215 | gnorm 1.318 | clip 64.8 | loss_scale 4 | train_wall 1505 | gb_free 21.6 | wall 1561 Start iterating over samples epoch 002: 14 / 1689 loss=5.163, nll_loss=3.754, ppl=13.49, wps=541888, ups=1.1, wpb=490777, bsz=16322.4, num_updates=1700, lr=0.000425, gnorm=0.638, clip=4, loss_scale=4, train_wall=88, gb_free=22, wall=1575 epoch 002: 14 / 1689 loss=5.163, nll_loss=3.754, ppl=13.49, wps=541888, ups=1.1, wpb=490777, bsz=16322.4, num_updates=1700, lr=0.000425, gnorm=0.638, clip=4, loss_scale=4, train_wall=88, gb_free=22, wall=1575 epoch 002: 114 / 1689 loss=5.03, nll_loss=3.607, ppl=12.18, wps=550048, ups=1.11, wpb=496688, bsz=16760.9, num_updates=1800, lr=0.00045, gnorm=0.57, clip=1, loss_scale=4, train_wall=89, gb_free=21.8, wall=1665 epoch 002: 114 / 1689 loss=5.03, nll_loss=3.607, ppl=12.18, wps=550048, ups=1.11, wpb=496688, bsz=16760.9, num_updates=1800, lr=0.00045, gnorm=0.57, clip=1, loss_scale=4, train_wall=89, gb_free=21.8, wall=1665 epoch 002: 214 / 1689 loss=4.948, nll_loss=3.517, ppl=11.45, wps=546210, ups=1.1, wpb=494526, bsz=16524.7, num_updates=1900, lr=0.000475, gnorm=0.588, clip=6, loss_scale=4, train_wall=89, gb_free=22.2, wall=1756 epoch 002: 214 / 1689 loss=4.948, nll_loss=3.517, ppl=11.45, wps=546210, ups=1.1, wpb=494526, bsz=16524.7, num_updates=1900, lr=0.000475, gnorm=0.588, clip=6, loss_scale=4, train_wall=89, gb_free=22.2, wall=1756 epoch 002: 314 / 1689 loss=4.844, nll_loss=3.402, ppl=10.57, wps=547195, ups=1.11, wpb=494140, bsz=16833.4, num_updates=2000, lr=0.0005, gnorm=0.465, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=1846 epoch 002: 314 / 1689 loss=4.844, nll_loss=3.402, ppl=10.57, wps=547195, ups=1.11, wpb=494140, bsz=16833.4, num_updates=2000, lr=0.0005, gnorm=0.465, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=1846 begin validation on "valid" subset epoch 002 | valid on 'valid' subset | loss 4.848 | nll_loss 3.333 | ppl 10.08 | wps 0 | wpb 44526 | bsz 2008 | num_updates 2000 | best_loss 4.848 epoch 002 | valid on 'valid' subset | loss 4.848 | nll_loss 3.333 | ppl 10.08 | wps 0 | wpb 44526 | bsz 2008 | num_updates 2000 | best_loss 4.848 epoch 002: 414 / 1689 loss=4.786, nll_loss=3.339, ppl=10.12, wps=455433, ups=0.92, wpb=495220, bsz=16290.6, num_updates=2100, lr=0.000525, gnorm=0.505, clip=5, loss_scale=4, train_wall=88, gb_free=22, wall=1955 epoch 002: 414 / 1689 loss=4.786, nll_loss=3.339, ppl=10.12, wps=455433, ups=0.92, wpb=495220, bsz=16290.6, num_updates=2100, lr=0.000525, gnorm=0.505, clip=5, loss_scale=4, train_wall=88, gb_free=22, wall=1955 epoch 002: 515 / 1689 loss=4.695, nll_loss=3.24, ppl=9.45, wps=545830, ups=1.1, wpb=495254, bsz=16571, num_updates=2200, lr=0.00055, gnorm=0.471, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=2046 epoch 002: 515 / 1689 loss=4.695, nll_loss=3.24, ppl=9.45, wps=545830, ups=1.1, wpb=495254, bsz=16571, num_updates=2200, lr=0.00055, gnorm=0.471, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=2046 epoch 002: 615 / 1689 loss=4.621, nll_loss=3.159, ppl=8.93, wps=551617, ups=1.11, wpb=495521, bsz=16630.4, num_updates=2300, lr=0.000575, gnorm=0.41, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=2135 epoch 002: 615 / 1689 loss=4.621, nll_loss=3.159, ppl=8.93, wps=551617, ups=1.11, wpb=495521, bsz=16630.4, num_updates=2300, lr=0.000575, gnorm=0.41, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=2135 epoch 002: 715 / 1689 loss=4.585, nll_loss=3.12, ppl=8.69, wps=549320, ups=1.11, wpb=495264, bsz=16162, num_updates=2400, lr=0.0006, gnorm=0.439, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=2226 epoch 002: 715 / 1689 loss=4.585, nll_loss=3.12, ppl=8.69, wps=549320, ups=1.11, wpb=495264, bsz=16162, num_updates=2400, lr=0.0006, gnorm=0.439, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=2226 epoch 002: 815 / 1689 loss=4.52, nll_loss=3.05, ppl=8.28, wps=543579, ups=1.1, wpb=494583, bsz=16776.5, num_updates=2500, lr=0.000625, gnorm=0.399, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=2317 epoch 002: 815 / 1689 loss=4.52, nll_loss=3.05, ppl=8.28, wps=543579, ups=1.1, wpb=494583, bsz=16776.5, num_updates=2500, lr=0.000625, gnorm=0.399, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=2317 epoch 002: 915 / 1689 loss=4.487, nll_loss=3.014, ppl=8.08, wps=547996, ups=1.11, wpb=495507, bsz=16601.8, num_updates=2600, lr=0.00065, gnorm=0.411, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=2407 epoch 002: 915 / 1689 loss=4.487, nll_loss=3.014, ppl=8.08, wps=547996, ups=1.11, wpb=495507, bsz=16601.8, num_updates=2600, lr=0.00065, gnorm=0.411, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=2407 epoch 002: 1015 / 1689 loss=4.43, nll_loss=2.952, ppl=7.74, wps=548578, ups=1.11, wpb=495244, bsz=16349.9, num_updates=2700, lr=0.000675, gnorm=0.377, clip=0, loss_scale=8, train_wall=89, gb_free=21.8, wall=2497 epoch 002: 1015 / 1689 loss=4.43, nll_loss=2.952, ppl=7.74, wps=548578, ups=1.11, wpb=495244, bsz=16349.9, num_updates=2700, lr=0.000675, gnorm=0.377, clip=0, loss_scale=8, train_wall=89, gb_free=21.8, wall=2497 epoch 002: 1116 / 1689 loss=4.407, nll_loss=2.927, ppl=7.61, wps=550552, ups=1.11, wpb=495128, bsz=16391, num_updates=2800, lr=0.0007, gnorm=0.39, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=2587 epoch 002: 1116 / 1689 loss=4.407, nll_loss=2.927, ppl=7.61, wps=550552, ups=1.11, wpb=495128, bsz=16391, num_updates=2800, lr=0.0007, gnorm=0.39, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=2587 epoch 002: 1216 / 1689 loss=4.37, nll_loss=2.887, ppl=7.4, wps=550362, ups=1.11, wpb=496100, bsz=16472.5, num_updates=2900, lr=0.000725, gnorm=0.367, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=2677 epoch 002: 1216 / 1689 loss=4.37, nll_loss=2.887, ppl=7.4, wps=550362, ups=1.11, wpb=496100, bsz=16472.5, num_updates=2900, lr=0.000725, gnorm=0.367, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=2677 epoch 002: 1316 / 1689 loss=4.34, nll_loss=2.856, ppl=7.24, wps=550510, ups=1.11, wpb=496192, bsz=16642.6, num_updates=3000, lr=0.00075, gnorm=0.378, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=2767 epoch 002: 1316 / 1689 loss=4.34, nll_loss=2.856, ppl=7.24, wps=550510, ups=1.11, wpb=496192, bsz=16642.6, num_updates=3000, lr=0.00075, gnorm=0.378, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=2767 begin validation on "valid" subset epoch 002 | valid on 'valid' subset | loss 4.327 | nll_loss 2.786 | ppl 6.9 | wps 0 | wpb 44526 | bsz 2008 | num_updates 3000 | best_loss 4.327 epoch 002 | valid on 'valid' subset | loss 4.327 | nll_loss 2.786 | ppl 6.9 | wps 0 | wpb 44526 | bsz 2008 | num_updates 3000 | best_loss 4.327 epoch 002: 1416 / 1689 loss=4.31, nll_loss=2.822, ppl=7.07, wps=458702, ups=0.93, wpb=494703, bsz=16277.2, num_updates=3100, lr=0.000775, gnorm=0.362, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=2875 epoch 002: 1416 / 1689 loss=4.31, nll_loss=2.822, ppl=7.07, wps=458702, ups=0.93, wpb=494703, bsz=16277.2, num_updates=3100, lr=0.000775, gnorm=0.362, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=2875 epoch 002: 1516 / 1689 loss=4.294, nll_loss=2.806, ppl=6.99, wps=552401, ups=1.11, wpb=495594, bsz=16465.1, num_updates=3200, lr=0.0008, gnorm=0.36, clip=0, loss_scale=4, train_wall=88, gb_free=23, wall=2965 epoch 002: 1516 / 1689 loss=4.294, nll_loss=2.806, ppl=6.99, wps=552401, ups=1.11, wpb=495594, bsz=16465.1, num_updates=3200, lr=0.0008, gnorm=0.36, clip=0, loss_scale=4, train_wall=88, gb_free=23, wall=2965 epoch 002: 1617 / 1689 loss=4.266, nll_loss=2.776, ppl=6.85, wps=543668, ups=1.09, wpb=496950, bsz=16306.1, num_updates=3300, lr=0.000825, gnorm=0.358, clip=0, loss_scale=4, train_wall=90, gb_free=21.8, wall=3056 epoch 002: 1617 / 1689 loss=4.266, nll_loss=2.776, ppl=6.85, wps=543668, ups=1.09, wpb=496950, bsz=16306.1, num_updates=3300, lr=0.000825, gnorm=0.358, clip=0, loss_scale=4, train_wall=90, gb_free=21.8, wall=3056 end of epoch 2 (average epoch stats below) epoch 002 | loss 4.55 | nll_loss 3.083 | ppl 8.48 | wps 535332 | ups 1.08 | wpb 495139 | bsz 16505.8 | num_updates 3372 | lr 0.000843 | gnorm 0.427 | clip 0.7 | loss_scale 4 | train_wall 1494 | gb_free 25.6 | wall 3121 epoch 002 | loss 4.55 | nll_loss 3.083 | ppl 8.48 | wps 535332 | ups 1.08 | wpb 495139 | bsz 16505.8 | num_updates 3372 | lr 0.000843 | gnorm 0.427 | clip 0.7 | loss_scale 4 | train_wall 1494 | gb_free 25.6 | wall 3121 Start iterating over samples epoch 003: 28 / 1689 loss=4.253, nll_loss=2.762, ppl=6.78, wps=541936, ups=1.1, wpb=490460, bsz=16511.3, num_updates=3400, lr=0.00085, gnorm=0.368, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=3147 epoch 003: 28 / 1689 loss=4.253, nll_loss=2.762, ppl=6.78, wps=541936, ups=1.1, wpb=490460, bsz=16511.3, num_updates=3400, lr=0.00085, gnorm=0.368, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=3147 epoch 003: 28 / 1689 loss=4.253, nll_loss=2.762, ppl=6.78, wps=541936, ups=1.1, wpb=490460, bsz=16511.3, num_updates=3400, lr=0.00085, gnorm=0.368, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=3147 epoch 003: 128 / 1689 loss=4.219, nll_loss=2.724, ppl=6.61, wps=549064, ups=1.11, wpb=495133, bsz=16849.4, num_updates=3500, lr=0.000875, gnorm=0.35, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=3237 epoch 003: 128 / 1689 loss=4.219, nll_loss=2.724, ppl=6.61, wps=549064, ups=1.11, wpb=495133, bsz=16849.4, num_updates=3500, lr=0.000875, gnorm=0.35, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=3237 epoch 003: 128 / 1689 loss=4.219, nll_loss=2.724, ppl=6.61, wps=549064, ups=1.11, wpb=495133, bsz=16849.4, num_updates=3500, lr=0.000875, gnorm=0.35, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=3237 epoch 003: 228 / 1689 loss=4.211, nll_loss=2.716, ppl=6.57, wps=551073, ups=1.11, wpb=495438, bsz=16848.2, num_updates=3600, lr=0.0009, gnorm=0.352, clip=0, loss_scale=4, train_wall=88, gb_free=22.9, wall=3327 epoch 003: 228 / 1689 loss=4.211, nll_loss=2.716, ppl=6.57, wps=551073, ups=1.11, wpb=495438, bsz=16848.2, num_updates=3600, lr=0.0009, gnorm=0.352, clip=0, loss_scale=4, train_wall=88, gb_free=22.9, wall=3327 epoch 003: 228 / 1689 loss=4.211, nll_loss=2.716, ppl=6.57, wps=551073, ups=1.11, wpb=495438, bsz=16848.2, num_updates=3600, lr=0.0009, gnorm=0.352, clip=0, loss_scale=4, train_wall=88, gb_free=22.9, wall=3327 epoch 003: 328 / 1689 loss=4.209, nll_loss=2.715, ppl=6.57, wps=542977, ups=1.1, wpb=494814, bsz=16504.7, num_updates=3700, lr=0.000925, gnorm=0.375, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=3418 epoch 003: 328 / 1689 loss=4.209, nll_loss=2.715, ppl=6.57, wps=542977, ups=1.1, wpb=494814, bsz=16504.7, num_updates=3700, lr=0.000925, gnorm=0.375, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=3418 epoch 003: 328 / 1689 loss=4.209, nll_loss=2.715, ppl=6.57, wps=542977, ups=1.1, wpb=494814, bsz=16504.7, num_updates=3700, lr=0.000925, gnorm=0.375, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=3418 epoch 003: 429 / 1689 loss=4.193, nll_loss=2.698, ppl=6.49, wps=542053, ups=1.1, wpb=494762, bsz=16420.4, num_updates=3800, lr=0.00095, gnorm=0.355, clip=0, loss_scale=2, train_wall=90, gb_free=21.1, wall=3509 epoch 003: 429 / 1689 loss=4.193, nll_loss=2.698, ppl=6.49, wps=542053, ups=1.1, wpb=494762, bsz=16420.4, num_updates=3800, lr=0.00095, gnorm=0.355, clip=0, loss_scale=2, train_wall=90, gb_free=21.1, wall=3509 epoch 003: 429 / 1689 loss=4.193, nll_loss=2.698, ppl=6.49, wps=542053, ups=1.1, wpb=494762, bsz=16420.4, num_updates=3800, lr=0.00095, gnorm=0.355, clip=0, loss_scale=2, train_wall=90, gb_free=21.1, wall=3509 epoch 003: 529 / 1689 loss=4.183, nll_loss=2.686, ppl=6.44, wps=548301, ups=1.11, wpb=496084, bsz=16501, num_updates=3900, lr=0.000975, gnorm=0.373, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=3600 epoch 003: 529 / 1689 loss=4.183, nll_loss=2.686, ppl=6.44, wps=548301, ups=1.11, wpb=496084, bsz=16501, num_updates=3900, lr=0.000975, gnorm=0.373, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=3600 epoch 003: 529 / 1689 loss=4.183, nll_loss=2.686, ppl=6.44, wps=548301, ups=1.11, wpb=496084, bsz=16501, num_updates=3900, lr=0.000975, gnorm=0.373, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=3600 epoch 003: 629 / 1689 loss=4.166, nll_loss=2.669, ppl=6.36, wps=551729, ups=1.11, wpb=495658, bsz=16451.8, num_updates=4000, lr=0.001, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=3690 epoch 003: 629 / 1689 loss=4.166, nll_loss=2.669, ppl=6.36, wps=551729, ups=1.11, wpb=495658, bsz=16451.8, num_updates=4000, lr=0.001, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=3690 epoch 003: 629 / 1689 loss=4.166, nll_loss=2.669, ppl=6.36, wps=551729, ups=1.11, wpb=495658, bsz=16451.8, num_updates=4000, lr=0.001, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=3690 begin validation on "valid" subset epoch 003 | valid on 'valid' subset | loss 4.224 | nll_loss 2.664 | ppl 6.34 | wps 0 | wpb 44526 | bsz 2008 | num_updates 4000 | best_loss 4.224 epoch 003 | valid on 'valid' subset | loss 4.224 | nll_loss 2.664 | ppl 6.34 | wps 0 | wpb 44526 | bsz 2008 | num_updates 4000 | best_loss 4.224 epoch 003 | valid on 'valid' subset | loss 4.224 | nll_loss 2.664 | ppl 6.34 | wps 0 | wpb 44526 | bsz 2008 | num_updates 4000 | best_loss 4.224 epoch 003: 729 / 1689 loss=4.164, nll_loss=2.667, ppl=6.35, wps=451566, ups=0.91, wpb=496213, bsz=16341, num_updates=4100, lr=0.00098773, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=3800 epoch 003: 729 / 1689 loss=4.164, nll_loss=2.667, ppl=6.35, wps=451566, ups=0.91, wpb=496213, bsz=16341, num_updates=4100, lr=0.00098773, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=3800 epoch 003: 729 / 1689 loss=4.164, nll_loss=2.667, ppl=6.35, wps=451566, ups=0.91, wpb=496213, bsz=16341, num_updates=4100, lr=0.00098773, gnorm=0.351, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=3800 epoch 003: 829 / 1689 loss=4.149, nll_loss=2.651, ppl=6.28, wps=549734, ups=1.11, wpb=493044, bsz=16311.8, num_updates=4200, lr=0.0009759, gnorm=0.338, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=3889 epoch 003: 829 / 1689 loss=4.149, nll_loss=2.651, ppl=6.28, wps=549734, ups=1.11, wpb=493044, bsz=16311.8, num_updates=4200, lr=0.0009759, gnorm=0.338, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=3889 epoch 003: 829 / 1689 loss=4.149, nll_loss=2.651, ppl=6.28, wps=549734, ups=1.11, wpb=493044, bsz=16311.8, num_updates=4200, lr=0.0009759, gnorm=0.338, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=3889 epoch 003: 929 / 1689 loss=4.134, nll_loss=2.635, ppl=6.21, wps=548109, ups=1.11, wpb=494194, bsz=16361.3, num_updates=4300, lr=0.000964486, gnorm=0.334, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=3980 epoch 003: 929 / 1689 loss=4.134, nll_loss=2.635, ppl=6.21, wps=548109, ups=1.11, wpb=494194, bsz=16361.3, num_updates=4300, lr=0.000964486, gnorm=0.334, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=3980 epoch 003: 929 / 1689 loss=4.134, nll_loss=2.635, ppl=6.21, wps=548109, ups=1.11, wpb=494194, bsz=16361.3, num_updates=4300, lr=0.000964486, gnorm=0.334, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=3980 epoch 003: 1029 / 1689 loss=4.126, nll_loss=2.626, ppl=6.17, wps=554207, ups=1.12, wpb=496554, bsz=16440.3, num_updates=4400, lr=0.000953463, gnorm=0.332, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=4069 epoch 003: 1029 / 1689 loss=4.126, nll_loss=2.626, ppl=6.17, wps=554207, ups=1.12, wpb=496554, bsz=16440.3, num_updates=4400, lr=0.000953463, gnorm=0.332, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=4069 epoch 003: 1029 / 1689 loss=4.126, nll_loss=2.626, ppl=6.17, wps=554207, ups=1.12, wpb=496554, bsz=16440.3, num_updates=4400, lr=0.000953463, gnorm=0.332, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=4069 epoch 003: 1129 / 1689 loss=4.102, nll_loss=2.601, ppl=6.07, wps=549996, ups=1.11, wpb=496113, bsz=16938.9, num_updates=4500, lr=0.000942809, gnorm=0.326, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4159 epoch 003: 1129 / 1689 loss=4.102, nll_loss=2.601, ppl=6.07, wps=549996, ups=1.11, wpb=496113, bsz=16938.9, num_updates=4500, lr=0.000942809, gnorm=0.326, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4159 epoch 003: 1129 / 1689 loss=4.102, nll_loss=2.601, ppl=6.07, wps=549996, ups=1.11, wpb=496113, bsz=16938.9, num_updates=4500, lr=0.000942809, gnorm=0.326, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4159 epoch 003: 1229 / 1689 loss=4.098, nll_loss=2.597, ppl=6.05, wps=552892, ups=1.12, wpb=494048, bsz=16474.2, num_updates=4600, lr=0.000932505, gnorm=0.325, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=4249 epoch 003: 1229 / 1689 loss=4.098, nll_loss=2.597, ppl=6.05, wps=552892, ups=1.12, wpb=494048, bsz=16474.2, num_updates=4600, lr=0.000932505, gnorm=0.325, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=4249 epoch 003: 1229 / 1689 loss=4.098, nll_loss=2.597, ppl=6.05, wps=552892, ups=1.12, wpb=494048, bsz=16474.2, num_updates=4600, lr=0.000932505, gnorm=0.325, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=4249 epoch 003: 1329 / 1689 loss=4.086, nll_loss=2.584, ppl=6, wps=552501, ups=1.11, wpb=495822, bsz=16321, num_updates=4700, lr=0.000922531, gnorm=0.321, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=4338 epoch 003: 1329 / 1689 loss=4.086, nll_loss=2.584, ppl=6, wps=552501, ups=1.11, wpb=495822, bsz=16321, num_updates=4700, lr=0.000922531, gnorm=0.321, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=4338 epoch 003: 1329 / 1689 loss=4.086, nll_loss=2.584, ppl=6, wps=552501, ups=1.11, wpb=495822, bsz=16321, num_updates=4700, lr=0.000922531, gnorm=0.321, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=4338 epoch 003: 1430 / 1689 loss=4.071, nll_loss=2.568, ppl=5.93, wps=543183, ups=1.09, wpb=496349, bsz=16693, num_updates=4800, lr=0.000912871, gnorm=0.31, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=4430 epoch 003: 1430 / 1689 loss=4.071, nll_loss=2.568, ppl=5.93, wps=543183, ups=1.09, wpb=496349, bsz=16693, num_updates=4800, lr=0.000912871, gnorm=0.31, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=4430 epoch 003: 1430 / 1689 loss=4.071, nll_loss=2.568, ppl=5.93, wps=543183, ups=1.09, wpb=496349, bsz=16693, num_updates=4800, lr=0.000912871, gnorm=0.31, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=4430 epoch 003: 1530 / 1689 loss=4.06, nll_loss=2.556, ppl=5.88, wps=549973, ups=1.11, wpb=495935, bsz=16441.5, num_updates=4900, lr=0.000903508, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=4520 epoch 003: 1530 / 1689 loss=4.06, nll_loss=2.556, ppl=5.88, wps=549973, ups=1.11, wpb=495935, bsz=16441.5, num_updates=4900, lr=0.000903508, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=4520 epoch 003: 1530 / 1689 loss=4.06, nll_loss=2.556, ppl=5.88, wps=549973, ups=1.11, wpb=495935, bsz=16441.5, num_updates=4900, lr=0.000903508, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=4520 epoch 003: 1630 / 1689 loss=4.056, nll_loss=2.551, ppl=5.86, wps=551824, ups=1.11, wpb=495488, bsz=16111.9, num_updates=5000, lr=0.000894427, gnorm=0.311, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4610 epoch 003: 1630 / 1689 loss=4.056, nll_loss=2.551, ppl=5.86, wps=551824, ups=1.11, wpb=495488, bsz=16111.9, num_updates=5000, lr=0.000894427, gnorm=0.311, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4610 epoch 003: 1630 / 1689 loss=4.056, nll_loss=2.551, ppl=5.86, wps=551824, ups=1.11, wpb=495488, bsz=16111.9, num_updates=5000, lr=0.000894427, gnorm=0.311, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=4610 begin validation on "valid" subset epoch 003 | valid on 'valid' subset | loss 4.048 | nll_loss 2.484 | ppl 5.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 5000 | best_loss 4.048 epoch 003 | valid on 'valid' subset | loss 4.048 | nll_loss 2.484 | ppl 5.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 5000 | best_loss 4.048 epoch 003 | valid on 'valid' subset | loss 4.048 | nll_loss 2.484 | ppl 5.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 5000 | best_loss 4.048 end of epoch 3 (average epoch stats below) epoch 003 | loss 4.138 | nll_loss 2.639 | ppl 6.23 | wps 529997 | ups 1.07 | wpb 495116 | bsz 16504.1 | num_updates 5059 | lr 0.000889196 | gnorm 0.338 | clip 0 | loss_scale 4 | train_wall 1494 | gb_free 23.3 | wall 4697 epoch 003 | loss 4.138 | nll_loss 2.639 | ppl 6.23 | wps 529997 | ups 1.07 | wpb 495116 | bsz 16504.1 | num_updates 5059 | lr 0.000889196 | gnorm 0.338 | clip 0 | loss_scale 4 | train_wall 1494 | gb_free 23.3 | wall 4697 epoch 003 | loss 4.138 | nll_loss 2.639 | ppl 6.23 | wps 529997 | ups 1.07 | wpb 495116 | bsz 16504.1 | num_updates 5059 | lr 0.000889196 | gnorm 0.338 | clip 0 | loss_scale 4 | train_wall 1494 | gb_free 23.3 | wall 4697 Start iterating over samples epoch 004: 41 / 1689 loss=4.033, nll_loss=2.526, ppl=5.76, wps=396590, ups=0.81, wpb=491687, bsz=16828.6, num_updates=5100, lr=0.000885615, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=4734 epoch 004: 41 / 1689 loss=4.033, nll_loss=2.526, ppl=5.76, wps=396590, ups=0.81, wpb=491687, bsz=16828.6, num_updates=5100, lr=0.000885615, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=4734 epoch 004: 41 / 1689 loss=4.033, nll_loss=2.526, ppl=5.76, wps=396590, ups=0.81, wpb=491687, bsz=16828.6, num_updates=5100, lr=0.000885615, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=4734 epoch 004: 41 / 1689 loss=4.033, nll_loss=2.526, ppl=5.76, wps=396590, ups=0.81, wpb=491687, bsz=16828.6, num_updates=5100, lr=0.000885615, gnorm=0.312, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=4734 epoch 004: 141 / 1689 loss=4.019, nll_loss=2.511, ppl=5.7, wps=562414, ups=1.13, wpb=496024, bsz=16396.1, num_updates=5200, lr=0.000877058, gnorm=0.299, clip=0, loss_scale=4, train_wall=87, gb_free=22.1, wall=4822 epoch 004: 141 / 1689 loss=4.019, nll_loss=2.511, ppl=5.7, wps=562414, ups=1.13, wpb=496024, bsz=16396.1, num_updates=5200, lr=0.000877058, gnorm=0.299, clip=0, loss_scale=4, train_wall=87, gb_free=22.1, wall=4822 epoch 004: 141 / 1689 loss=4.019, nll_loss=2.511, ppl=5.7, wps=562414, ups=1.13, wpb=496024, bsz=16396.1, num_updates=5200, lr=0.000877058, gnorm=0.299, clip=0, loss_scale=4, train_wall=87, gb_free=22.1, wall=4822 epoch 004: 141 / 1689 loss=4.019, nll_loss=2.511, ppl=5.7, wps=562414, ups=1.13, wpb=496024, bsz=16396.1, num_updates=5200, lr=0.000877058, gnorm=0.299, clip=0, loss_scale=4, train_wall=87, gb_free=22.1, wall=4822 epoch 004: 242 / 1689 loss=4.012, nll_loss=2.503, ppl=5.67, wps=554165, ups=1.12, wpb=496338, bsz=16307.9, num_updates=5300, lr=0.000868744, gnorm=0.303, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=4912 epoch 004: 242 / 1689 loss=4.012, nll_loss=2.503, ppl=5.67, wps=554165, ups=1.12, wpb=496338, bsz=16307.9, num_updates=5300, lr=0.000868744, gnorm=0.303, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=4912 epoch 004: 242 / 1689 loss=4.012, nll_loss=2.503, ppl=5.67, wps=554165, ups=1.12, wpb=496338, bsz=16307.9, num_updates=5300, lr=0.000868744, gnorm=0.303, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=4912 epoch 004: 242 / 1689 loss=4.012, nll_loss=2.503, ppl=5.67, wps=554165, ups=1.12, wpb=496338, bsz=16307.9, num_updates=5300, lr=0.000868744, gnorm=0.303, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=4912 epoch 004: 342 / 1689 loss=4.005, nll_loss=2.496, ppl=5.64, wps=551301, ups=1.12, wpb=493975, bsz=16735.7, num_updates=5400, lr=0.000860663, gnorm=0.312, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=5001 epoch 004: 342 / 1689 loss=4.005, nll_loss=2.496, ppl=5.64, wps=551301, ups=1.12, wpb=493975, bsz=16735.7, num_updates=5400, lr=0.000860663, gnorm=0.312, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=5001 epoch 004: 342 / 1689 loss=4.005, nll_loss=2.496, ppl=5.64, wps=551301, ups=1.12, wpb=493975, bsz=16735.7, num_updates=5400, lr=0.000860663, gnorm=0.312, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=5001 epoch 004: 342 / 1689 loss=4.005, nll_loss=2.496, ppl=5.64, wps=551301, ups=1.12, wpb=493975, bsz=16735.7, num_updates=5400, lr=0.000860663, gnorm=0.312, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=5001 epoch 004: 442 / 1689 loss=3.993, nll_loss=2.483, ppl=5.59, wps=563260, ups=1.13, wpb=496824, bsz=16678.8, num_updates=5500, lr=0.000852803, gnorm=0.298, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=5089 epoch 004: 442 / 1689 loss=3.993, nll_loss=2.483, ppl=5.59, wps=563260, ups=1.13, wpb=496824, bsz=16678.8, num_updates=5500, lr=0.000852803, gnorm=0.298, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=5089 epoch 004: 442 / 1689 loss=3.993, nll_loss=2.483, ppl=5.59, wps=563260, ups=1.13, wpb=496824, bsz=16678.8, num_updates=5500, lr=0.000852803, gnorm=0.298, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=5089 epoch 004: 442 / 1689 loss=3.993, nll_loss=2.483, ppl=5.59, wps=563260, ups=1.13, wpb=496824, bsz=16678.8, num_updates=5500, lr=0.000852803, gnorm=0.298, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=5089 epoch 004: 542 / 1689 loss=4.003, nll_loss=2.494, ppl=5.64, wps=558098, ups=1.13, wpb=495916, bsz=16423.7, num_updates=5600, lr=0.000845154, gnorm=0.295, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5178 epoch 004: 542 / 1689 loss=4.003, nll_loss=2.494, ppl=5.64, wps=558098, ups=1.13, wpb=495916, bsz=16423.7, num_updates=5600, lr=0.000845154, gnorm=0.295, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5178 epoch 004: 542 / 1689 loss=4.003, nll_loss=2.494, ppl=5.64, wps=558098, ups=1.13, wpb=495916, bsz=16423.7, num_updates=5600, lr=0.000845154, gnorm=0.295, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5178 epoch 004: 542 / 1689 loss=4.003, nll_loss=2.494, ppl=5.64, wps=558098, ups=1.13, wpb=495916, bsz=16423.7, num_updates=5600, lr=0.000845154, gnorm=0.295, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5178 epoch 004: 642 / 1689 loss=3.989, nll_loss=2.48, ppl=5.58, wps=552409, ups=1.12, wpb=494379, bsz=16563, num_updates=5700, lr=0.000837708, gnorm=0.305, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5268 epoch 004: 642 / 1689 loss=3.989, nll_loss=2.48, ppl=5.58, wps=552409, ups=1.12, wpb=494379, bsz=16563, num_updates=5700, lr=0.000837708, gnorm=0.305, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5268 epoch 004: 642 / 1689 loss=3.989, nll_loss=2.48, ppl=5.58, wps=552409, ups=1.12, wpb=494379, bsz=16563, num_updates=5700, lr=0.000837708, gnorm=0.305, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5268 epoch 004: 642 / 1689 loss=3.989, nll_loss=2.48, ppl=5.58, wps=552409, ups=1.12, wpb=494379, bsz=16563, num_updates=5700, lr=0.000837708, gnorm=0.305, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=5268 epoch 004: 742 / 1689 loss=3.982, nll_loss=2.472, ppl=5.55, wps=555559, ups=1.12, wpb=495835, bsz=16531, num_updates=5800, lr=0.000830455, gnorm=0.301, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5357 epoch 004: 742 / 1689 loss=3.982, nll_loss=2.472, ppl=5.55, wps=555559, ups=1.12, wpb=495835, bsz=16531, num_updates=5800, lr=0.000830455, gnorm=0.301, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5357 epoch 004: 742 / 1689 loss=3.982, nll_loss=2.472, ppl=5.55, wps=555559, ups=1.12, wpb=495835, bsz=16531, num_updates=5800, lr=0.000830455, gnorm=0.301, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5357 epoch 004: 742 / 1689 loss=3.982, nll_loss=2.472, ppl=5.55, wps=555559, ups=1.12, wpb=495835, bsz=16531, num_updates=5800, lr=0.000830455, gnorm=0.301, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5357 epoch 004: 842 / 1689 loss=3.987, nll_loss=2.478, ppl=5.57, wps=553150, ups=1.12, wpb=495047, bsz=16309.8, num_updates=5900, lr=0.000823387, gnorm=0.293, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5446 epoch 004: 842 / 1689 loss=3.987, nll_loss=2.478, ppl=5.57, wps=553150, ups=1.12, wpb=495047, bsz=16309.8, num_updates=5900, lr=0.000823387, gnorm=0.293, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5446 epoch 004: 842 / 1689 loss=3.987, nll_loss=2.478, ppl=5.57, wps=553150, ups=1.12, wpb=495047, bsz=16309.8, num_updates=5900, lr=0.000823387, gnorm=0.293, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5446 epoch 004: 842 / 1689 loss=3.987, nll_loss=2.478, ppl=5.57, wps=553150, ups=1.12, wpb=495047, bsz=16309.8, num_updates=5900, lr=0.000823387, gnorm=0.293, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=5446 epoch 004: 942 / 1689 loss=3.968, nll_loss=2.457, ppl=5.49, wps=549782, ups=1.11, wpb=495867, bsz=16303.1, num_updates=6000, lr=0.000816497, gnorm=0.3, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=5537 epoch 004: 942 / 1689 loss=3.968, nll_loss=2.457, ppl=5.49, wps=549782, ups=1.11, wpb=495867, bsz=16303.1, num_updates=6000, lr=0.000816497, gnorm=0.3, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=5537 epoch 004: 942 / 1689 loss=3.968, nll_loss=2.457, ppl=5.49, wps=549782, ups=1.11, wpb=495867, bsz=16303.1, num_updates=6000, lr=0.000816497, gnorm=0.3, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=5537 epoch 004: 942 / 1689 loss=3.968, nll_loss=2.457, ppl=5.49, wps=549782, ups=1.11, wpb=495867, bsz=16303.1, num_updates=6000, lr=0.000816497, gnorm=0.3, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=5537 begin validation on "valid" subset epoch 004 | valid on 'valid' subset | loss 3.975 | nll_loss 2.415 | ppl 5.33 | wps 0 | wpb 44526 | bsz 2008 | num_updates 6000 | best_loss 3.975 epoch 004 | valid on 'valid' subset | loss 3.975 | nll_loss 2.415 | ppl 5.33 | wps 0 | wpb 44526 | bsz 2008 | num_updates 6000 | best_loss 3.975 epoch 004 | valid on 'valid' subset | loss 3.975 | nll_loss 2.415 | ppl 5.33 | wps 0 | wpb 44526 | bsz 2008 | num_updates 6000 | best_loss 3.975 epoch 004 | valid on 'valid' subset | loss 3.975 | nll_loss 2.415 | ppl 5.33 | wps 0 | wpb 44526 | bsz 2008 | num_updates 6000 | best_loss 3.975 epoch 004: 1042 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=461325, ups=0.93, wpb=495105, bsz=16619.3, num_updates=6100, lr=0.000809776, gnorm=0.288, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=5644 epoch 004: 1042 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=461325, ups=0.93, wpb=495105, bsz=16619.3, num_updates=6100, lr=0.000809776, gnorm=0.288, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=5644 epoch 004: 1042 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=461325, ups=0.93, wpb=495105, bsz=16619.3, num_updates=6100, lr=0.000809776, gnorm=0.288, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=5644 epoch 004: 1042 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=461325, ups=0.93, wpb=495105, bsz=16619.3, num_updates=6100, lr=0.000809776, gnorm=0.288, clip=0, loss_scale=4, train_wall=88, gb_free=21.6, wall=5644 epoch 004: 1142 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=551894, ups=1.12, wpb=494956, bsz=16745.6, num_updates=6200, lr=0.000803219, gnorm=0.291, clip=0, loss_scale=4, train_wall=88, gb_free=21.2, wall=5734 epoch 004: 1142 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=551894, ups=1.12, wpb=494956, bsz=16745.6, num_updates=6200, lr=0.000803219, gnorm=0.291, clip=0, loss_scale=4, train_wall=88, gb_free=21.2, wall=5734 epoch 004: 1142 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=551894, ups=1.12, wpb=494956, bsz=16745.6, num_updates=6200, lr=0.000803219, gnorm=0.291, clip=0, loss_scale=4, train_wall=88, gb_free=21.2, wall=5734 epoch 004: 1142 / 1689 loss=3.963, nll_loss=2.452, ppl=5.47, wps=551894, ups=1.12, wpb=494956, bsz=16745.6, num_updates=6200, lr=0.000803219, gnorm=0.291, clip=0, loss_scale=4, train_wall=88, gb_free=21.2, wall=5734 epoch 004: 1242 / 1689 loss=3.957, nll_loss=2.445, ppl=5.45, wps=555070, ups=1.12, wpb=494927, bsz=16264.9, num_updates=6300, lr=0.000796819, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=5823 epoch 004: 1242 / 1689 loss=3.957, nll_loss=2.445, ppl=5.45, wps=555070, ups=1.12, wpb=494927, bsz=16264.9, num_updates=6300, lr=0.000796819, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=5823 epoch 004: 1242 / 1689 loss=3.957, nll_loss=2.445, ppl=5.45, wps=555070, ups=1.12, wpb=494927, bsz=16264.9, num_updates=6300, lr=0.000796819, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=5823 epoch 004: 1242 / 1689 loss=3.957, nll_loss=2.445, ppl=5.45, wps=555070, ups=1.12, wpb=494927, bsz=16264.9, num_updates=6300, lr=0.000796819, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=5823 epoch 004: 1343 / 1689 loss=3.953, nll_loss=2.441, ppl=5.43, wps=541875, ups=1.1, wpb=494640, bsz=16639.3, num_updates=6400, lr=0.000790569, gnorm=0.29, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=5914 epoch 004: 1343 / 1689 loss=3.953, nll_loss=2.441, ppl=5.43, wps=541875, ups=1.1, wpb=494640, bsz=16639.3, num_updates=6400, lr=0.000790569, gnorm=0.29, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=5914 epoch 004: 1343 / 1689 loss=3.953, nll_loss=2.441, ppl=5.43, wps=541875, ups=1.1, wpb=494640, bsz=16639.3, num_updates=6400, lr=0.000790569, gnorm=0.29, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=5914 epoch 004: 1343 / 1689 loss=3.953, nll_loss=2.441, ppl=5.43, wps=541875, ups=1.1, wpb=494640, bsz=16639.3, num_updates=6400, lr=0.000790569, gnorm=0.29, clip=0, loss_scale=4, train_wall=90, gb_free=21.9, wall=5914 epoch 004: 1443 / 1689 loss=3.947, nll_loss=2.435, ppl=5.41, wps=551348, ups=1.12, wpb=494315, bsz=16775.5, num_updates=6500, lr=0.000784465, gnorm=0.308, clip=0, loss_scale=4, train_wall=88, gb_free=23.1, wall=6004 epoch 004: 1443 / 1689 loss=3.947, nll_loss=2.435, ppl=5.41, wps=551348, ups=1.12, wpb=494315, bsz=16775.5, num_updates=6500, lr=0.000784465, gnorm=0.308, clip=0, loss_scale=4, train_wall=88, gb_free=23.1, wall=6004 epoch 004: 1443 / 1689 loss=3.947, nll_loss=2.435, ppl=5.41, wps=551348, ups=1.12, wpb=494315, bsz=16775.5, num_updates=6500, lr=0.000784465, gnorm=0.308, clip=0, loss_scale=4, train_wall=88, gb_free=23.1, wall=6004 epoch 004: 1443 / 1689 loss=3.947, nll_loss=2.435, ppl=5.41, wps=551348, ups=1.12, wpb=494315, bsz=16775.5, num_updates=6500, lr=0.000784465, gnorm=0.308, clip=0, loss_scale=4, train_wall=88, gb_free=23.1, wall=6004 epoch 004: 1543 / 1689 loss=3.943, nll_loss=2.43, ppl=5.39, wps=556637, ups=1.13, wpb=494708, bsz=16402, num_updates=6600, lr=0.000778499, gnorm=0.293, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=6093 epoch 004: 1543 / 1689 loss=3.943, nll_loss=2.43, ppl=5.39, wps=556637, ups=1.13, wpb=494708, bsz=16402, num_updates=6600, lr=0.000778499, gnorm=0.293, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=6093 epoch 004: 1543 / 1689 loss=3.943, nll_loss=2.43, ppl=5.39, wps=556637, ups=1.13, wpb=494708, bsz=16402, num_updates=6600, lr=0.000778499, gnorm=0.293, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=6093 epoch 004: 1543 / 1689 loss=3.943, nll_loss=2.43, ppl=5.39, wps=556637, ups=1.13, wpb=494708, bsz=16402, num_updates=6600, lr=0.000778499, gnorm=0.293, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=6093 epoch 004: 1643 / 1689 loss=3.931, nll_loss=2.417, ppl=5.34, wps=558018, ups=1.12, wpb=496980, bsz=16207, num_updates=6700, lr=0.000772667, gnorm=0.274, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=6182 epoch 004: 1643 / 1689 loss=3.931, nll_loss=2.417, ppl=5.34, wps=558018, ups=1.12, wpb=496980, bsz=16207, num_updates=6700, lr=0.000772667, gnorm=0.274, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=6182 epoch 004: 1643 / 1689 loss=3.931, nll_loss=2.417, ppl=5.34, wps=558018, ups=1.12, wpb=496980, bsz=16207, num_updates=6700, lr=0.000772667, gnorm=0.274, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=6182 epoch 004: 1643 / 1689 loss=3.931, nll_loss=2.417, ppl=5.34, wps=558018, ups=1.12, wpb=496980, bsz=16207, num_updates=6700, lr=0.000772667, gnorm=0.274, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=6182 end of epoch 4 (average epoch stats below) epoch 004 | loss 3.976 | nll_loss 2.465 | ppl 5.52 | wps 547595 | ups 1.11 | wpb 495127 | bsz 16501.5 | num_updates 6746 | lr 0.000770029 | gnorm 0.296 | clip 0 | loss_scale 4 | train_wall 1486 | gb_free 23.6 | wall 6222 epoch 004 | loss 3.976 | nll_loss 2.465 | ppl 5.52 | wps 547595 | ups 1.11 | wpb 495127 | bsz 16501.5 | num_updates 6746 | lr 0.000770029 | gnorm 0.296 | clip 0 | loss_scale 4 | train_wall 1486 | gb_free 23.6 | wall 6222 epoch 004 | loss 3.976 | nll_loss 2.465 | ppl 5.52 | wps 547595 | ups 1.11 | wpb 495127 | bsz 16501.5 | num_updates 6746 | lr 0.000770029 | gnorm 0.296 | clip 0 | loss_scale 4 | train_wall 1486 | gb_free 23.6 | wall 6222 epoch 004 | loss 3.976 | nll_loss 2.465 | ppl 5.52 | wps 547595 | ups 1.11 | wpb 495127 | bsz 16501.5 | num_updates 6746 | lr 0.000770029 | gnorm 0.296 | clip 0 | loss_scale 4 | train_wall 1486 | gb_free 23.6 | wall 6222 Start iterating over samples epoch 005: 54 / 1689 loss=3.921, nll_loss=2.406, ppl=5.3, wps=540585, ups=1.1, wpb=491068, bsz=16227.2, num_updates=6800, lr=0.000766965, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6273 epoch 005: 54 / 1689 loss=3.921, nll_loss=2.406, ppl=5.3, wps=540585, ups=1.1, wpb=491068, bsz=16227.2, num_updates=6800, lr=0.000766965, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6273 epoch 005: 54 / 1689 loss=3.921, nll_loss=2.406, ppl=5.3, wps=540585, ups=1.1, wpb=491068, bsz=16227.2, num_updates=6800, lr=0.000766965, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6273 epoch 005: 54 / 1689 loss=3.921, nll_loss=2.406, ppl=5.3, wps=540585, ups=1.1, wpb=491068, bsz=16227.2, num_updates=6800, lr=0.000766965, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6273 epoch 005: 54 / 1689 loss=3.921, nll_loss=2.406, ppl=5.3, wps=540585, ups=1.1, wpb=491068, bsz=16227.2, num_updates=6800, lr=0.000766965, gnorm=0.292, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6273 epoch 005: 154 / 1689 loss=3.904, nll_loss=2.386, ppl=5.23, wps=553657, ups=1.12, wpb=494999, bsz=16482.5, num_updates=6900, lr=0.000761387, gnorm=0.287, clip=0, loss_scale=8, train_wall=88, gb_free=22.3, wall=6362 epoch 005: 154 / 1689 loss=3.904, nll_loss=2.386, ppl=5.23, wps=553657, ups=1.12, wpb=494999, bsz=16482.5, num_updates=6900, lr=0.000761387, gnorm=0.287, clip=0, loss_scale=8, train_wall=88, gb_free=22.3, wall=6362 epoch 005: 154 / 1689 loss=3.904, nll_loss=2.386, ppl=5.23, wps=553657, ups=1.12, wpb=494999, bsz=16482.5, num_updates=6900, lr=0.000761387, gnorm=0.287, clip=0, loss_scale=8, train_wall=88, gb_free=22.3, wall=6362 epoch 005: 154 / 1689 loss=3.904, nll_loss=2.386, ppl=5.23, wps=553657, ups=1.12, wpb=494999, bsz=16482.5, num_updates=6900, lr=0.000761387, gnorm=0.287, clip=0, loss_scale=8, train_wall=88, gb_free=22.3, wall=6362 epoch 005: 154 / 1689 loss=3.904, nll_loss=2.386, ppl=5.23, wps=553657, ups=1.12, wpb=494999, bsz=16482.5, num_updates=6900, lr=0.000761387, gnorm=0.287, clip=0, loss_scale=8, train_wall=88, gb_free=22.3, wall=6362 epoch 005: 255 / 1689 loss=3.898, nll_loss=2.38, ppl=5.2, wps=548642, ups=1.11, wpb=494694, bsz=16334.9, num_updates=7000, lr=0.000755929, gnorm=0.29, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6452 epoch 005: 255 / 1689 loss=3.898, nll_loss=2.38, ppl=5.2, wps=548642, ups=1.11, wpb=494694, bsz=16334.9, num_updates=7000, lr=0.000755929, gnorm=0.29, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6452 epoch 005: 255 / 1689 loss=3.898, nll_loss=2.38, ppl=5.2, wps=548642, ups=1.11, wpb=494694, bsz=16334.9, num_updates=7000, lr=0.000755929, gnorm=0.29, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6452 epoch 005: 255 / 1689 loss=3.898, nll_loss=2.38, ppl=5.2, wps=548642, ups=1.11, wpb=494694, bsz=16334.9, num_updates=7000, lr=0.000755929, gnorm=0.29, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6452 epoch 005: 255 / 1689 loss=3.898, nll_loss=2.38, ppl=5.2, wps=548642, ups=1.11, wpb=494694, bsz=16334.9, num_updates=7000, lr=0.000755929, gnorm=0.29, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6452 begin validation on "valid" subset epoch 005 | valid on 'valid' subset | loss 3.918 | nll_loss 2.358 | ppl 5.13 | wps 0 | wpb 44526 | bsz 2008 | num_updates 7000 | best_loss 3.918 epoch 005 | valid on 'valid' subset | loss 3.918 | nll_loss 2.358 | ppl 5.13 | wps 0 | wpb 44526 | bsz 2008 | num_updates 7000 | best_loss 3.918 epoch 005 | valid on 'valid' subset | loss 3.918 | nll_loss 2.358 | ppl 5.13 | wps 0 | wpb 44526 | bsz 2008 | num_updates 7000 | best_loss 3.918 epoch 005 | valid on 'valid' subset | loss 3.918 | nll_loss 2.358 | ppl 5.13 | wps 0 | wpb 44526 | bsz 2008 | num_updates 7000 | best_loss 3.918 epoch 005 | valid on 'valid' subset | loss 3.918 | nll_loss 2.358 | ppl 5.13 | wps 0 | wpb 44526 | bsz 2008 | num_updates 7000 | best_loss 3.918 epoch 005: 355 / 1689 loss=3.906, nll_loss=2.389, ppl=5.24, wps=457679, ups=0.92, wpb=495012, bsz=16464.3, num_updates=7100, lr=0.000750587, gnorm=0.298, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6560 epoch 005: 355 / 1689 loss=3.906, nll_loss=2.389, ppl=5.24, wps=457679, ups=0.92, wpb=495012, bsz=16464.3, num_updates=7100, lr=0.000750587, gnorm=0.298, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6560 epoch 005: 355 / 1689 loss=3.906, nll_loss=2.389, ppl=5.24, wps=457679, ups=0.92, wpb=495012, bsz=16464.3, num_updates=7100, lr=0.000750587, gnorm=0.298, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6560 epoch 005: 355 / 1689 loss=3.906, nll_loss=2.389, ppl=5.24, wps=457679, ups=0.92, wpb=495012, bsz=16464.3, num_updates=7100, lr=0.000750587, gnorm=0.298, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6560 epoch 005: 355 / 1689 loss=3.906, nll_loss=2.389, ppl=5.24, wps=457679, ups=0.92, wpb=495012, bsz=16464.3, num_updates=7100, lr=0.000750587, gnorm=0.298, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=6560 epoch 005: 455 / 1689 loss=3.894, nll_loss=2.376, ppl=5.19, wps=554665, ups=1.12, wpb=496196, bsz=16668.8, num_updates=7200, lr=0.000745356, gnorm=0.276, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6650 epoch 005: 455 / 1689 loss=3.894, nll_loss=2.376, ppl=5.19, wps=554665, ups=1.12, wpb=496196, bsz=16668.8, num_updates=7200, lr=0.000745356, gnorm=0.276, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6650 epoch 005: 455 / 1689 loss=3.894, nll_loss=2.376, ppl=5.19, wps=554665, ups=1.12, wpb=496196, bsz=16668.8, num_updates=7200, lr=0.000745356, gnorm=0.276, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6650 epoch 005: 455 / 1689 loss=3.894, nll_loss=2.376, ppl=5.19, wps=554665, ups=1.12, wpb=496196, bsz=16668.8, num_updates=7200, lr=0.000745356, gnorm=0.276, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6650 epoch 005: 455 / 1689 loss=3.894, nll_loss=2.376, ppl=5.19, wps=554665, ups=1.12, wpb=496196, bsz=16668.8, num_updates=7200, lr=0.000745356, gnorm=0.276, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=6650 epoch 005: 555 / 1689 loss=3.892, nll_loss=2.375, ppl=5.19, wps=554133, ups=1.12, wpb=495952, bsz=16605.6, num_updates=7300, lr=0.000740233, gnorm=0.278, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=6739 epoch 005: 555 / 1689 loss=3.892, nll_loss=2.375, ppl=5.19, wps=554133, ups=1.12, wpb=495952, bsz=16605.6, num_updates=7300, lr=0.000740233, gnorm=0.278, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=6739 epoch 005: 555 / 1689 loss=3.892, nll_loss=2.375, ppl=5.19, wps=554133, ups=1.12, wpb=495952, bsz=16605.6, num_updates=7300, lr=0.000740233, gnorm=0.278, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=6739 epoch 005: 555 / 1689 loss=3.892, nll_loss=2.375, ppl=5.19, wps=554133, ups=1.12, wpb=495952, bsz=16605.6, num_updates=7300, lr=0.000740233, gnorm=0.278, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=6739 epoch 005: 555 / 1689 loss=3.892, nll_loss=2.375, ppl=5.19, wps=554133, ups=1.12, wpb=495952, bsz=16605.6, num_updates=7300, lr=0.000740233, gnorm=0.278, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=6739 epoch 005: 655 / 1689 loss=3.896, nll_loss=2.378, ppl=5.2, wps=550094, ups=1.11, wpb=495698, bsz=16362.2, num_updates=7400, lr=0.000735215, gnorm=0.288, clip=0, loss_scale=4, train_wall=89, gb_free=21.8, wall=6829 epoch 005: 655 / 1689 loss=3.896, nll_loss=2.378, ppl=5.2, wps=550094, ups=1.11, wpb=495698, bsz=16362.2, num_updates=7400, lr=0.000735215, gnorm=0.288, clip=0, loss_scale=4, train_wall=89, gb_free=21.8, wall=6829 epoch 005: 655 / 1689 loss=3.896, nll_loss=2.378, ppl=5.2, wps=550094, ups=1.11, wpb=495698, bsz=16362.2, num_updates=7400, lr=0.000735215, gnorm=0.288, clip=0, loss_scale=4, train_wall=89, gb_free=21.8, wall=6829 epoch 005: 655 / 1689 loss=3.896, nll_loss=2.378, ppl=5.2, wps=550094, ups=1.11, wpb=495698, bsz=16362.2, num_updates=7400, lr=0.000735215, gnorm=0.288, clip=0, loss_scale=4, train_wall=89, gb_free=21.8, wall=6829 epoch 005: 655 / 1689 loss=3.896, nll_loss=2.378, ppl=5.2, wps=550094, ups=1.11, wpb=495698, bsz=16362.2, num_updates=7400, lr=0.000735215, gnorm=0.288, clip=0, loss_scale=4, train_wall=89, gb_free=21.8, wall=6829 epoch 005: 756 / 1689 loss=3.898, nll_loss=2.382, ppl=5.21, wps=545564, ups=1.1, wpb=494518, bsz=16453.4, num_updates=7500, lr=0.000730297, gnorm=0.297, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=6920 epoch 005: 756 / 1689 loss=3.898, nll_loss=2.382, ppl=5.21, wps=545564, ups=1.1, wpb=494518, bsz=16453.4, num_updates=7500, lr=0.000730297, gnorm=0.297, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=6920 epoch 005: 756 / 1689 loss=3.898, nll_loss=2.382, ppl=5.21, wps=545564, ups=1.1, wpb=494518, bsz=16453.4, num_updates=7500, lr=0.000730297, gnorm=0.297, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=6920 epoch 005: 756 / 1689 loss=3.898, nll_loss=2.382, ppl=5.21, wps=545564, ups=1.1, wpb=494518, bsz=16453.4, num_updates=7500, lr=0.000730297, gnorm=0.297, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=6920 epoch 005: 756 / 1689 loss=3.898, nll_loss=2.382, ppl=5.21, wps=545564, ups=1.1, wpb=494518, bsz=16453.4, num_updates=7500, lr=0.000730297, gnorm=0.297, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=6920 epoch 005: 856 / 1689 loss=3.891, nll_loss=2.374, ppl=5.18, wps=548314, ups=1.11, wpb=495323, bsz=16568.4, num_updates=7600, lr=0.000725476, gnorm=0.28, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=7010 epoch 005: 856 / 1689 loss=3.891, nll_loss=2.374, ppl=5.18, wps=548314, ups=1.11, wpb=495323, bsz=16568.4, num_updates=7600, lr=0.000725476, gnorm=0.28, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=7010 epoch 005: 856 / 1689 loss=3.891, nll_loss=2.374, ppl=5.18, wps=548314, ups=1.11, wpb=495323, bsz=16568.4, num_updates=7600, lr=0.000725476, gnorm=0.28, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=7010 epoch 005: 856 / 1689 loss=3.891, nll_loss=2.374, ppl=5.18, wps=548314, ups=1.11, wpb=495323, bsz=16568.4, num_updates=7600, lr=0.000725476, gnorm=0.28, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=7010 epoch 005: 856 / 1689 loss=3.891, nll_loss=2.374, ppl=5.18, wps=548314, ups=1.11, wpb=495323, bsz=16568.4, num_updates=7600, lr=0.000725476, gnorm=0.28, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=7010 epoch 005: 956 / 1689 loss=3.897, nll_loss=2.381, ppl=5.21, wps=548813, ups=1.11, wpb=495898, bsz=16465, num_updates=7700, lr=0.00072075, gnorm=0.293, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=7101 epoch 005: 956 / 1689 loss=3.897, nll_loss=2.381, ppl=5.21, wps=548813, ups=1.11, wpb=495898, bsz=16465, num_updates=7700, lr=0.00072075, gnorm=0.293, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=7101 epoch 005: 956 / 1689 loss=3.897, nll_loss=2.381, ppl=5.21, wps=548813, ups=1.11, wpb=495898, bsz=16465, num_updates=7700, lr=0.00072075, gnorm=0.293, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=7101 epoch 005: 956 / 1689 loss=3.897, nll_loss=2.381, ppl=5.21, wps=548813, ups=1.11, wpb=495898, bsz=16465, num_updates=7700, lr=0.00072075, gnorm=0.293, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=7101 epoch 005: 956 / 1689 loss=3.897, nll_loss=2.381, ppl=5.21, wps=548813, ups=1.11, wpb=495898, bsz=16465, num_updates=7700, lr=0.00072075, gnorm=0.293, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=7101 epoch 005: 1056 / 1689 loss=3.883, nll_loss=2.365, ppl=5.15, wps=552672, ups=1.11, wpb=497649, bsz=16667.8, num_updates=7800, lr=0.000716115, gnorm=0.279, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=7191 epoch 005: 1056 / 1689 loss=3.883, nll_loss=2.365, ppl=5.15, wps=552672, ups=1.11, wpb=497649, bsz=16667.8, num_updates=7800, lr=0.000716115, gnorm=0.279, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=7191 epoch 005: 1056 / 1689 loss=3.883, nll_loss=2.365, ppl=5.15, wps=552672, ups=1.11, wpb=497649, bsz=16667.8, num_updates=7800, lr=0.000716115, gnorm=0.279, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=7191 epoch 005: 1056 / 1689 loss=3.883, nll_loss=2.365, ppl=5.15, wps=552672, ups=1.11, wpb=497649, bsz=16667.8, num_updates=7800, lr=0.000716115, gnorm=0.279, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=7191 epoch 005: 1056 / 1689 loss=3.883, nll_loss=2.365, ppl=5.15, wps=552672, ups=1.11, wpb=497649, bsz=16667.8, num_updates=7800, lr=0.000716115, gnorm=0.279, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=7191 epoch 005: 1156 / 1689 loss=3.878, nll_loss=2.36, ppl=5.13, wps=552796, ups=1.12, wpb=494045, bsz=16677.1, num_updates=7900, lr=0.000711568, gnorm=0.285, clip=0, loss_scale=4, train_wall=88, gb_free=22.7, wall=7280 epoch 005: 1156 / 1689 loss=3.878, nll_loss=2.36, ppl=5.13, wps=552796, ups=1.12, wpb=494045, bsz=16677.1, num_updates=7900, lr=0.000711568, gnorm=0.285, clip=0, loss_scale=4, train_wall=88, gb_free=22.7, wall=7280 epoch 005: 1156 / 1689 loss=3.878, nll_loss=2.36, ppl=5.13, wps=552796, ups=1.12, wpb=494045, bsz=16677.1, num_updates=7900, lr=0.000711568, gnorm=0.285, clip=0, loss_scale=4, train_wall=88, gb_free=22.7, wall=7280 epoch 005: 1156 / 1689 loss=3.878, nll_loss=2.36, ppl=5.13, wps=552796, ups=1.12, wpb=494045, bsz=16677.1, num_updates=7900, lr=0.000711568, gnorm=0.285, clip=0, loss_scale=4, train_wall=88, gb_free=22.7, wall=7280 epoch 005: 1156 / 1689 loss=3.878, nll_loss=2.36, ppl=5.13, wps=552796, ups=1.12, wpb=494045, bsz=16677.1, num_updates=7900, lr=0.000711568, gnorm=0.285, clip=0, loss_scale=4, train_wall=88, gb_free=22.7, wall=7280 epoch 005: 1257 / 1689 loss=3.879, nll_loss=2.361, ppl=5.14, wps=541310, ups=1.1, wpb=494066, bsz=16369, num_updates=8000, lr=0.000707107, gnorm=0.282, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=7371 epoch 005: 1257 / 1689 loss=3.879, nll_loss=2.361, ppl=5.14, wps=541310, ups=1.1, wpb=494066, bsz=16369, num_updates=8000, lr=0.000707107, gnorm=0.282, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=7371 epoch 005: 1257 / 1689 loss=3.879, nll_loss=2.361, ppl=5.14, wps=541310, ups=1.1, wpb=494066, bsz=16369, num_updates=8000, lr=0.000707107, gnorm=0.282, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=7371 epoch 005: 1257 / 1689 loss=3.879, nll_loss=2.361, ppl=5.14, wps=541310, ups=1.1, wpb=494066, bsz=16369, num_updates=8000, lr=0.000707107, gnorm=0.282, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=7371 epoch 005: 1257 / 1689 loss=3.879, nll_loss=2.361, ppl=5.14, wps=541310, ups=1.1, wpb=494066, bsz=16369, num_updates=8000, lr=0.000707107, gnorm=0.282, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=7371 begin validation on "valid" subset epoch 005 | valid on 'valid' subset | loss 3.876 | nll_loss 2.317 | ppl 4.98 | wps 0 | wpb 44526 | bsz 2008 | num_updates 8000 | best_loss 3.876 epoch 005 | valid on 'valid' subset | loss 3.876 | nll_loss 2.317 | ppl 4.98 | wps 0 | wpb 44526 | bsz 2008 | num_updates 8000 | best_loss 3.876 epoch 005 | valid on 'valid' subset | loss 3.876 | nll_loss 2.317 | ppl 4.98 | wps 0 | wpb 44526 | bsz 2008 | num_updates 8000 | best_loss 3.876 epoch 005 | valid on 'valid' subset | loss 3.876 | nll_loss 2.317 | ppl 4.98 | wps 0 | wpb 44526 | bsz 2008 | num_updates 8000 | best_loss 3.876 epoch 005 | valid on 'valid' subset | loss 3.876 | nll_loss 2.317 | ppl 4.98 | wps 0 | wpb 44526 | bsz 2008 | num_updates 8000 | best_loss 3.876 epoch 005: 1357 / 1689 loss=3.877, nll_loss=2.36, ppl=5.13, wps=460650, ups=0.93, wpb=494962, bsz=16785.4, num_updates=8100, lr=0.000702728, gnorm=0.297, clip=0, loss_scale=2, train_wall=87, gb_free=22, wall=7479 epoch 005: 1357 / 1689 loss=3.877, nll_loss=2.36, ppl=5.13, wps=460650, ups=0.93, wpb=494962, bsz=16785.4, num_updates=8100, lr=0.000702728, gnorm=0.297, clip=0, loss_scale=2, train_wall=87, gb_free=22, wall=7479 epoch 005: 1357 / 1689 loss=3.877, nll_loss=2.36, ppl=5.13, wps=460650, ups=0.93, wpb=494962, bsz=16785.4, num_updates=8100, lr=0.000702728, gnorm=0.297, clip=0, loss_scale=2, train_wall=87, gb_free=22, wall=7479 epoch 005: 1357 / 1689 loss=3.877, nll_loss=2.36, ppl=5.13, wps=460650, ups=0.93, wpb=494962, bsz=16785.4, num_updates=8100, lr=0.000702728, gnorm=0.297, clip=0, loss_scale=2, train_wall=87, gb_free=22, wall=7479 epoch 005: 1357 / 1689 loss=3.877, nll_loss=2.36, ppl=5.13, wps=460650, ups=0.93, wpb=494962, bsz=16785.4, num_updates=8100, lr=0.000702728, gnorm=0.297, clip=0, loss_scale=2, train_wall=87, gb_free=22, wall=7479 epoch 005: 1457 / 1689 loss=3.871, nll_loss=2.353, ppl=5.11, wps=550298, ups=1.11, wpb=495674, bsz=16477.2, num_updates=8200, lr=0.00069843, gnorm=0.27, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=7569 epoch 005: 1457 / 1689 loss=3.871, nll_loss=2.353, ppl=5.11, wps=550298, ups=1.11, wpb=495674, bsz=16477.2, num_updates=8200, lr=0.00069843, gnorm=0.27, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=7569 epoch 005: 1457 / 1689 loss=3.871, nll_loss=2.353, ppl=5.11, wps=550298, ups=1.11, wpb=495674, bsz=16477.2, num_updates=8200, lr=0.00069843, gnorm=0.27, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=7569 epoch 005: 1457 / 1689 loss=3.871, nll_loss=2.353, ppl=5.11, wps=550298, ups=1.11, wpb=495674, bsz=16477.2, num_updates=8200, lr=0.00069843, gnorm=0.27, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=7569 epoch 005: 1457 / 1689 loss=3.871, nll_loss=2.353, ppl=5.11, wps=550298, ups=1.11, wpb=495674, bsz=16477.2, num_updates=8200, lr=0.00069843, gnorm=0.27, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=7569 epoch 005: 1557 / 1689 loss=3.872, nll_loss=2.354, ppl=5.11, wps=552457, ups=1.11, wpb=495932, bsz=16143.1, num_updates=8300, lr=0.00069421, gnorm=0.273, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=7659 epoch 005: 1557 / 1689 loss=3.872, nll_loss=2.354, ppl=5.11, wps=552457, ups=1.11, wpb=495932, bsz=16143.1, num_updates=8300, lr=0.00069421, gnorm=0.273, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=7659 epoch 005: 1557 / 1689 loss=3.872, nll_loss=2.354, ppl=5.11, wps=552457, ups=1.11, wpb=495932, bsz=16143.1, num_updates=8300, lr=0.00069421, gnorm=0.273, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=7659 epoch 005: 1557 / 1689 loss=3.872, nll_loss=2.354, ppl=5.11, wps=552457, ups=1.11, wpb=495932, bsz=16143.1, num_updates=8300, lr=0.00069421, gnorm=0.273, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=7659 epoch 005: 1557 / 1689 loss=3.872, nll_loss=2.354, ppl=5.11, wps=552457, ups=1.11, wpb=495932, bsz=16143.1, num_updates=8300, lr=0.00069421, gnorm=0.273, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=7659 epoch 005: 1657 / 1689 loss=3.868, nll_loss=2.35, ppl=5.1, wps=549599, ups=1.11, wpb=495486, bsz=16830.4, num_updates=8400, lr=0.000690066, gnorm=0.293, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=7749 epoch 005: 1657 / 1689 loss=3.868, nll_loss=2.35, ppl=5.1, wps=549599, ups=1.11, wpb=495486, bsz=16830.4, num_updates=8400, lr=0.000690066, gnorm=0.293, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=7749 epoch 005: 1657 / 1689 loss=3.868, nll_loss=2.35, ppl=5.1, wps=549599, ups=1.11, wpb=495486, bsz=16830.4, num_updates=8400, lr=0.000690066, gnorm=0.293, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=7749 epoch 005: 1657 / 1689 loss=3.868, nll_loss=2.35, ppl=5.1, wps=549599, ups=1.11, wpb=495486, bsz=16830.4, num_updates=8400, lr=0.000690066, gnorm=0.293, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=7749 epoch 005: 1657 / 1689 loss=3.868, nll_loss=2.35, ppl=5.1, wps=549599, ups=1.11, wpb=495486, bsz=16830.4, num_updates=8400, lr=0.000690066, gnorm=0.293, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=7749 end of epoch 5 (average epoch stats below) epoch 005 | loss 3.888 | nll_loss 2.371 | ppl 5.17 | wps 536930 | ups 1.08 | wpb 495133 | bsz 16503 | num_updates 8432 | lr 0.000688755 | gnorm 0.285 | clip 0 | loss_scale 2 | train_wall 1489 | gb_free 22.5 | wall 7777 epoch 005 | loss 3.888 | nll_loss 2.371 | ppl 5.17 | wps 536930 | ups 1.08 | wpb 495133 | bsz 16503 | num_updates 8432 | lr 0.000688755 | gnorm 0.285 | clip 0 | loss_scale 2 | train_wall 1489 | gb_free 22.5 | wall 7777 epoch 005 | loss 3.888 | nll_loss 2.371 | ppl 5.17 | wps 536930 | ups 1.08 | wpb 495133 | bsz 16503 | num_updates 8432 | lr 0.000688755 | gnorm 0.285 | clip 0 | loss_scale 2 | train_wall 1489 | gb_free 22.5 | wall 7777 epoch 005 | loss 3.888 | nll_loss 2.371 | ppl 5.17 | wps 536930 | ups 1.08 | wpb 495133 | bsz 16503 | num_updates 8432 | lr 0.000688755 | gnorm 0.285 | clip 0 | loss_scale 2 | train_wall 1489 | gb_free 22.5 | wall 7777 epoch 005 | loss 3.888 | nll_loss 2.371 | ppl 5.17 | wps 536930 | ups 1.08 | wpb 495133 | bsz 16503 | num_updates 8432 | lr 0.000688755 | gnorm 0.285 | clip 0 | loss_scale 2 | train_wall 1489 | gb_free 22.5 | wall 7777 Start iterating over samples epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 68 / 1689 loss=3.842, nll_loss=2.319, ppl=4.99, wps=554032, ups=1.12, wpb=492620, bsz=16013.7, num_updates=8500, lr=0.000685994, gnorm=0.275, clip=0, loss_scale=4, train_wall=87, gb_free=21.6, wall=7838 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 168 / 1689 loss=3.838, nll_loss=2.316, ppl=4.98, wps=558973, ups=1.13, wpb=495533, bsz=16295.3, num_updates=8600, lr=0.000681994, gnorm=0.277, clip=0, loss_scale=4, train_wall=88, gb_free=21.3, wall=7926 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 268 / 1689 loss=3.831, nll_loss=2.308, ppl=4.95, wps=556023, ups=1.12, wpb=495847, bsz=16619.6, num_updates=8700, lr=0.000678064, gnorm=0.273, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=8016 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 368 / 1689 loss=3.841, nll_loss=2.319, ppl=4.99, wps=556158, ups=1.13, wpb=494280, bsz=16324.2, num_updates=8800, lr=0.0006742, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=8104 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 468 / 1689 loss=3.833, nll_loss=2.311, ppl=4.96, wps=558112, ups=1.13, wpb=496041, bsz=16477.8, num_updates=8900, lr=0.000670402, gnorm=0.286, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=8193 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 epoch 006: 568 / 1689 loss=3.836, nll_loss=2.314, ppl=4.97, wps=556572, ups=1.12, wpb=495655, bsz=16606.7, num_updates=9000, lr=0.000666667, gnorm=0.273, clip=0, loss_scale=8, train_wall=88, gb_free=21.5, wall=8282 begin validation on "valid" subset epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006 | valid on 'valid' subset | loss 3.849 | nll_loss 2.297 | ppl 4.91 | wps 0 | wpb 44526 | bsz 2008 | num_updates 9000 | best_loss 3.849 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 669 / 1689 loss=3.834, nll_loss=2.312, ppl=4.96, wps=456122, ups=0.92, wpb=494182, bsz=16803.5, num_updates=9100, lr=0.000662994, gnorm=0.284, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=8391 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 770 / 1689 loss=3.835, nll_loss=2.313, ppl=4.97, wps=550332, ups=1.11, wpb=495871, bsz=16373.1, num_updates=9200, lr=0.00065938, gnorm=0.271, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=8481 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 870 / 1689 loss=3.836, nll_loss=2.315, ppl=4.98, wps=553154, ups=1.12, wpb=495285, bsz=16919.1, num_updates=9300, lr=0.000655826, gnorm=0.289, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=8570 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 970 / 1689 loss=3.838, nll_loss=2.317, ppl=4.98, wps=550851, ups=1.12, wpb=493156, bsz=16292.8, num_updates=9400, lr=0.000652328, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=8660 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1070 / 1689 loss=3.831, nll_loss=2.309, ppl=4.95, wps=552248, ups=1.12, wpb=494574, bsz=16490.6, num_updates=9500, lr=0.000648886, gnorm=0.274, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=8750 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1170 / 1689 loss=3.829, nll_loss=2.307, ppl=4.95, wps=551637, ups=1.12, wpb=494476, bsz=16754.3, num_updates=9600, lr=0.000645497, gnorm=0.279, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=8839 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1270 / 1689 loss=3.833, nll_loss=2.312, ppl=4.97, wps=557304, ups=1.12, wpb=495553, bsz=16170.1, num_updates=9700, lr=0.000642161, gnorm=0.276, clip=0, loss_scale=4, train_wall=87, gb_free=21.3, wall=8928 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1370 / 1689 loss=3.824, nll_loss=2.301, ppl=4.93, wps=551249, ups=1.11, wpb=497314, bsz=16725.4, num_updates=9800, lr=0.000638877, gnorm=0.276, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=9018 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1470 / 1689 loss=3.831, nll_loss=2.309, ppl=4.96, wps=547878, ups=1.11, wpb=494850, bsz=16358.3, num_updates=9900, lr=0.000635642, gnorm=0.271, clip=0, loss_scale=4, train_wall=89, gb_free=22.3, wall=9109 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 epoch 006: 1570 / 1689 loss=3.819, nll_loss=2.297, ppl=4.91, wps=546716, ups=1.1, wpb=495856, bsz=16750.3, num_updates=10000, lr=0.000632456, gnorm=0.27, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=9199 begin validation on "valid" subset epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006 | valid on 'valid' subset | loss 3.829 | nll_loss 2.27 | ppl 4.82 | wps 0 | wpb 44526 | bsz 2008 | num_updates 10000 | best_loss 3.829 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 epoch 006: 1670 / 1689 loss=3.823, nll_loss=2.301, ppl=4.93, wps=460895, ups=0.93, wpb=495675, bsz=16481.7, num_updates=10100, lr=0.000629317, gnorm=0.264, clip=0, loss_scale=4, train_wall=87, gb_free=21.9, wall=9307 end of epoch 6 (average epoch stats below) epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 epoch 006 | loss 3.832 | nll_loss 2.31 | ppl 4.96 | wps 540280 | ups 1.09 | wpb 495132 | bsz 16502.7 | num_updates 10119 | lr 0.000628726 | gnorm 0.276 | clip 0 | loss_scale 4 | train_wall 1487 | gb_free 23 | wall 9323 Start iterating over samples epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 81 / 1689 loss=3.799, nll_loss=2.273, ppl=4.83, wps=546100, ups=1.11, wpb=490479, bsz=16313.8, num_updates=10200, lr=0.000626224, gnorm=0.278, clip=0, loss_scale=8, train_wall=87, gb_free=21.6, wall=9397 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 182 / 1689 loss=3.784, nll_loss=2.257, ppl=4.78, wps=549602, ups=1.11, wpb=495690, bsz=16727.4, num_updates=10300, lr=0.000623177, gnorm=0.255, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=9487 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 282 / 1689 loss=3.792, nll_loss=2.266, ppl=4.81, wps=551167, ups=1.11, wpb=495341, bsz=16592.9, num_updates=10400, lr=0.000620174, gnorm=0.28, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9577 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 382 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=556559, ups=1.12, wpb=495481, bsz=16103.1, num_updates=10500, lr=0.000617213, gnorm=0.266, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=9666 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 482 / 1689 loss=3.8, nll_loss=2.275, ppl=4.84, wps=553429, ups=1.12, wpb=494968, bsz=16600.6, num_updates=10600, lr=0.000614295, gnorm=0.265, clip=0, loss_scale=4, train_wall=88, gb_free=21.5, wall=9755 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 582 / 1689 loss=3.794, nll_loss=2.268, ppl=4.82, wps=552330, ups=1.11, wpb=496079, bsz=16431.6, num_updates=10700, lr=0.000611418, gnorm=0.27, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=9845 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 683 / 1689 loss=3.796, nll_loss=2.271, ppl=4.83, wps=548312, ups=1.11, wpb=494828, bsz=16295.2, num_updates=10800, lr=0.000608581, gnorm=0.264, clip=0, loss_scale=4, train_wall=89, gb_free=21.9, wall=9935 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 783 / 1689 loss=3.795, nll_loss=2.27, ppl=4.82, wps=550923, ups=1.11, wpb=494245, bsz=16231.8, num_updates=10900, lr=0.000605783, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=10025 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 epoch 007: 883 / 1689 loss=3.793, nll_loss=2.268, ppl=4.82, wps=552433, ups=1.12, wpb=494758, bsz=16427.9, num_updates=11000, lr=0.000603023, gnorm=0.272, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=10115 begin validation on "valid" subset epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007 | valid on 'valid' subset | loss 3.813 | nll_loss 2.257 | ppl 4.78 | wps 0 | wpb 44526 | bsz 2008 | num_updates 11000 | best_loss 3.813 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 983 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=374975, ups=0.76, wpb=496550, bsz=16253.8, num_updates=11100, lr=0.0006003, gnorm=0.26, clip=0, loss_scale=4, train_wall=86, gb_free=21.8, wall=10247 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1083 / 1689 loss=3.792, nll_loss=2.268, ppl=4.82, wps=562650, ups=1.13, wpb=496079, bsz=16901.5, num_updates=11200, lr=0.000597614, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=10335 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1184 / 1689 loss=3.792, nll_loss=2.267, ppl=4.81, wps=558298, ups=1.13, wpb=496037, bsz=16664.2, num_updates=11300, lr=0.000594964, gnorm=0.264, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=10424 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1284 / 1689 loss=3.793, nll_loss=2.269, ppl=4.82, wps=560552, ups=1.13, wpb=495333, bsz=16460.6, num_updates=11400, lr=0.000592349, gnorm=0.263, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=10512 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1385 / 1689 loss=3.796, nll_loss=2.272, ppl=4.83, wps=548230, ups=1.11, wpb=494861, bsz=16332.3, num_updates=11500, lr=0.000589768, gnorm=0.266, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=10603 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1485 / 1689 loss=3.782, nll_loss=2.257, ppl=4.78, wps=557318, ups=1.12, wpb=496165, bsz=16659.5, num_updates=11600, lr=0.00058722, gnorm=0.251, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=10692 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1585 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=550135, ups=1.11, wpb=495847, bsz=16996.2, num_updates=11700, lr=0.000584705, gnorm=0.248, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=10782 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 epoch 007: 1685 / 1689 loss=3.781, nll_loss=2.255, ppl=4.77, wps=548495, ups=1.11, wpb=494720, bsz=16535.1, num_updates=11800, lr=0.000582223, gnorm=0.262, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=10872 end of epoch 7 (average epoch stats below) epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 epoch 007 | loss 3.791 | nll_loss 2.266 | ppl 4.81 | wps 537578 | ups 1.09 | wpb 495103 | bsz 16500.2 | num_updates 11804 | lr 0.000582124 | gnorm 0.264 | clip 0 | loss_scale 2 | train_wall 1484 | gb_free 23 | wall 10875 Start iterating over samples epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 96 / 1689 loss=3.751, nll_loss=2.221, ppl=4.66, wps=545394, ups=1.11, wpb=491549, bsz=16528.6, num_updates=11900, lr=0.000579771, gnorm=0.266, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=10962 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 epoch 008: 196 / 1689 loss=3.754, nll_loss=2.224, ppl=4.67, wps=551249, ups=1.11, wpb=494513, bsz=16695.1, num_updates=12000, lr=0.00057735, gnorm=0.269, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11052 begin validation on "valid" subset epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008 | valid on 'valid' subset | loss 3.806 | nll_loss 2.242 | ppl 4.73 | wps 0 | wpb 44526 | bsz 2008 | num_updates 12000 | best_loss 3.806 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 296 / 1689 loss=3.758, nll_loss=2.23, ppl=4.69, wps=460972, ups=0.93, wpb=495279, bsz=16524.9, num_updates=12100, lr=0.00057496, gnorm=0.262, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=11159 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 396 / 1689 loss=3.756, nll_loss=2.227, ppl=4.68, wps=547923, ups=1.11, wpb=495200, bsz=16683.3, num_updates=12200, lr=0.000572598, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=11250 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 496 / 1689 loss=3.767, nll_loss=2.239, ppl=4.72, wps=552031, ups=1.12, wpb=493323, bsz=15931, num_updates=12300, lr=0.000570266, gnorm=0.252, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=11339 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 597 / 1689 loss=3.768, nll_loss=2.241, ppl=4.73, wps=549703, ups=1.11, wpb=495236, bsz=16436.6, num_updates=12400, lr=0.000567962, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11429 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 697 / 1689 loss=3.76, nll_loss=2.231, ppl=4.7, wps=553237, ups=1.12, wpb=495765, bsz=16581, num_updates=12500, lr=0.000565685, gnorm=0.272, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=11519 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 797 / 1689 loss=3.766, nll_loss=2.239, ppl=4.72, wps=554288, ups=1.12, wpb=495839, bsz=16430.6, num_updates=12600, lr=0.000563436, gnorm=0.25, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=11608 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 897 / 1689 loss=3.763, nll_loss=2.236, ppl=4.71, wps=550987, ups=1.11, wpb=496104, bsz=16689, num_updates=12700, lr=0.000561214, gnorm=0.257, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=11698 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 997 / 1689 loss=3.767, nll_loss=2.24, ppl=4.72, wps=547570, ups=1.11, wpb=494239, bsz=16444.2, num_updates=12800, lr=0.000559017, gnorm=0.258, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=11789 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1097 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=554332, ups=1.12, wpb=494973, bsz=16402.6, num_updates=12900, lr=0.000556846, gnorm=0.242, clip=0, loss_scale=4, train_wall=87, gb_free=21.8, wall=11878 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 epoch 008: 1197 / 1689 loss=3.756, nll_loss=2.228, ppl=4.68, wps=553167, ups=1.11, wpb=497463, bsz=16235.4, num_updates=13000, lr=0.0005547, gnorm=0.253, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=11968 begin validation on "valid" subset epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008 | valid on 'valid' subset | loss 3.794 | nll_loss 2.238 | ppl 4.72 | wps 0 | wpb 44526 | bsz 2008 | num_updates 13000 | best_loss 3.794 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1297 / 1689 loss=3.755, nll_loss=2.227, ppl=4.68, wps=345027, ups=0.7, wpb=495386, bsz=16704.1, num_updates=13100, lr=0.000552579, gnorm=0.241, clip=0, loss_scale=4, train_wall=117, gb_free=21.6, wall=12111 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1397 / 1689 loss=3.759, nll_loss=2.232, ppl=4.7, wps=555156, ups=1.12, wpb=496355, bsz=16590.8, num_updates=13200, lr=0.000550482, gnorm=0.267, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=12201 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1497 / 1689 loss=3.764, nll_loss=2.237, ppl=4.71, wps=551795, ups=1.11, wpb=495317, bsz=16443.7, num_updates=13300, lr=0.000548408, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12290 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 epoch 008: 1597 / 1689 loss=3.762, nll_loss=2.235, ppl=4.71, wps=551410, ups=1.11, wpb=494717, bsz=16612.9, num_updates=13400, lr=0.000546358, gnorm=0.264, clip=0, loss_scale=8, train_wall=88, gb_free=21.3, wall=12380 end of epoch 8 (average epoch stats below) epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 epoch 008 | loss 3.76 | nll_loss 2.232 | ppl 4.7 | wps 526287 | ups 1.06 | wpb 495129 | bsz 16507.3 | num_updates 13491 | lr 0.000544513 | gnorm 0.255 | clip 0 | loss_scale 4 | train_wall 1515 | gb_free 22.7 | wall 12462 Start iterating over samples epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 9 / 1689 loss=3.76, nll_loss=2.233, ppl=4.7, wps=537612, ups=1.09, wpb=491822, bsz=16531.4, num_updates=13500, lr=0.000544331, gnorm=0.26, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=12472 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 109 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=553780, ups=1.12, wpb=495608, bsz=16332.6, num_updates=13600, lr=0.000542326, gnorm=0.243, clip=0, loss_scale=4, train_wall=88, gb_free=22.1, wall=12561 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 209 / 1689 loss=3.735, nll_loss=2.204, ppl=4.61, wps=548363, ups=1.11, wpb=493659, bsz=16754, num_updates=13700, lr=0.000540343, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=22.5, wall=12651 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 309 / 1689 loss=3.726, nll_loss=2.194, ppl=4.58, wps=550725, ups=1.11, wpb=497012, bsz=16546.1, num_updates=13800, lr=0.000538382, gnorm=0.237, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=12741 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 409 / 1689 loss=3.737, nll_loss=2.206, ppl=4.61, wps=553243, ups=1.12, wpb=495285, bsz=16275.8, num_updates=13900, lr=0.000536442, gnorm=0.241, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=12831 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 epoch 009: 510 / 1689 loss=3.732, nll_loss=2.202, ppl=4.6, wps=540071, ups=1.09, wpb=494984, bsz=16511.7, num_updates=14000, lr=0.000534522, gnorm=0.25, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=12923 begin validation on "valid" subset epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009 | valid on 'valid' subset | loss 3.791 | nll_loss 2.228 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 14000 | best_loss 3.791 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 610 / 1689 loss=3.734, nll_loss=2.203, ppl=4.61, wps=322298, ups=0.65, wpb=495231, bsz=16346.6, num_updates=14100, lr=0.000532624, gnorm=0.252, clip=0, loss_scale=4, train_wall=129, gb_free=21.3, wall=13076 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 711 / 1689 loss=3.731, nll_loss=2.2, ppl=4.59, wps=543915, ups=1.1, wpb=495578, bsz=16521.7, num_updates=14200, lr=0.000530745, gnorm=0.24, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=13167 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 811 / 1689 loss=3.738, nll_loss=2.208, ppl=4.62, wps=550094, ups=1.11, wpb=494861, bsz=16637.5, num_updates=14300, lr=0.000528886, gnorm=0.246, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=13257 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 911 / 1689 loss=3.742, nll_loss=2.212, ppl=4.63, wps=549282, ups=1.11, wpb=495369, bsz=16696.6, num_updates=14400, lr=0.000527046, gnorm=0.25, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=13348 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1011 / 1689 loss=3.737, nll_loss=2.208, ppl=4.62, wps=551115, ups=1.11, wpb=495114, bsz=16360.1, num_updates=14500, lr=0.000525226, gnorm=0.241, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=13437 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1111 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=549721, ups=1.11, wpb=494900, bsz=16756.2, num_updates=14600, lr=0.000523424, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=13527 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1211 / 1689 loss=3.735, nll_loss=2.205, ppl=4.61, wps=552604, ups=1.12, wpb=495309, bsz=16216.9, num_updates=14700, lr=0.000521641, gnorm=0.245, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=13617 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1311 / 1689 loss=3.736, nll_loss=2.206, ppl=4.61, wps=549938, ups=1.11, wpb=495766, bsz=16353.8, num_updates=14800, lr=0.000519875, gnorm=0.246, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13707 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1411 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=552958, ups=1.12, wpb=495312, bsz=16649.7, num_updates=14900, lr=0.000518128, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=13797 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 epoch 009: 1511 / 1689 loss=3.742, nll_loss=2.213, ppl=4.64, wps=555040, ups=1.12, wpb=495654, bsz=16451.1, num_updates=15000, lr=0.000516398, gnorm=0.249, clip=0, loss_scale=4, train_wall=88, gb_free=21.9, wall=13886 begin validation on "valid" subset epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009 | valid on 'valid' subset | loss 3.784 | nll_loss 2.229 | ppl 4.69 | wps 0 | wpb 44526 | bsz 2008 | num_updates 15000 | best_loss 3.784 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 epoch 009: 1611 / 1689 loss=3.734, nll_loss=2.204, ppl=4.61, wps=446365, ups=0.9, wpb=495442, bsz=16697.3, num_updates=15100, lr=0.000514685, gnorm=0.249, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=13997 end of epoch 9 (average epoch stats below) epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 epoch 009 | loss 3.735 | nll_loss 2.205 | ppl 4.61 | wps 520620 | ups 1.05 | wpb 495133 | bsz 16501.7 | num_updates 15177 | lr 0.000513378 | gnorm 0.244 | clip 0 | loss_scale 4 | train_wall 1531 | gb_free 25.4 | wall 14065 Start iterating over samples epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 23 / 1689 loss=3.733, nll_loss=2.203, ppl=4.6, wps=541920, ups=1.1, wpb=491781, bsz=16549.7, num_updates=15200, lr=0.000512989, gnorm=0.23, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=14088 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 124 / 1689 loss=3.697, nll_loss=2.162, ppl=4.47, wps=551456, ups=1.11, wpb=495515, bsz=16349.3, num_updates=15300, lr=0.00051131, gnorm=0.244, clip=0, loss_scale=2, train_wall=89, gb_free=22.9, wall=14178 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 224 / 1689 loss=3.712, nll_loss=2.179, ppl=4.53, wps=561088, ups=1.13, wpb=495280, bsz=16744.1, num_updates=15400, lr=0.000509647, gnorm=0.242, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=14266 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 324 / 1689 loss=3.718, nll_loss=2.185, ppl=4.55, wps=555628, ups=1.12, wpb=495349, bsz=16496.1, num_updates=15500, lr=0.000508001, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=21.5, wall=14355 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 424 / 1689 loss=3.709, nll_loss=2.176, ppl=4.52, wps=554862, ups=1.12, wpb=495917, bsz=16609.7, num_updates=15600, lr=0.00050637, gnorm=0.24, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=14445 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 524 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=552091, ups=1.11, wpb=496666, bsz=16304.5, num_updates=15700, lr=0.000504754, gnorm=0.235, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=14534 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 624 / 1689 loss=3.705, nll_loss=2.172, ppl=4.51, wps=549450, ups=1.11, wpb=495407, bsz=16805.3, num_updates=15800, lr=0.000503155, gnorm=0.245, clip=0, loss_scale=4, train_wall=89, gb_free=22.5, wall=14625 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 724 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=552450, ups=1.12, wpb=495419, bsz=16520.5, num_updates=15900, lr=0.00050157, gnorm=0.236, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=14714 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 epoch 010: 824 / 1689 loss=3.716, nll_loss=2.184, ppl=4.54, wps=550758, ups=1.11, wpb=495783, bsz=16543.2, num_updates=16000, lr=0.0005, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=14804 begin validation on "valid" subset epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010 | valid on 'valid' subset | loss 3.769 | nll_loss 2.206 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 16000 | best_loss 3.769 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 924 / 1689 loss=3.715, nll_loss=2.183, ppl=4.54, wps=457081, ups=0.92, wpb=496619, bsz=16738.4, num_updates=16100, lr=0.000498445, gnorm=0.233, clip=0, loss_scale=4, train_wall=89, gb_free=21.4, wall=14913 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1024 / 1689 loss=3.722, nll_loss=2.191, ppl=4.56, wps=550180, ups=1.11, wpb=495302, bsz=16598.1, num_updates=16200, lr=0.000496904, gnorm=0.239, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=15003 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1125 / 1689 loss=3.72, nll_loss=2.189, ppl=4.56, wps=547428, ups=1.11, wpb=494788, bsz=16364.2, num_updates=16300, lr=0.000495377, gnorm=0.234, clip=0, loss_scale=4, train_wall=90, gb_free=22, wall=15093 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1225 / 1689 loss=3.718, nll_loss=2.187, ppl=4.55, wps=551116, ups=1.11, wpb=495627, bsz=16405.8, num_updates=16400, lr=0.000493865, gnorm=0.234, clip=0, loss_scale=4, train_wall=89, gb_free=21.6, wall=15183 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1325 / 1689 loss=3.721, nll_loss=2.191, ppl=4.56, wps=545390, ups=1.1, wpb=494456, bsz=16599.7, num_updates=16500, lr=0.000492366, gnorm=0.236, clip=0, loss_scale=4, train_wall=89, gb_free=21.3, wall=15274 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1425 / 1689 loss=3.715, nll_loss=2.184, ppl=4.54, wps=549232, ups=1.11, wpb=495787, bsz=16575.2, num_updates=16600, lr=0.000490881, gnorm=0.237, clip=0, loss_scale=4, train_wall=89, gb_free=22, wall=15364 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1525 / 1689 loss=3.722, nll_loss=2.191, ppl=4.57, wps=548520, ups=1.11, wpb=493411, bsz=16423.4, num_updates=16700, lr=0.000489409, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.2, wall=15454 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 epoch 010: 1626 / 1689 loss=3.711, nll_loss=2.179, ppl=4.53, wps=547242, ups=1.1, wpb=495825, bsz=16272, num_updates=16800, lr=0.00048795, gnorm=0.228, clip=0, loss_scale=4, train_wall=89, gb_free=21.5, wall=15545 end of epoch 10 (average epoch stats below) epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 epoch 010 | loss 3.714 | nll_loss 2.182 | ppl 4.54 | wps 543758 | ups 1.1 | wpb 495120 | bsz 16503.7 | num_updates 16863 | lr 0.000487038 | gnorm 0.237 | clip 0 | loss_scale 4 | train_wall 1498 | gb_free 23.6 | wall 15601 Start iterating over samples epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 37 / 1689 loss=3.701, nll_loss=2.167, ppl=4.49, wps=547940, ups=1.12, wpb=490586, bsz=16064, num_updates=16900, lr=0.000486504, gnorm=0.235, clip=0, loss_scale=4, train_wall=87, gb_free=21.7, wall=15634 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 epoch 011: 137 / 1689 loss=3.687, nll_loss=2.151, ppl=4.44, wps=548568, ups=1.11, wpb=496275, bsz=16701.5, num_updates=17000, lr=0.000485071, gnorm=0.229, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=15725 begin validation on "valid" subset epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011 | valid on 'valid' subset | loss 3.777 | nll_loss 2.219 | ppl 4.66 | wps 0 | wpb 44526 | bsz 2008 | num_updates 17000 | best_loss 3.769 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 237 / 1689 loss=3.686, nll_loss=2.15, ppl=4.44, wps=403808, ups=0.81, wpb=495725, bsz=16513.3, num_updates=17100, lr=0.000483651, gnorm=0.233, clip=0, loss_scale=4, train_wall=93, gb_free=21.9, wall=15848 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 337 / 1689 loss=3.696, nll_loss=2.162, ppl=4.48, wps=560439, ups=1.13, wpb=496083, bsz=16943.4, num_updates=17200, lr=0.000482243, gnorm=0.233, clip=0, loss_scale=4, train_wall=88, gb_free=22, wall=15936 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 437 / 1689 loss=3.703, nll_loss=2.17, ppl=4.5, wps=559962, ups=1.13, wpb=493728, bsz=16388.2, num_updates=17300, lr=0.000480847, gnorm=0.231, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=16024 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 538 / 1689 loss=3.704, nll_loss=2.171, ppl=4.5, wps=549782, ups=1.11, wpb=493748, bsz=16621.9, num_updates=17400, lr=0.000479463, gnorm=0.223, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=16114 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 639 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=555312, ups=1.12, wpb=496719, bsz=16222.2, num_updates=17500, lr=0.000478091, gnorm=0.236, clip=0, loss_scale=2, train_wall=88, gb_free=20.6, wall=16204 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 739 / 1689 loss=3.699, nll_loss=2.165, ppl=4.49, wps=551571, ups=1.11, wpb=494728, bsz=16490.2, num_updates=17600, lr=0.000476731, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16293 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 839 / 1689 loss=3.694, nll_loss=2.16, ppl=4.47, wps=556106, ups=1.12, wpb=495642, bsz=16615, num_updates=17700, lr=0.000475383, gnorm=0.225, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=16382 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 939 / 1689 loss=3.693, nll_loss=2.158, ppl=4.46, wps=551732, ups=1.11, wpb=496919, bsz=16518.8, num_updates=17800, lr=0.000474045, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=16472 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1039 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=547356, ups=1.11, wpb=493359, bsz=16746.8, num_updates=17900, lr=0.000472719, gnorm=0.219, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=16563 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 epoch 011: 1139 / 1689 loss=3.702, nll_loss=2.17, ppl=4.5, wps=553258, ups=1.12, wpb=496117, bsz=16615.5, num_updates=18000, lr=0.000471405, gnorm=0.244, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=16652 begin validation on "valid" subset epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011 | valid on 'valid' subset | loss 3.767 | nll_loss 2.205 | ppl 4.61 | wps 0 | wpb 44526 | bsz 2008 | num_updates 18000 | best_loss 3.767 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1240 / 1689 loss=3.697, nll_loss=2.163, ppl=4.48, wps=425664, ups=0.86, wpb=495213, bsz=16693, num_updates=18100, lr=0.0004701, gnorm=0.235, clip=0, loss_scale=2, train_wall=95, gb_free=22.4, wall=16769 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1340 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=552909, ups=1.11, wpb=495931, bsz=16401, num_updates=18200, lr=0.000468807, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=16858 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1440 / 1689 loss=3.701, nll_loss=2.168, ppl=4.5, wps=551372, ups=1.12, wpb=493728, bsz=16125.3, num_updates=18300, lr=0.000467525, gnorm=0.23, clip=0, loss_scale=2, train_wall=88, gb_free=21.3, wall=16948 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1540 / 1689 loss=3.702, nll_loss=2.169, ppl=4.5, wps=551227, ups=1.11, wpb=495044, bsz=16039.4, num_updates=18400, lr=0.000466252, gnorm=0.238, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=17038 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 epoch 011: 1640 / 1689 loss=3.696, nll_loss=2.163, ppl=4.48, wps=553376, ups=1.11, wpb=496810, bsz=16394.2, num_updates=18500, lr=0.000464991, gnorm=0.232, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=17127 end of epoch 11 (average epoch stats below) epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 epoch 011 | loss 3.697 | nll_loss 2.163 | ppl 4.48 | wps 531477 | ups 1.07 | wpb 495132 | bsz 16501.7 | num_updates 18549 | lr 0.000464376 | gnorm 0.23 | clip 0 | loss_scale 2 | train_wall 1498 | gb_free 22.9 | wall 17171 Start iterating over samples epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 51 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=542314, ups=1.1, wpb=491031, bsz=16377, num_updates=18600, lr=0.000463739, gnorm=0.221, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=17218 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 151 / 1689 loss=3.667, nll_loss=2.129, ppl=4.37, wps=552798, ups=1.11, wpb=496225, bsz=16312.6, num_updates=18700, lr=0.000462497, gnorm=0.221, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=17308 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 251 / 1689 loss=3.677, nll_loss=2.141, ppl=4.41, wps=552324, ups=1.12, wpb=494887, bsz=16667.9, num_updates=18800, lr=0.000461266, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=21.4, wall=17397 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 351 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=549141, ups=1.11, wpb=494357, bsz=16474.9, num_updates=18900, lr=0.000460044, gnorm=0.227, clip=0, loss_scale=4, train_wall=88, gb_free=20.8, wall=17487 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 epoch 012: 451 / 1689 loss=3.682, nll_loss=2.147, ppl=4.43, wps=551975, ups=1.12, wpb=494311, bsz=16398.4, num_updates=19000, lr=0.000458831, gnorm=0.228, clip=0, loss_scale=4, train_wall=88, gb_free=22.2, wall=17577 begin validation on "valid" subset epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012 | valid on 'valid' subset | loss 3.76 | nll_loss 2.2 | ppl 4.6 | wps 0 | wpb 44526 | bsz 2008 | num_updates 19000 | best_loss 3.76 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 552 / 1689 loss=3.68, nll_loss=2.144, ppl=4.42, wps=339801, ups=0.69, wpb=495970, bsz=16423.6, num_updates=19100, lr=0.000457629, gnorm=0.212, clip=0, loss_scale=4, train_wall=120, gb_free=21.4, wall=17723 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 653 / 1689 loss=3.679, nll_loss=2.143, ppl=4.42, wps=550959, ups=1.11, wpb=496772, bsz=16706.1, num_updates=19200, lr=0.000456435, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=17813 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 753 / 1689 loss=3.686, nll_loss=2.151, ppl=4.44, wps=551771, ups=1.11, wpb=495899, bsz=16454.3, num_updates=19300, lr=0.000455251, gnorm=0.234, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=17903 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 853 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=546849, ups=1.1, wpb=495923, bsz=16741.9, num_updates=19400, lr=0.000454077, gnorm=0.219, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=17994 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 953 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555086, ups=1.12, wpb=495202, bsz=16724.2, num_updates=19500, lr=0.000452911, gnorm=0.223, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=18083 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1053 / 1689 loss=3.683, nll_loss=2.148, ppl=4.43, wps=564188, ups=1.14, wpb=495498, bsz=16036.2, num_updates=19600, lr=0.000451754, gnorm=0.217, clip=0, loss_scale=2, train_wall=87, gb_free=19.1, wall=18171 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1153 / 1689 loss=3.684, nll_loss=2.149, ppl=4.43, wps=557502, ups=1.12, wpb=495890, bsz=16418, num_updates=19700, lr=0.000450606, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18260 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1253 / 1689 loss=3.685, nll_loss=2.151, ppl=4.44, wps=554342, ups=1.12, wpb=494408, bsz=16744.4, num_updates=19800, lr=0.000449467, gnorm=0.227, clip=0, loss_scale=4, train_wall=89, gb_free=22.4, wall=18349 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1353 / 1689 loss=3.688, nll_loss=2.154, ppl=4.45, wps=556032, ups=1.12, wpb=495301, bsz=16893, num_updates=19900, lr=0.000448336, gnorm=0.226, clip=0, loss_scale=4, train_wall=88, gb_free=21.8, wall=18438 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 epoch 012: 1453 / 1689 loss=3.689, nll_loss=2.155, ppl=4.45, wps=555491, ups=1.12, wpb=494568, bsz=16294.8, num_updates=20000, lr=0.000447214, gnorm=0.218, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18527 begin validation on "valid" subset epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012 | valid on 'valid' subset | loss 3.744 | nll_loss 2.189 | ppl 4.56 | wps 0 | wpb 44526 | bsz 2008 | num_updates 20000 | best_loss 3.744 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1553 / 1689 loss=3.684, nll_loss=2.149, ppl=4.44, wps=425601, ups=0.86, wpb=495443, bsz=16580.7, num_updates=20100, lr=0.0004461, gnorm=0.213, clip=0, loss_scale=4, train_wall=94, gb_free=22, wall=18643 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 epoch 012: 1654 / 1689 loss=3.687, nll_loss=2.153, ppl=4.45, wps=555188, ups=1.12, wpb=496046, bsz=16551.5, num_updates=20200, lr=0.000444994, gnorm=0.224, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=18733 end of epoch 12 (average epoch stats below) epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 epoch 012 | loss 3.682 | nll_loss 2.147 | ppl 4.43 | wps 524217 | ups 1.06 | wpb 495112 | bsz 16505.5 | num_updates 20235 | lr 0.000444609 | gnorm 0.222 | clip 0 | loss_scale 4 | train_wall 1524 | gb_free 23 | wall 18764 Start iterating over samples epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 65 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=544192, ups=1.11, wpb=491838, bsz=16115.6, num_updates=20300, lr=0.000443897, gnorm=0.229, clip=0, loss_scale=4, train_wall=87, gb_free=22, wall=18823 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 166 / 1689 loss=3.658, nll_loss=2.12, ppl=4.35, wps=552332, ups=1.12, wpb=494968, bsz=16340.2, num_updates=20400, lr=0.000442807, gnorm=0.227, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=18913 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 266 / 1689 loss=3.657, nll_loss=2.119, ppl=4.34, wps=549528, ups=1.11, wpb=495013, bsz=16496, num_updates=20500, lr=0.000441726, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=19003 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 366 / 1689 loss=3.664, nll_loss=2.127, ppl=4.37, wps=546759, ups=1.11, wpb=494323, bsz=16842.7, num_updates=20600, lr=0.000440653, gnorm=0.208, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=19093 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 466 / 1689 loss=3.672, nll_loss=2.136, ppl=4.39, wps=545630, ups=1.1, wpb=494067, bsz=16499.2, num_updates=20700, lr=0.000439587, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=19184 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 566 / 1689 loss=3.668, nll_loss=2.131, ppl=4.38, wps=556272, ups=1.12, wpb=495680, bsz=16061.8, num_updates=20800, lr=0.000438529, gnorm=0.205, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=19273 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 666 / 1689 loss=3.671, nll_loss=2.134, ppl=4.39, wps=550184, ups=1.11, wpb=496242, bsz=17018.2, num_updates=20900, lr=0.000437479, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=21.7, wall=19363 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 epoch 013: 766 / 1689 loss=3.673, nll_loss=2.137, ppl=4.4, wps=552396, ups=1.12, wpb=494990, bsz=16509.3, num_updates=21000, lr=0.000436436, gnorm=0.22, clip=0, loss_scale=4, train_wall=88, gb_free=22.8, wall=19453 begin validation on "valid" subset epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013 | valid on 'valid' subset | loss 3.753 | nll_loss 2.197 | ppl 4.58 | wps 0 | wpb 44526 | bsz 2008 | num_updates 21000 | best_loss 3.744 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 868 / 1689 loss=3.67, nll_loss=2.133, ppl=4.39, wps=416245, ups=0.84, wpb=495602, bsz=16780.4, num_updates=21100, lr=0.0004354, gnorm=0.223, clip=0, loss_scale=1, train_wall=92, gb_free=21.7, wall=19572 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 968 / 1689 loss=3.665, nll_loss=2.128, ppl=4.37, wps=559679, ups=1.13, wpb=496294, bsz=16492.2, num_updates=21200, lr=0.000434372, gnorm=0.21, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=19660 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1068 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=555952, ups=1.12, wpb=495287, bsz=16471.4, num_updates=21300, lr=0.000433351, gnorm=0.233, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=19750 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1168 / 1689 loss=3.678, nll_loss=2.142, ppl=4.41, wps=556483, ups=1.13, wpb=494232, bsz=16459.7, num_updates=21400, lr=0.000432338, gnorm=0.222, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=19838 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1268 / 1689 loss=3.676, nll_loss=2.14, ppl=4.41, wps=566370, ups=1.14, wpb=495556, bsz=16234, num_updates=21500, lr=0.000431331, gnorm=0.223, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=19926 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1368 / 1689 loss=3.671, nll_loss=2.135, ppl=4.39, wps=552644, ups=1.12, wpb=494992, bsz=16767.5, num_updates=21600, lr=0.000430331, gnorm=0.213, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=20015 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1468 / 1689 loss=3.665, nll_loss=2.129, ppl=4.37, wps=552432, ups=1.11, wpb=495675, bsz=16767, num_updates=21700, lr=0.000429339, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=20105 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1568 / 1689 loss=3.678, nll_loss=2.144, ppl=4.42, wps=555013, ups=1.12, wpb=495073, bsz=16437, num_updates=21800, lr=0.000428353, gnorm=0.221, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=20194 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 epoch 013: 1668 / 1689 loss=3.672, nll_loss=2.136, ppl=4.4, wps=552943, ups=1.12, wpb=495714, bsz=16411.5, num_updates=21900, lr=0.000427374, gnorm=0.218, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=20284 end of epoch 13 (average epoch stats below) epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 epoch 013 | loss 3.669 | nll_loss 2.132 | ppl 4.38 | wps 542565 | ups 1.1 | wpb 495104 | bsz 16507.7 | num_updates 21921 | lr 0.000427169 | gnorm 0.218 | clip 0 | loss_scale 2 | train_wall 1490 | gb_free 23.2 | wall 20302 Start iterating over samples epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 epoch 014: 79 / 1689 loss=3.65, nll_loss=2.111, ppl=4.32, wps=543663, ups=1.1, wpb=492338, bsz=16229.8, num_updates=22000, lr=0.000426401, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=20375 begin validation on "valid" subset epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014 | valid on 'valid' subset | loss 3.744 | nll_loss 2.185 | ppl 4.55 | wps 0 | wpb 44526 | bsz 2008 | num_updates 22000 | best_loss 3.744 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 179 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=457826, ups=0.92, wpb=497567, bsz=16512.9, num_updates=22100, lr=0.000425436, gnorm=0.21, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20483 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 279 / 1689 loss=3.649, nll_loss=2.11, ppl=4.32, wps=553643, ups=1.11, wpb=496704, bsz=16381.4, num_updates=22200, lr=0.000424476, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.3, wall=20573 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 379 / 1689 loss=3.652, nll_loss=2.114, ppl=4.33, wps=552484, ups=1.11, wpb=495870, bsz=16728.3, num_updates=22300, lr=0.000423524, gnorm=0.216, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20663 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 479 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=552003, ups=1.12, wpb=494989, bsz=16530.9, num_updates=22400, lr=0.000422577, gnorm=0.217, clip=0, loss_scale=4, train_wall=88, gb_free=22.4, wall=20752 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 580 / 1689 loss=3.655, nll_loss=2.117, ppl=4.34, wps=548709, ups=1.11, wpb=495680, bsz=16455.3, num_updates=22500, lr=0.000421637, gnorm=0.217, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=20843 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 680 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=554764, ups=1.12, wpb=494082, bsz=16366.4, num_updates=22600, lr=0.000420703, gnorm=0.209, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=20932 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 780 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=544841, ups=1.1, wpb=494848, bsz=16441, num_updates=22700, lr=0.000419775, gnorm=0.215, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=21023 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 880 / 1689 loss=3.656, nll_loss=2.118, ppl=4.34, wps=548808, ups=1.11, wpb=495581, bsz=16610.2, num_updates=22800, lr=0.000418854, gnorm=0.214, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=21113 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 980 / 1689 loss=3.66, nll_loss=2.123, ppl=4.36, wps=551478, ups=1.12, wpb=494528, bsz=16750, num_updates=22900, lr=0.000417938, gnorm=0.217, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=21203 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 epoch 014: 1080 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=548638, ups=1.11, wpb=494801, bsz=16575.7, num_updates=23000, lr=0.000417029, gnorm=0.218, clip=0, loss_scale=2, train_wall=89, gb_free=21.7, wall=21293 begin validation on "valid" subset epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014 | valid on 'valid' subset | loss 3.733 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 23000 | best_loss 3.733 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1180 / 1689 loss=3.661, nll_loss=2.124, ppl=4.36, wps=397977, ups=0.81, wpb=493877, bsz=16457.9, num_updates=23100, lr=0.000416125, gnorm=0.217, clip=0, loss_scale=4, train_wall=101, gb_free=21.9, wall=21417 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1280 / 1689 loss=3.662, nll_loss=2.126, ppl=4.36, wps=559050, ups=1.13, wpb=496123, bsz=16284.1, num_updates=23200, lr=0.000415227, gnorm=0.208, clip=0, loss_scale=4, train_wall=87, gb_free=22.5, wall=21506 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1380 / 1689 loss=3.664, nll_loss=2.128, ppl=4.37, wps=551720, ups=1.11, wpb=495380, bsz=16312.1, num_updates=23300, lr=0.000414335, gnorm=0.205, clip=0, loss_scale=4, train_wall=88, gb_free=22.6, wall=21595 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1480 / 1689 loss=3.663, nll_loss=2.126, ppl=4.37, wps=546286, ups=1.1, wpb=495087, bsz=16655, num_updates=23400, lr=0.000413449, gnorm=0.201, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=21686 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1580 / 1689 loss=3.654, nll_loss=2.116, ppl=4.33, wps=551305, ups=1.11, wpb=496824, bsz=16741, num_updates=23500, lr=0.000412568, gnorm=0.22, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21776 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 epoch 014: 1681 / 1689 loss=3.67, nll_loss=2.134, ppl=4.39, wps=544296, ups=1.1, wpb=493548, bsz=16443, num_updates=23600, lr=0.000411693, gnorm=0.211, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=21867 end of epoch 14 (average epoch stats below) epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 epoch 014 | loss 3.657 | nll_loss 2.119 | ppl 4.34 | wps 531671 | ups 1.07 | wpb 495130 | bsz 16505.8 | num_updates 23608 | lr 0.000411624 | gnorm 0.213 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 23.9 | wall 21873 Start iterating over samples epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 93 / 1689 loss=3.633, nll_loss=2.091, ppl=4.26, wps=538138, ups=1.1, wpb=490339, bsz=16270, num_updates=23700, lr=0.000410824, gnorm=0.224, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=21958 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 193 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=555500, ups=1.12, wpb=494696, bsz=16409.6, num_updates=23800, lr=0.00040996, gnorm=0.207, clip=0, loss_scale=2, train_wall=87, gb_free=21.4, wall=22047 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 293 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=549778, ups=1.11, wpb=495433, bsz=16520.5, num_updates=23900, lr=0.000409101, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22137 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 epoch 015: 393 / 1689 loss=3.641, nll_loss=2.101, ppl=4.29, wps=548904, ups=1.11, wpb=496466, bsz=16830.1, num_updates=24000, lr=0.000408248, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22228 begin validation on "valid" subset epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015 | valid on 'valid' subset | loss 3.74 | nll_loss 2.182 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 24000 | best_loss 3.733 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 493 / 1689 loss=3.641, nll_loss=2.102, ppl=4.29, wps=490957, ups=0.99, wpb=495617, bsz=16743.2, num_updates=24100, lr=0.0004074, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=22329 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 594 / 1689 loss=3.645, nll_loss=2.105, ppl=4.3, wps=549205, ups=1.11, wpb=494299, bsz=16332.8, num_updates=24200, lr=0.000406558, gnorm=0.213, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=22419 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 694 / 1689 loss=3.644, nll_loss=2.105, ppl=4.3, wps=546646, ups=1.1, wpb=495018, bsz=16705.3, num_updates=24300, lr=0.00040572, gnorm=0.211, clip=0, loss_scale=2, train_wall=89, gb_free=22.4, wall=22509 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 794 / 1689 loss=3.647, nll_loss=2.108, ppl=4.31, wps=555385, ups=1.12, wpb=495342, bsz=16290.8, num_updates=24400, lr=0.000404888, gnorm=0.203, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22598 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 894 / 1689 loss=3.649, nll_loss=2.111, ppl=4.32, wps=554052, ups=1.12, wpb=494458, bsz=16219.9, num_updates=24500, lr=0.000404061, gnorm=0.204, clip=0, loss_scale=2, train_wall=87, gb_free=21.9, wall=22688 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 994 / 1689 loss=3.653, nll_loss=2.115, ppl=4.33, wps=556170, ups=1.12, wpb=497488, bsz=16258.6, num_updates=24600, lr=0.000403239, gnorm=0.205, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=22777 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1095 / 1689 loss=3.645, nll_loss=2.106, ppl=4.31, wps=547009, ups=1.1, wpb=495181, bsz=16749.1, num_updates=24700, lr=0.000402422, gnorm=0.21, clip=0, loss_scale=2, train_wall=89, gb_free=22.8, wall=22868 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1195 / 1689 loss=3.659, nll_loss=2.122, ppl=4.35, wps=553512, ups=1.12, wpb=493832, bsz=16218.3, num_updates=24800, lr=0.00040161, gnorm=0.216, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=22957 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1295 / 1689 loss=3.643, nll_loss=2.105, ppl=4.3, wps=550145, ups=1.11, wpb=495357, bsz=17017.6, num_updates=24900, lr=0.000400802, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=23047 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 epoch 015: 1395 / 1689 loss=3.648, nll_loss=2.11, ppl=4.32, wps=552247, ups=1.11, wpb=497008, bsz=16499, num_updates=25000, lr=0.0004, gnorm=0.209, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=23137 begin validation on "valid" subset epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015 | valid on 'valid' subset | loss 3.726 | nll_loss 2.163 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 25000 | best_loss 3.726 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1495 / 1689 loss=3.65, nll_loss=2.112, ppl=4.32, wps=388341, ups=0.78, wpb=494793, bsz=16897.5, num_updates=25100, lr=0.000399202, gnorm=0.208, clip=0, loss_scale=2, train_wall=103, gb_free=22.3, wall=23264 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 epoch 015: 1595 / 1689 loss=3.657, nll_loss=2.12, ppl=4.35, wps=550160, ups=1.11, wpb=496171, bsz=16480.2, num_updates=25200, lr=0.00039841, gnorm=0.217, clip=0, loss_scale=4, train_wall=89, gb_free=21.7, wall=23354 end of epoch 15 (average epoch stats below) epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 epoch 015 | loss 3.646 | nll_loss 2.107 | ppl 4.31 | wps 533888 | ups 1.08 | wpb 495128 | bsz 16505.7 | num_updates 25294 | lr 0.000397669 | gnorm 0.208 | clip 0 | loss_scale 4 | train_wall 1502 | gb_free 24.3 | wall 23437 Start iterating over samples epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 6 / 1689 loss=3.656, nll_loss=2.119, ppl=4.34, wps=379269, ups=0.77, wpb=491657, bsz=16028.5, num_updates=25300, lr=0.000397621, gnorm=0.2, clip=0, loss_scale=4, train_wall=101, gb_free=20.9, wall=23484 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 107 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=540462, ups=1.09, wpb=495078, bsz=16349.9, num_updates=25400, lr=0.000396838, gnorm=0.204, clip=0, loss_scale=2, train_wall=90, gb_free=22, wall=23576 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 207 / 1689 loss=3.626, nll_loss=2.084, ppl=4.24, wps=556738, ups=1.12, wpb=496509, bsz=16657, num_updates=25500, lr=0.000396059, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=23665 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 307 / 1689 loss=3.633, nll_loss=2.092, ppl=4.26, wps=554392, ups=1.12, wpb=493618, bsz=16222.7, num_updates=25600, lr=0.000395285, gnorm=0.208, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=23754 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 407 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=562903, ups=1.14, wpb=494967, bsz=16209.4, num_updates=25700, lr=0.000394515, gnorm=0.202, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=23842 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 507 / 1689 loss=3.637, nll_loss=2.097, ppl=4.28, wps=557921, ups=1.12, wpb=496125, bsz=16169.8, num_updates=25800, lr=0.00039375, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.4, wall=23931 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 607 / 1689 loss=3.633, nll_loss=2.093, ppl=4.27, wps=554518, ups=1.12, wpb=494840, bsz=16623.5, num_updates=25900, lr=0.000392989, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=24020 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 epoch 016: 708 / 1689 loss=3.636, nll_loss=2.096, ppl=4.28, wps=549437, ups=1.11, wpb=494332, bsz=16620.9, num_updates=26000, lr=0.000392232, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=24110 begin validation on "valid" subset epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016 | valid on 'valid' subset | loss 3.734 | nll_loss 2.181 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 26000 | best_loss 3.726 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 808 / 1689 loss=3.639, nll_loss=2.099, ppl=4.28, wps=486781, ups=0.98, wpb=494230, bsz=16607.3, num_updates=26100, lr=0.00039148, gnorm=0.204, clip=0, loss_scale=2, train_wall=89, gb_free=21.9, wall=24211 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 908 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=549591, ups=1.11, wpb=495695, bsz=16647.6, num_updates=26200, lr=0.000390732, gnorm=0.203, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=24302 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1008 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=550403, ups=1.11, wpb=495793, bsz=16314, num_updates=26300, lr=0.000389989, gnorm=0.198, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=24392 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1108 / 1689 loss=3.639, nll_loss=2.101, ppl=4.29, wps=553703, ups=1.12, wpb=496248, bsz=16571.3, num_updates=26400, lr=0.000389249, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=24481 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1208 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=550782, ups=1.11, wpb=495003, bsz=17110.2, num_updates=26500, lr=0.000388514, gnorm=0.203, clip=0, loss_scale=4, train_wall=88, gb_free=21.1, wall=24571 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1308 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=547768, ups=1.1, wpb=496281, bsz=16738, num_updates=26600, lr=0.000387783, gnorm=0.202, clip=0, loss_scale=4, train_wall=89, gb_free=22.1, wall=24662 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1409 / 1689 loss=3.638, nll_loss=2.098, ppl=4.28, wps=552264, ups=1.11, wpb=495328, bsz=16379.5, num_updates=26700, lr=0.000387056, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=24752 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1509 / 1689 loss=3.641, nll_loss=2.103, ppl=4.3, wps=548586, ups=1.11, wpb=495249, bsz=16547.4, num_updates=26800, lr=0.000386334, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=24842 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 epoch 016: 1609 / 1689 loss=3.64, nll_loss=2.101, ppl=4.29, wps=554525, ups=1.12, wpb=496717, bsz=16258.9, num_updates=26900, lr=0.000385615, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=24931 end of epoch 16 (average epoch stats below) epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 epoch 016 | loss 3.636 | nll_loss 2.097 | ppl 4.28 | wps 533219 | ups 1.08 | wpb 495123 | bsz 16505 | num_updates 26980 | lr 0.000385043 | gnorm 0.202 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 23.3 | wall 25002 Start iterating over samples epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 epoch 017: 20 / 1689 loss=3.639, nll_loss=2.1, ppl=4.29, wps=529997, ups=1.08, wpb=490647, bsz=16599.9, num_updates=27000, lr=0.0003849, gnorm=0.204, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=25024 begin validation on "valid" subset epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017 | valid on 'valid' subset | loss 3.741 | nll_loss 2.184 | ppl 4.54 | wps 0 | wpb 44526 | bsz 2008 | num_updates 27000 | best_loss 3.726 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 120 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=427445, ups=0.86, wpb=496106, bsz=16510.4, num_updates=27100, lr=0.000384189, gnorm=0.196, clip=0, loss_scale=2, train_wall=86, gb_free=21.2, wall=25140 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 221 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=561450, ups=1.13, wpb=495241, bsz=16307.8, num_updates=27200, lr=0.000383482, gnorm=0.21, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=25228 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 321 / 1689 loss=3.618, nll_loss=2.075, ppl=4.21, wps=552663, ups=1.12, wpb=494318, bsz=16494.1, num_updates=27300, lr=0.00038278, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=25318 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 421 / 1689 loss=3.62, nll_loss=2.078, ppl=4.22, wps=556497, ups=1.12, wpb=495854, bsz=16600.5, num_updates=27400, lr=0.00038208, gnorm=0.202, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=25407 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 521 / 1689 loss=3.632, nll_loss=2.092, ppl=4.26, wps=558934, ups=1.13, wpb=494365, bsz=16416, num_updates=27500, lr=0.000381385, gnorm=0.206, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=25495 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 621 / 1689 loss=3.629, nll_loss=2.088, ppl=4.25, wps=551691, ups=1.12, wpb=493574, bsz=16241.4, num_updates=27600, lr=0.000380693, gnorm=0.201, clip=0, loss_scale=2, train_wall=88, gb_free=22.6, wall=25585 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 722 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=546437, ups=1.1, wpb=495195, bsz=16309.9, num_updates=27700, lr=0.000380006, gnorm=0.199, clip=0, loss_scale=2, train_wall=90, gb_free=21.8, wall=25675 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 822 / 1689 loss=3.631, nll_loss=2.091, ppl=4.26, wps=554065, ups=1.11, wpb=496930, bsz=17043.4, num_updates=27800, lr=0.000379322, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=21, wall=25765 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 922 / 1689 loss=3.628, nll_loss=2.087, ppl=4.25, wps=553236, ups=1.12, wpb=495411, bsz=16672.9, num_updates=27900, lr=0.000378641, gnorm=0.21, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25855 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 epoch 017: 1022 / 1689 loss=3.633, nll_loss=2.094, ppl=4.27, wps=551384, ups=1.11, wpb=494784, bsz=16371.9, num_updates=28000, lr=0.000377964, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=25944 begin validation on "valid" subset epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017 | valid on 'valid' subset | loss 3.723 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 28000 | best_loss 3.723 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1122 / 1689 loss=3.627, nll_loss=2.087, ppl=4.25, wps=453395, ups=0.91, wpb=496626, bsz=16690.5, num_updates=28100, lr=0.000377291, gnorm=0.2, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=26054 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1223 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=550150, ups=1.11, wpb=495251, bsz=16369.3, num_updates=28200, lr=0.000376622, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=26144 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1323 / 1689 loss=3.63, nll_loss=2.09, ppl=4.26, wps=551651, ups=1.11, wpb=494783, bsz=16410.8, num_updates=28300, lr=0.000375956, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26234 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1423 / 1689 loss=3.629, nll_loss=2.089, ppl=4.25, wps=551430, ups=1.11, wpb=495478, bsz=16525.7, num_updates=28400, lr=0.000375293, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=26323 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1523 / 1689 loss=3.629, nll_loss=2.089, ppl=4.26, wps=552186, ups=1.11, wpb=496729, bsz=16691.4, num_updates=28500, lr=0.000374634, gnorm=0.209, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=26413 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 epoch 017: 1623 / 1689 loss=3.636, nll_loss=2.097, ppl=4.28, wps=548581, ups=1.11, wpb=495626, bsz=16864.9, num_updates=28600, lr=0.000373979, gnorm=0.206, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=26504 end of epoch 17 (average epoch stats below) epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 epoch 017 | loss 3.628 | nll_loss 2.087 | ppl 4.25 | wps 535295 | ups 1.08 | wpb 495114 | bsz 16508.2 | num_updates 28666 | lr 0.000373548 | gnorm 0.2 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 21.4 | wall 26562 Start iterating over samples epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 34 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=548932, ups=1.12, wpb=490199, bsz=16047, num_updates=28700, lr=0.000373327, gnorm=0.2, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=26593 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 134 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=546941, ups=1.1, wpb=494983, bsz=16423.6, num_updates=28800, lr=0.000372678, gnorm=0.196, clip=0, loss_scale=2, train_wall=89, gb_free=21.3, wall=26684 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 234 / 1689 loss=3.618, nll_loss=2.076, ppl=4.22, wps=551203, ups=1.11, wpb=495985, bsz=16379.2, num_updates=28900, lr=0.000372033, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=26774 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 epoch 018: 335 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=547877, ups=1.1, wpb=495861, bsz=16514.2, num_updates=29000, lr=0.000371391, gnorm=0.205, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=26864 begin validation on "valid" subset epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.729 | nll_loss 2.175 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 29000 | best_loss 3.723 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 435 / 1689 loss=3.608, nll_loss=2.065, ppl=4.18, wps=486197, ups=0.98, wpb=498460, bsz=16628.4, num_updates=29100, lr=0.000370752, gnorm=0.208, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=26967 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 535 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556958, ups=1.12, wpb=495571, bsz=16509.4, num_updates=29200, lr=0.000370117, gnorm=0.201, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=27056 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 635 / 1689 loss=3.617, nll_loss=2.075, ppl=4.21, wps=557302, ups=1.12, wpb=495853, bsz=16322.2, num_updates=29300, lr=0.000369484, gnorm=0.207, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=27145 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 735 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=557192, ups=1.12, wpb=495670, bsz=16543, num_updates=29400, lr=0.000368856, gnorm=0.199, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=27233 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 835 / 1689 loss=3.62, nll_loss=2.079, ppl=4.22, wps=556020, ups=1.12, wpb=494889, bsz=16930.9, num_updates=29500, lr=0.00036823, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=27322 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 935 / 1689 loss=3.621, nll_loss=2.08, ppl=4.23, wps=563140, ups=1.13, wpb=496166, bsz=16399.5, num_updates=29600, lr=0.000367607, gnorm=0.19, clip=0, loss_scale=2, train_wall=87, gb_free=21.8, wall=27411 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1035 / 1689 loss=3.624, nll_loss=2.083, ppl=4.24, wps=558589, ups=1.13, wpb=494349, bsz=16325.8, num_updates=29700, lr=0.000366988, gnorm=0.196, clip=0, loss_scale=2, train_wall=88, gb_free=22.9, wall=27499 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1135 / 1689 loss=3.619, nll_loss=2.077, ppl=4.22, wps=553011, ups=1.12, wpb=495762, bsz=16620, num_updates=29800, lr=0.000366372, gnorm=0.202, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=27589 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1236 / 1689 loss=3.619, nll_loss=2.078, ppl=4.22, wps=547095, ups=1.1, wpb=495469, bsz=16643, num_updates=29900, lr=0.000365758, gnorm=0.193, clip=0, loss_scale=1, train_wall=90, gb_free=21.4, wall=27679 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 epoch 018: 1336 / 1689 loss=3.62, nll_loss=2.079, ppl=4.23, wps=551267, ups=1.11, wpb=495090, bsz=16285.4, num_updates=30000, lr=0.000365148, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=27769 begin validation on "valid" subset epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018 | valid on 'valid' subset | loss 3.726 | nll_loss 2.168 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 30000 | best_loss 3.723 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1436 / 1689 loss=3.627, nll_loss=2.086, ppl=4.25, wps=482922, ups=0.98, wpb=495217, bsz=16364.5, num_updates=30100, lr=0.000364541, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=27872 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1536 / 1689 loss=3.628, nll_loss=2.088, ppl=4.25, wps=558752, ups=1.13, wpb=495062, bsz=16302.6, num_updates=30200, lr=0.000363937, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=27960 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 epoch 018: 1636 / 1689 loss=3.636, nll_loss=2.098, ppl=4.28, wps=553735, ups=1.13, wpb=492165, bsz=16866.5, num_updates=30300, lr=0.000363336, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=28049 end of epoch 18 (average epoch stats below) epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 epoch 018 | loss 3.62 | nll_loss 2.078 | ppl 4.22 | wps 544636 | ups 1.1 | wpb 495124 | bsz 16505.2 | num_updates 30353 | lr 0.000363019 | gnorm 0.198 | clip 0 | loss_scale 2 | train_wall 1486 | gb_free 22.9 | wall 28095 Start iterating over samples epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 47 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=546306, ups=1.11, wpb=492154, bsz=16531.8, num_updates=30400, lr=0.000362738, gnorm=0.192, clip=0, loss_scale=2, train_wall=87, gb_free=22.1, wall=28139 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 147 / 1689 loss=3.6, nll_loss=2.056, ppl=4.16, wps=554742, ups=1.12, wpb=496311, bsz=16890.3, num_updates=30500, lr=0.000362143, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=28229 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 248 / 1689 loss=3.604, nll_loss=2.06, ppl=4.17, wps=548745, ups=1.1, wpb=496675, bsz=16538.4, num_updates=30600, lr=0.000361551, gnorm=0.201, clip=0, loss_scale=1, train_wall=89, gb_free=21.2, wall=28319 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 348 / 1689 loss=3.605, nll_loss=2.061, ppl=4.17, wps=550214, ups=1.11, wpb=495948, bsz=16655.1, num_updates=30700, lr=0.000360961, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=28409 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 448 / 1689 loss=3.607, nll_loss=2.064, ppl=4.18, wps=555866, ups=1.12, wpb=495919, bsz=16096.3, num_updates=30800, lr=0.000360375, gnorm=0.198, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=28499 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 548 / 1689 loss=3.614, nll_loss=2.072, ppl=4.2, wps=555195, ups=1.12, wpb=496495, bsz=16586.9, num_updates=30900, lr=0.000359791, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=28588 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 epoch 019: 648 / 1689 loss=3.614, nll_loss=2.073, ppl=4.21, wps=554377, ups=1.12, wpb=495882, bsz=16894.3, num_updates=31000, lr=0.000359211, gnorm=0.201, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=28677 begin validation on "valid" subset epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.708 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 31000 | best_loss 3.708 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 748 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=456510, ups=0.92, wpb=494963, bsz=16687.4, num_updates=31100, lr=0.000358633, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.2, wall=28786 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 848 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=556849, ups=1.13, wpb=494751, bsz=16400.2, num_updates=31200, lr=0.000358057, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=28875 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 948 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=555003, ups=1.12, wpb=496266, bsz=16263.1, num_updates=31300, lr=0.000357485, gnorm=0.193, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=28964 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1048 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552024, ups=1.12, wpb=494116, bsz=16322.3, num_updates=31400, lr=0.000356915, gnorm=0.197, clip=0, loss_scale=2, train_wall=89, gb_free=22, wall=29054 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1148 / 1689 loss=3.621, nll_loss=2.081, ppl=4.23, wps=549907, ups=1.11, wpb=493697, bsz=16410, num_updates=31500, lr=0.000356348, gnorm=0.194, clip=0, loss_scale=2, train_wall=89, gb_free=21.5, wall=29143 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1249 / 1689 loss=3.624, nll_loss=2.084, ppl=4.24, wps=550644, ups=1.11, wpb=494545, bsz=16791.8, num_updates=31600, lr=0.000355784, gnorm=0.195, clip=0, loss_scale=2, train_wall=89, gb_free=22.2, wall=29233 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1349 / 1689 loss=3.615, nll_loss=2.074, ppl=4.21, wps=552488, ups=1.12, wpb=495434, bsz=16753.5, num_updates=31700, lr=0.000355222, gnorm=0.201, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=29323 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1450 / 1689 loss=3.617, nll_loss=2.076, ppl=4.22, wps=543934, ups=1.1, wpb=495027, bsz=16242, num_updates=31800, lr=0.000354663, gnorm=0.203, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=29414 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1550 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=555574, ups=1.12, wpb=495491, bsz=16212.5, num_updates=31900, lr=0.000354107, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=29503 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 epoch 019: 1650 / 1689 loss=3.62, nll_loss=2.08, ppl=4.23, wps=553081, ups=1.12, wpb=495034, bsz=16641.8, num_updates=32000, lr=0.000353553, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=29593 begin validation on "valid" subset epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 epoch 019 | valid on 'valid' subset | loss 3.718 | nll_loss 2.162 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 32000 | best_loss 3.708 end of epoch 19 (average epoch stats below) epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 epoch 019 | loss 3.612 | nll_loss 2.07 | ppl 4.2 | wps 526566 | ups 1.06 | wpb 495140 | bsz 16508.4 | num_updates 32039 | lr 0.000353338 | gnorm 0.195 | clip 0 | loss_scale 1 | train_wall 1524 | gb_free 23.7 | wall 29681 Start iterating over samples epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 61 / 1689 loss=3.594, nll_loss=2.049, ppl=4.14, wps=342792, ups=0.7, wpb=490993, bsz=16291.7, num_updates=32100, lr=0.000353002, gnorm=0.202, clip=0, loss_scale=1, train_wall=123, gb_free=21.9, wall=29736 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 161 / 1689 loss=3.595, nll_loss=2.05, ppl=4.14, wps=558655, ups=1.13, wpb=495912, bsz=16196.7, num_updates=32200, lr=0.000352454, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=29825 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 261 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=553476, ups=1.12, wpb=495030, bsz=16438.5, num_updates=32300, lr=0.000351908, gnorm=0.206, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=29914 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 362 / 1689 loss=3.592, nll_loss=2.047, ppl=4.13, wps=543176, ups=1.1, wpb=494892, bsz=16596.6, num_updates=32400, lr=0.000351364, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=30005 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 462 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=550994, ups=1.11, wpb=497431, bsz=16695.8, num_updates=32500, lr=0.000350823, gnorm=0.195, clip=0, loss_scale=1, train_wall=89, gb_free=22.4, wall=30096 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 562 / 1689 loss=3.605, nll_loss=2.062, ppl=4.18, wps=552911, ups=1.12, wpb=494670, bsz=16784.1, num_updates=32600, lr=0.000350285, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=30185 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 662 / 1689 loss=3.613, nll_loss=2.071, ppl=4.2, wps=553594, ups=1.12, wpb=495158, bsz=16410.6, num_updates=32700, lr=0.000349749, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30274 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 762 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=545521, ups=1.1, wpb=496788, bsz=16955.4, num_updates=32800, lr=0.000349215, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=20.5, wall=30365 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 862 / 1689 loss=3.607, nll_loss=2.065, ppl=4.18, wps=553872, ups=1.12, wpb=494028, bsz=16410, num_updates=32900, lr=0.000348684, gnorm=0.189, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=30455 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 epoch 020: 963 / 1689 loss=3.614, nll_loss=2.072, ppl=4.21, wps=543460, ups=1.1, wpb=494434, bsz=16440.8, num_updates=33000, lr=0.000348155, gnorm=0.194, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=30546 begin validation on "valid" subset epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020 | valid on 'valid' subset | loss 3.721 | nll_loss 2.165 | ppl 4.49 | wps 0 | wpb 44526 | bsz 2008 | num_updates 33000 | best_loss 3.708 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1063 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=323265, ups=0.65, wpb=495090, bsz=16377.8, num_updates=33100, lr=0.000347629, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=30699 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1163 / 1689 loss=3.603, nll_loss=2.06, ppl=4.17, wps=558722, ups=1.13, wpb=494608, bsz=16617.4, num_updates=33200, lr=0.000347105, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=30787 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1263 / 1689 loss=3.61, nll_loss=2.068, ppl=4.19, wps=559029, ups=1.13, wpb=495067, bsz=16579, num_updates=33300, lr=0.000346583, gnorm=0.189, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=30876 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1363 / 1689 loss=3.616, nll_loss=2.074, ppl=4.21, wps=559234, ups=1.13, wpb=494951, bsz=16409, num_updates=33400, lr=0.000346064, gnorm=0.192, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=30964 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1463 / 1689 loss=3.613, nll_loss=2.072, ppl=4.2, wps=549313, ups=1.11, wpb=494957, bsz=16535.6, num_updates=33500, lr=0.000345547, gnorm=0.191, clip=0, loss_scale=2, train_wall=89, gb_free=21.8, wall=31055 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1563 / 1689 loss=3.612, nll_loss=2.07, ppl=4.2, wps=551694, ups=1.11, wpb=494931, bsz=16194.5, num_updates=33600, lr=0.000345033, gnorm=0.188, clip=0, loss_scale=2, train_wall=89, gb_free=22.1, wall=31144 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 epoch 020: 1663 / 1689 loss=3.61, nll_loss=2.069, ppl=4.19, wps=555247, ups=1.12, wpb=496440, bsz=16587.8, num_updates=33700, lr=0.00034452, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=31234 end of epoch 20 (average epoch stats below) epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 epoch 020 | loss 3.605 | nll_loss 2.063 | ppl 4.18 | wps 530350 | ups 1.07 | wpb 495109 | bsz 16504.9 | num_updates 33726 | lr 0.000344388 | gnorm 0.192 | clip 0 | loss_scale 2 | train_wall 1487 | gb_free 23.5 | wall 31256 Start iterating over samples epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 74 / 1689 loss=3.59, nll_loss=2.045, ppl=4.13, wps=550877, ups=1.12, wpb=492097, bsz=16228.6, num_updates=33800, lr=0.00034401, gnorm=0.199, clip=0, loss_scale=2, train_wall=87, gb_free=21.5, wall=31323 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 174 / 1689 loss=3.589, nll_loss=2.044, ppl=4.12, wps=552144, ups=1.12, wpb=495010, bsz=16470, num_updates=33900, lr=0.000343503, gnorm=0.188, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=31413 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 epoch 021: 276 / 1689 loss=3.594, nll_loss=2.05, ppl=4.14, wps=542998, ups=1.1, wpb=495675, bsz=16536.1, num_updates=34000, lr=0.000342997, gnorm=0.197, clip=0, loss_scale=1, train_wall=89, gb_free=22.5, wall=31504 begin validation on "valid" subset epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.733 | nll_loss 2.174 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 34000 | best_loss 3.708 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 376 / 1689 loss=3.596, nll_loss=2.052, ppl=4.15, wps=483024, ups=0.98, wpb=494287, bsz=16787.7, num_updates=34100, lr=0.000342494, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=31606 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 476 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=552838, ups=1.12, wpb=494601, bsz=16683.7, num_updates=34200, lr=0.000341993, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=31696 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 576 / 1689 loss=3.598, nll_loss=2.054, ppl=4.15, wps=554820, ups=1.12, wpb=494618, bsz=16648.8, num_updates=34300, lr=0.000341494, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=31785 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 676 / 1689 loss=3.602, nll_loss=2.059, ppl=4.17, wps=553058, ups=1.12, wpb=495366, bsz=16543.1, num_updates=34400, lr=0.000340997, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=31874 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 776 / 1689 loss=3.606, nll_loss=2.063, ppl=4.18, wps=553456, ups=1.12, wpb=494624, bsz=16620.2, num_updates=34500, lr=0.000340503, gnorm=0.196, clip=0, loss_scale=2, train_wall=87, gb_free=22.5, wall=31964 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 876 / 1689 loss=3.597, nll_loss=2.054, ppl=4.15, wps=556571, ups=1.12, wpb=495586, bsz=16262.5, num_updates=34600, lr=0.00034001, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32053 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 976 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=553640, ups=1.11, wpb=496633, bsz=16287, num_updates=34700, lr=0.00033952, gnorm=0.199, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=32143 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1077 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=550508, ups=1.11, wpb=494916, bsz=16432.2, num_updates=34800, lr=0.000339032, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32232 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1177 / 1689 loss=3.603, nll_loss=2.061, ppl=4.17, wps=551854, ups=1.11, wpb=495410, bsz=16737, num_updates=34900, lr=0.000338546, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=32322 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 epoch 021: 1277 / 1689 loss=3.604, nll_loss=2.062, ppl=4.17, wps=551054, ups=1.11, wpb=494735, bsz=16685.6, num_updates=35000, lr=0.000338062, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=32412 begin validation on "valid" subset epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021 | valid on 'valid' subset | loss 3.717 | nll_loss 2.161 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 35000 | best_loss 3.708 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1377 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=481903, ups=0.97, wpb=495253, bsz=16146.5, num_updates=35100, lr=0.00033758, gnorm=0.195, clip=0, loss_scale=1, train_wall=88, gb_free=22.7, wall=32515 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1477 / 1689 loss=3.604, nll_loss=2.062, ppl=4.18, wps=559690, ups=1.13, wpb=496903, bsz=16361.3, num_updates=35200, lr=0.0003371, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=32604 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1577 / 1689 loss=3.608, nll_loss=2.066, ppl=4.19, wps=554535, ups=1.12, wpb=496768, bsz=16438.5, num_updates=35300, lr=0.000336622, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=32693 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 epoch 021: 1678 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=548333, ups=1.11, wpb=494630, bsz=16527.3, num_updates=35400, lr=0.000336146, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32783 end of epoch 21 (average epoch stats below) epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 epoch 021 | loss 3.599 | nll_loss 2.056 | ppl 4.16 | wps 542898 | ups 1.1 | wpb 495118 | bsz 16506.8 | num_updates 35411 | lr 0.000336094 | gnorm 0.191 | clip 0 | loss_scale 1 | train_wall 1482 | gb_free 24 | wall 32792 Start iterating over samples epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 89 / 1689 loss=3.578, nll_loss=2.031, ppl=4.09, wps=549877, ups=1.12, wpb=491112, bsz=16526, num_updates=35500, lr=0.000335673, gnorm=0.197, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=32873 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 189 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=552150, ups=1.12, wpb=493570, bsz=16081.5, num_updates=35600, lr=0.000335201, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=32962 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 289 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=550405, ups=1.11, wpb=494230, bsz=16687.9, num_updates=35700, lr=0.000334731, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.9, wall=33052 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 389 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=548850, ups=1.11, wpb=495664, bsz=16765, num_updates=35800, lr=0.000334263, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=33142 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 489 / 1689 loss=3.586, nll_loss=2.041, ppl=4.12, wps=551888, ups=1.11, wpb=495564, bsz=16555, num_updates=35900, lr=0.000333797, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=21.6, wall=33232 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 epoch 022: 590 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=547918, ups=1.11, wpb=495748, bsz=16567, num_updates=36000, lr=0.000333333, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.5, wall=33322 begin validation on "valid" subset epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.709 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 36000 | best_loss 3.708 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 690 / 1689 loss=3.593, nll_loss=2.048, ppl=4.14, wps=488088, ups=0.98, wpb=495734, bsz=16464.4, num_updates=36100, lr=0.000332871, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=33424 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 790 / 1689 loss=3.593, nll_loss=2.049, ppl=4.14, wps=553427, ups=1.12, wpb=494668, bsz=16353.9, num_updates=36200, lr=0.000332411, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=33513 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 890 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=551710, ups=1.12, wpb=494578, bsz=16751, num_updates=36300, lr=0.000331953, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=33603 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 990 / 1689 loss=3.6, nll_loss=2.057, ppl=4.16, wps=556704, ups=1.12, wpb=496460, bsz=16477.8, num_updates=36400, lr=0.000331497, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=33692 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1090 / 1689 loss=3.595, nll_loss=2.051, ppl=4.14, wps=554988, ups=1.12, wpb=494926, bsz=16594.7, num_updates=36500, lr=0.000331042, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=33781 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1190 / 1689 loss=3.602, nll_loss=2.06, ppl=4.17, wps=552595, ups=1.11, wpb=495925, bsz=16382.7, num_updates=36600, lr=0.00033059, gnorm=0.183, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=33871 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1290 / 1689 loss=3.595, nll_loss=2.052, ppl=4.15, wps=554041, ups=1.11, wpb=497004, bsz=16455.8, num_updates=36700, lr=0.000330139, gnorm=0.192, clip=0, loss_scale=2, train_wall=88, gb_free=22.3, wall=33961 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1391 / 1689 loss=3.605, nll_loss=2.063, ppl=4.18, wps=550244, ups=1.11, wpb=496044, bsz=16687.1, num_updates=36800, lr=0.00032969, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=34051 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1492 / 1689 loss=3.601, nll_loss=2.058, ppl=4.16, wps=545399, ups=1.1, wpb=495049, bsz=16251.4, num_updates=36900, lr=0.000329243, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=34142 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 epoch 022: 1592 / 1689 loss=3.599, nll_loss=2.056, ppl=4.16, wps=558630, ups=1.13, wpb=495270, bsz=16576.4, num_updates=37000, lr=0.000328798, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.8, wall=34230 begin validation on "valid" subset epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 epoch 022 | valid on 'valid' subset | loss 3.718 | nll_loss 2.164 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 37000 | best_loss 3.708 end of epoch 22 (average epoch stats below) epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 epoch 022 | loss 3.593 | nll_loss 2.049 | ppl 4.14 | wps 538193 | ups 1.09 | wpb 495109 | bsz 16503 | num_updates 37097 | lr 0.000328368 | gnorm 0.19 | clip 0 | loss_scale 0.5 | train_wall 1485 | gb_free 22.7 | wall 34344 Start iterating over samples epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 3 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=422438, ups=0.86, wpb=491692, bsz=16391.3, num_updates=37100, lr=0.000328355, gnorm=0.198, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.5, wall=34347 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 103 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=564922, ups=1.14, wpb=495372, bsz=16652.6, num_updates=37200, lr=0.000327913, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=87, gb_free=22.3, wall=34435 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 203 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=552854, ups=1.12, wpb=495036, bsz=17039.4, num_updates=37300, lr=0.000327473, gnorm=0.2, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=34524 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 303 / 1689 loss=3.582, nll_loss=2.036, ppl=4.1, wps=564140, ups=1.14, wpb=496937, bsz=16699.8, num_updates=37400, lr=0.000327035, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=34612 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 403 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=561744, ups=1.13, wpb=495736, bsz=16631.8, num_updates=37500, lr=0.000326599, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=34700 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 503 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=557780, ups=1.13, wpb=495584, bsz=16722.6, num_updates=37600, lr=0.000326164, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34789 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 603 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=554415, ups=1.12, wpb=494395, bsz=16210.4, num_updates=37700, lr=0.000325731, gnorm=0.19, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=34879 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 703 / 1689 loss=3.585, nll_loss=2.041, ppl=4.11, wps=555705, ups=1.12, wpb=494606, bsz=16542.6, num_updates=37800, lr=0.0003253, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=34968 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 803 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=552999, ups=1.12, wpb=495916, bsz=16497.4, num_updates=37900, lr=0.000324871, gnorm=0.183, clip=0, loss_scale=2, train_wall=89, gb_free=22.3, wall=35057 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 epoch 023: 903 / 1689 loss=3.596, nll_loss=2.053, ppl=4.15, wps=554200, ups=1.12, wpb=495050, bsz=16859.4, num_updates=38000, lr=0.000324443, gnorm=0.189, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=35147 begin validation on "valid" subset epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023 | valid on 'valid' subset | loss 3.713 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 38000 | best_loss 3.708 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1004 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=373629, ups=0.75, wpb=495305, bsz=16199.1, num_updates=38100, lr=0.000324017, gnorm=0.194, clip=0, loss_scale=1, train_wall=114, gb_free=22.4, wall=35279 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1104 / 1689 loss=3.598, nll_loss=2.055, ppl=4.16, wps=556309, ups=1.13, wpb=494254, bsz=16301.7, num_updates=38200, lr=0.000323592, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=35368 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1204 / 1689 loss=3.589, nll_loss=2.045, ppl=4.13, wps=551653, ups=1.11, wpb=495937, bsz=16195.2, num_updates=38300, lr=0.00032317, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=35458 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1304 / 1689 loss=3.597, nll_loss=2.053, ppl=4.15, wps=554186, ups=1.12, wpb=495088, bsz=16557.8, num_updates=38400, lr=0.000322749, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=35547 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1404 / 1689 loss=3.59, nll_loss=2.046, ppl=4.13, wps=553648, ups=1.12, wpb=494603, bsz=16460, num_updates=38500, lr=0.000322329, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=21, wall=35637 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1505 / 1689 loss=3.587, nll_loss=2.042, ppl=4.12, wps=542370, ups=1.1, wpb=494732, bsz=16553.1, num_updates=38600, lr=0.000321911, gnorm=0.187, clip=0, loss_scale=1, train_wall=90, gb_free=21.7, wall=35728 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 epoch 023: 1605 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555284, ups=1.12, wpb=496558, bsz=16291, num_updates=38700, lr=0.000321495, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=35817 end of epoch 23 (average epoch stats below) epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 epoch 023 | loss 3.588 | nll_loss 2.043 | ppl 4.12 | wps 539440 | ups 1.09 | wpb 495113 | bsz 16505.4 | num_updates 38784 | lr 0.000321147 | gnorm 0.188 | clip 0 | loss_scale 1 | train_wall 1510 | gb_free 23.7 | wall 35892 Start iterating over samples epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 16 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=521367, ups=1.06, wpb=491748, bsz=16232.1, num_updates=38800, lr=0.000321081, gnorm=0.185, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=35911 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 116 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=568817, ups=1.15, wpb=496300, bsz=16697.8, num_updates=38900, lr=0.000320668, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=35999 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 epoch 024: 216 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=557934, ups=1.13, wpb=495898, bsz=16792.2, num_updates=39000, lr=0.000320256, gnorm=0.202, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=36088 begin validation on "valid" subset epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.72 | nll_loss 2.162 | ppl 4.48 | wps 0 | wpb 44526 | bsz 2008 | num_updates 39000 | best_loss 3.708 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 316 / 1689 loss=3.573, nll_loss=2.026, ppl=4.07, wps=490338, ups=0.99, wpb=495760, bsz=16244.6, num_updates=39100, lr=0.000319847, gnorm=0.184, clip=0, loss_scale=2, train_wall=87, gb_free=21.7, wall=36189 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 416 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=553788, ups=1.12, wpb=494422, bsz=16427.8, num_updates=39200, lr=0.000319438, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36278 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 516 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=558454, ups=1.12, wpb=497013, bsz=16628.7, num_updates=39300, lr=0.000319032, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=21.8, wall=36367 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 617 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=546757, ups=1.1, wpb=494888, bsz=16373.3, num_updates=39400, lr=0.000318626, gnorm=0.192, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=36458 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 717 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=548329, ups=1.11, wpb=494274, bsz=16267.9, num_updates=39500, lr=0.000318223, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.3, wall=36548 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 817 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=554171, ups=1.12, wpb=496588, bsz=16342.2, num_updates=39600, lr=0.000317821, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=36637 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 917 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=552598, ups=1.11, wpb=495924, bsz=16616.9, num_updates=39700, lr=0.00031742, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=36727 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1017 / 1689 loss=3.587, nll_loss=2.043, ppl=4.12, wps=558153, ups=1.13, wpb=495071, bsz=16588.8, num_updates=39800, lr=0.000317021, gnorm=0.177, clip=0, loss_scale=1, train_wall=87, gb_free=21.8, wall=36816 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1117 / 1689 loss=3.585, nll_loss=2.04, ppl=4.11, wps=550006, ups=1.11, wpb=496334, bsz=16905.7, num_updates=39900, lr=0.000316624, gnorm=0.187, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36906 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 epoch 024: 1217 / 1689 loss=3.584, nll_loss=2.039, ppl=4.11, wps=549649, ups=1.11, wpb=493992, bsz=16687, num_updates=40000, lr=0.000316228, gnorm=0.198, clip=0, loss_scale=2, train_wall=88, gb_free=22.4, wall=36996 begin validation on "valid" subset epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024 | valid on 'valid' subset | loss 3.709 | nll_loss 2.152 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 40000 | best_loss 3.708 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1317 / 1689 loss=3.583, nll_loss=2.039, ppl=4.11, wps=477519, ups=0.96, wpb=495037, bsz=16512, num_updates=40100, lr=0.000315833, gnorm=0.195, clip=0, loss_scale=2, train_wall=88, gb_free=21.9, wall=37100 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1418 / 1689 loss=3.588, nll_loss=2.043, ppl=4.12, wps=553359, ups=1.12, wpb=494409, bsz=16279.2, num_updates=40200, lr=0.00031544, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37189 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1518 / 1689 loss=3.593, nll_loss=2.05, ppl=4.14, wps=553358, ups=1.12, wpb=495356, bsz=16381.2, num_updates=40300, lr=0.000315049, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=37278 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 epoch 024: 1618 / 1689 loss=3.592, nll_loss=2.049, ppl=4.14, wps=555432, ups=1.12, wpb=495859, bsz=16599, num_updates=40400, lr=0.000314658, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=37368 end of epoch 24 (average epoch stats below) epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 epoch 024 | loss 3.582 | nll_loss 2.037 | ppl 4.1 | wps 542999 | ups 1.1 | wpb 495139 | bsz 16508.2 | num_updates 40471 | lr 0.000314382 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1483 | gb_free 23 | wall 37430 Start iterating over samples epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 29 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=546138, ups=1.11, wpb=490809, bsz=16174.3, num_updates=40500, lr=0.00031427, gnorm=0.196, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=37458 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 129 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=551078, ups=1.11, wpb=494999, bsz=16337.8, num_updates=40600, lr=0.000313882, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=37547 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 229 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=554415, ups=1.12, wpb=496597, bsz=16534.2, num_updates=40700, lr=0.000313497, gnorm=0.185, clip=0, loss_scale=2, train_wall=88, gb_free=22.5, wall=37637 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 329 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550082, ups=1.11, wpb=496444, bsz=16737.3, num_updates=40800, lr=0.000313112, gnorm=0.19, clip=0, loss_scale=2, train_wall=89, gb_free=22.6, wall=37727 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 430 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=545850, ups=1.1, wpb=495710, bsz=16470.9, num_updates=40900, lr=0.000312729, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=37818 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 epoch 025: 530 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=552377, ups=1.12, wpb=495004, bsz=16587.2, num_updates=41000, lr=0.000312348, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=37908 begin validation on "valid" subset epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025 | valid on 'valid' subset | loss 3.712 | nll_loss 2.154 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 41000 | best_loss 3.708 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 630 / 1689 loss=3.58, nll_loss=2.034, ppl=4.1, wps=491594, ups=0.99, wpb=494898, bsz=16404.9, num_updates=41100, lr=0.000311967, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=20.6, wall=38008 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 730 / 1689 loss=3.579, nll_loss=2.034, ppl=4.09, wps=552506, ups=1.11, wpb=495679, bsz=16774.6, num_updates=41200, lr=0.000311588, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=38098 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 830 / 1689 loss=3.582, nll_loss=2.037, ppl=4.1, wps=555679, ups=1.12, wpb=495159, bsz=16322.9, num_updates=41300, lr=0.000311211, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=38187 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 930 / 1689 loss=3.588, nll_loss=2.044, ppl=4.12, wps=556421, ups=1.12, wpb=494680, bsz=16934.1, num_updates=41400, lr=0.000310835, gnorm=0.19, clip=0, loss_scale=2, train_wall=88, gb_free=20.8, wall=38276 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1031 / 1689 loss=3.581, nll_loss=2.036, ppl=4.1, wps=549567, ups=1.11, wpb=494018, bsz=16466.9, num_updates=41500, lr=0.00031046, gnorm=0.188, clip=0, loss_scale=1, train_wall=89, gb_free=22.7, wall=38366 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1132 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=550482, ups=1.11, wpb=494882, bsz=16065.7, num_updates=41600, lr=0.000310087, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=38456 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1232 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557708, ups=1.13, wpb=495084, bsz=16419.2, num_updates=41700, lr=0.000309715, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.5, wall=38545 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1332 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553118, ups=1.11, wpb=497090, bsz=16643.7, num_updates=41800, lr=0.000309344, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=38634 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1432 / 1689 loss=3.579, nll_loss=2.034, ppl=4.1, wps=551656, ups=1.11, wpb=495482, bsz=16460.2, num_updates=41900, lr=0.000308975, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.7, wall=38724 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 epoch 025: 1532 / 1689 loss=3.584, nll_loss=2.04, ppl=4.11, wps=550772, ups=1.11, wpb=494190, bsz=16763.2, num_updates=42000, lr=0.000308607, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=38814 begin validation on "valid" subset epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025 | valid on 'valid' subset | loss 3.705 | nll_loss 2.148 | ppl 4.43 | wps 0 | wpb 44526 | bsz 2008 | num_updates 42000 | best_loss 3.705 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 epoch 025: 1632 / 1689 loss=3.58, nll_loss=2.035, ppl=4.1, wps=386304, ups=0.78, wpb=495293, bsz=16362.2, num_updates=42100, lr=0.00030824, gnorm=0.187, clip=0, loss_scale=1, train_wall=107, gb_free=22.3, wall=38942 end of epoch 25 (average epoch stats below) epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 epoch 025 | loss 3.578 | nll_loss 2.032 | ppl 4.09 | wps 534336 | ups 1.08 | wpb 495128 | bsz 16506.1 | num_updates 42157 | lr 0.000308032 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1506 | gb_free 23.7 | wall 38992 Start iterating over samples epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 43 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=548469, ups=1.12, wpb=491703, bsz=16496.9, num_updates=42200, lr=0.000307875, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21.7, wall=39032 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 143 / 1689 loss=3.559, nll_loss=2.01, ppl=4.03, wps=551844, ups=1.11, wpb=495093, bsz=16447.9, num_updates=42300, lr=0.00030751, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=39122 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 243 / 1689 loss=3.568, nll_loss=2.021, ppl=4.06, wps=555816, ups=1.12, wpb=496528, bsz=16881.5, num_updates=42400, lr=0.000307148, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=39211 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 343 / 1689 loss=3.57, nll_loss=2.023, ppl=4.06, wps=551861, ups=1.12, wpb=494253, bsz=16774.7, num_updates=42500, lr=0.000306786, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=39301 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 443 / 1689 loss=3.567, nll_loss=2.02, ppl=4.06, wps=552448, ups=1.12, wpb=494432, bsz=16187.1, num_updates=42600, lr=0.000306426, gnorm=0.197, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=39390 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 544 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=543916, ups=1.1, wpb=495902, bsz=16558.4, num_updates=42700, lr=0.000306067, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.8, wall=39481 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 644 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=554717, ups=1.12, wpb=495437, bsz=16067.5, num_updates=42800, lr=0.000305709, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=39571 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 744 / 1689 loss=3.572, nll_loss=2.025, ppl=4.07, wps=550310, ups=1.11, wpb=496248, bsz=16727.9, num_updates=42900, lr=0.000305352, gnorm=0.199, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=39661 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 epoch 026: 844 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=548224, ups=1.11, wpb=494954, bsz=16511, num_updates=43000, lr=0.000304997, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.6, wall=39751 begin validation on "valid" subset epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026 | valid on 'valid' subset | loss 3.707 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 43000 | best_loss 3.705 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 944 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=483347, ups=0.97, wpb=495860, bsz=16544.6, num_updates=43100, lr=0.000304643, gnorm=0.187, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=39854 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1044 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=557690, ups=1.13, wpb=495459, bsz=16298.2, num_updates=43200, lr=0.00030429, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=39942 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1145 / 1689 loss=3.578, nll_loss=2.033, ppl=4.09, wps=554210, ups=1.12, wpb=495558, bsz=16956.9, num_updates=43300, lr=0.000303939, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=40032 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1245 / 1689 loss=3.579, nll_loss=2.033, ppl=4.09, wps=560688, ups=1.13, wpb=495598, bsz=16222.1, num_updates=43400, lr=0.000303588, gnorm=0.187, clip=0, loss_scale=1, train_wall=87, gb_free=22.4, wall=40120 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1345 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=560797, ups=1.13, wpb=495343, bsz=16441.1, num_updates=43500, lr=0.000303239, gnorm=0.19, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=40209 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1445 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=559133, ups=1.13, wpb=495736, bsz=16696, num_updates=43600, lr=0.000302891, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=40297 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1545 / 1689 loss=3.583, nll_loss=2.038, ppl=4.11, wps=557797, ups=1.13, wpb=495015, bsz=16405.8, num_updates=43700, lr=0.000302545, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.8, wall=40386 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 epoch 026: 1645 / 1689 loss=3.577, nll_loss=2.031, ppl=4.09, wps=555665, ups=1.12, wpb=494936, bsz=16407.7, num_updates=43800, lr=0.000302199, gnorm=0.194, clip=0, loss_scale=2, train_wall=88, gb_free=22.1, wall=40475 end of epoch 26 (average epoch stats below) epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 epoch 026 | loss 3.573 | nll_loss 2.027 | ppl 4.08 | wps 548745 | ups 1.11 | wpb 495134 | bsz 16507 | num_updates 43843 | lr 0.000302051 | gnorm 0.189 | clip 0 | loss_scale 1 | train_wall 1485 | gb_free 23.1 | wall 40514 Start iterating over samples epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 57 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=545470, ups=1.11, wpb=490499, bsz=16018.8, num_updates=43900, lr=0.000301855, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=40565 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 epoch 027: 157 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=556146, ups=1.12, wpb=495787, bsz=16520.3, num_updates=44000, lr=0.000301511, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=40654 begin validation on "valid" subset epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027 | valid on 'valid' subset | loss 3.708 | nll_loss 2.153 | ppl 4.45 | wps 0 | wpb 44526 | bsz 2008 | num_updates 44000 | best_loss 3.705 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 257 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=483386, ups=0.97, wpb=495798, bsz=16062.8, num_updates=44100, lr=0.000301169, gnorm=0.18, clip=0, loss_scale=1, train_wall=89, gb_free=21.7, wall=40757 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 357 / 1689 loss=3.56, nll_loss=2.012, ppl=4.03, wps=556922, ups=1.12, wpb=495798, bsz=16531.7, num_updates=44200, lr=0.000300828, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=40846 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 457 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=552202, ups=1.11, wpb=496328, bsz=16350, num_updates=44300, lr=0.000300489, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=40936 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 558 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=551577, ups=1.11, wpb=496549, bsz=16326.2, num_updates=44400, lr=0.00030015, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41026 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 658 / 1689 loss=3.566, nll_loss=2.019, ppl=4.05, wps=548603, ups=1.11, wpb=493491, bsz=16722.8, num_updates=44500, lr=0.000299813, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=18.9, wall=41116 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 758 / 1689 loss=3.57, nll_loss=2.024, ppl=4.07, wps=544754, ups=1.1, wpb=494018, bsz=16787.9, num_updates=44600, lr=0.000299476, gnorm=0.191, clip=0, loss_scale=1, train_wall=89, gb_free=21.9, wall=41206 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 858 / 1689 loss=3.572, nll_loss=2.026, ppl=4.07, wps=552930, ups=1.12, wpb=494626, bsz=16688.6, num_updates=44700, lr=0.000299141, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=41296 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 958 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=554092, ups=1.12, wpb=495242, bsz=16203, num_updates=44800, lr=0.000298807, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=41385 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1058 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=553375, ups=1.12, wpb=494235, bsz=16903, num_updates=44900, lr=0.000298474, gnorm=0.179, clip=0, loss_scale=2, train_wall=87, gb_free=22.4, wall=41474 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 epoch 027: 1159 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=547632, ups=1.11, wpb=495270, bsz=16923.9, num_updates=45000, lr=0.000298142, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=41565 begin validation on "valid" subset epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027 | valid on 'valid' subset | loss 3.69 | nll_loss 2.134 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 45000 | best_loss 3.69 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1259 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=402061, ups=0.81, wpb=495840, bsz=16404.9, num_updates=45100, lr=0.000297812, gnorm=0.187, clip=0, loss_scale=1, train_wall=101, gb_free=22.2, wall=41688 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1359 / 1689 loss=3.576, nll_loss=2.03, ppl=4.08, wps=555129, ups=1.12, wpb=495692, bsz=16399.3, num_updates=45200, lr=0.000297482, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41777 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1459 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=550086, ups=1.11, wpb=495992, bsz=16518.2, num_updates=45300, lr=0.000297154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41868 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1559 / 1689 loss=3.576, nll_loss=2.031, ppl=4.09, wps=552227, ups=1.11, wpb=495352, bsz=16342, num_updates=45400, lr=0.000296826, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=41957 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 epoch 027: 1660 / 1689 loss=3.574, nll_loss=2.028, ppl=4.08, wps=538473, ups=1.09, wpb=495604, bsz=16799.7, num_updates=45500, lr=0.0002965, gnorm=0.187, clip=0, loss_scale=1, train_wall=91, gb_free=21.7, wall=42049 end of epoch 27 (average epoch stats below) epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 epoch 027 | loss 3.569 | nll_loss 2.022 | ppl 4.06 | wps 534819 | ups 1.08 | wpb 495126 | bsz 16507.9 | num_updates 45529 | lr 0.000296405 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1500 | gb_free 23.2 | wall 42075 Start iterating over samples epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 72 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=535552, ups=1.09, wpb=490816, bsz=16563.8, num_updates=45600, lr=0.000296174, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=42141 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 172 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=550759, ups=1.11, wpb=497853, bsz=16387.4, num_updates=45700, lr=0.00029585, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.9, wall=42231 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 272 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=549190, ups=1.11, wpb=494754, bsz=16510.1, num_updates=45800, lr=0.000295527, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=42322 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 372 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=552464, ups=1.12, wpb=494347, bsz=16558, num_updates=45900, lr=0.000295205, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=88, gb_free=20.7, wall=42411 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 epoch 028: 472 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=555288, ups=1.12, wpb=496562, bsz=16736.2, num_updates=46000, lr=0.000294884, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=42500 begin validation on "valid" subset epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.712 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 46000 | best_loss 3.69 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 572 / 1689 loss=3.563, nll_loss=2.015, ppl=4.04, wps=491975, ups=1, wpb=494213, bsz=17086.6, num_updates=46100, lr=0.000294564, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.3, wall=42601 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 672 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557016, ups=1.12, wpb=495836, bsz=16464.5, num_updates=46200, lr=0.000294245, gnorm=0.188, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=42690 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 772 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=550914, ups=1.11, wpb=496542, bsz=16452.6, num_updates=46300, lr=0.000293927, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=42780 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 872 / 1689 loss=3.565, nll_loss=2.018, ppl=4.05, wps=553585, ups=1.12, wpb=496256, bsz=16380.6, num_updates=46400, lr=0.00029361, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=42870 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 972 / 1689 loss=3.571, nll_loss=2.025, ppl=4.07, wps=556211, ups=1.12, wpb=495350, bsz=16498.1, num_updates=46500, lr=0.000293294, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=42959 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1072 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=558318, ups=1.13, wpb=495827, bsz=16184.3, num_updates=46600, lr=0.000292979, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=43048 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1173 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=552132, ups=1.11, wpb=496804, bsz=16327.6, num_updates=46700, lr=0.000292666, gnorm=0.197, clip=0, loss_scale=1, train_wall=88, gb_free=21.3, wall=43138 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1273 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=555048, ups=1.12, wpb=494667, bsz=16337.4, num_updates=46800, lr=0.000292353, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=43227 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1373 / 1689 loss=3.566, nll_loss=2.02, ppl=4.06, wps=553756, ups=1.11, wpb=496681, bsz=16606.9, num_updates=46900, lr=0.000292041, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=43316 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 epoch 028: 1473 / 1689 loss=3.572, nll_loss=2.027, ppl=4.07, wps=552521, ups=1.12, wpb=494584, bsz=16508.2, num_updates=47000, lr=0.00029173, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=43406 begin validation on "valid" subset epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028 | valid on 'valid' subset | loss 3.697 | nll_loss 2.14 | ppl 4.41 | wps 0 | wpb 44526 | bsz 2008 | num_updates 47000 | best_loss 3.69 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1573 / 1689 loss=3.575, nll_loss=2.03, ppl=4.08, wps=439902, ups=0.89, wpb=494072, bsz=16454.6, num_updates=47100, lr=0.00029142, gnorm=0.188, clip=0, loss_scale=1, train_wall=96, gb_free=21.8, wall=43518 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 epoch 028: 1673 / 1689 loss=3.577, nll_loss=2.032, ppl=4.09, wps=553684, ups=1.12, wpb=492936, bsz=16809.1, num_updates=47200, lr=0.000291111, gnorm=0.176, clip=0, loss_scale=2, train_wall=88, gb_free=21.7, wall=43607 end of epoch 28 (average epoch stats below) epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 epoch 028 | loss 3.564 | nll_loss 2.017 | ppl 4.05 | wps 540165 | ups 1.09 | wpb 495112 | bsz 16509.4 | num_updates 47216 | lr 0.000291062 | gnorm 0.186 | clip 0 | loss_scale 2 | train_wall 1491 | gb_free 22.6 | wall 43621 Start iterating over samples epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 85 / 1689 loss=3.553, nll_loss=2.004, ppl=4.01, wps=364660, ups=0.74, wpb=491069, bsz=16459.4, num_updates=47300, lr=0.000290803, gnorm=0.19, clip=0, loss_scale=1, train_wall=86, gb_free=21.8, wall=43742 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 185 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=562874, ups=1.14, wpb=495188, bsz=16838.4, num_updates=47400, lr=0.000290496, gnorm=0.175, clip=0, loss_scale=1, train_wall=87, gb_free=22.7, wall=43830 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 285 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=564268, ups=1.14, wpb=495112, bsz=16302.6, num_updates=47500, lr=0.000290191, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=22.6, wall=43918 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 385 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=559076, ups=1.13, wpb=494024, bsz=16035.2, num_updates=47600, lr=0.000289886, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=44006 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 485 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=554666, ups=1.12, wpb=495622, bsz=16748.3, num_updates=47700, lr=0.000289581, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=44095 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 586 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=544437, ups=1.1, wpb=495087, bsz=16779.5, num_updates=47800, lr=0.000289278, gnorm=0.183, clip=0, loss_scale=1, train_wall=90, gb_free=22.6, wall=44186 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 687 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=546936, ups=1.1, wpb=496589, bsz=16504.9, num_updates=47900, lr=0.000288976, gnorm=0.195, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=44277 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 epoch 029: 787 / 1689 loss=3.564, nll_loss=2.017, ppl=4.05, wps=549349, ups=1.11, wpb=496279, bsz=16438.2, num_updates=48000, lr=0.000288675, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.9, wall=44367 begin validation on "valid" subset epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029 | valid on 'valid' subset | loss 3.706 | nll_loss 2.151 | ppl 4.44 | wps 0 | wpb 44526 | bsz 2008 | num_updates 48000 | best_loss 3.69 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 887 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=378577, ups=0.77, wpb=493333, bsz=16669.2, num_updates=48100, lr=0.000288375, gnorm=0.19, clip=0, loss_scale=0.5, train_wall=112, gb_free=22.2, wall=44498 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 987 / 1689 loss=3.565, nll_loss=2.019, ppl=4.05, wps=551198, ups=1.12, wpb=493020, bsz=16519.9, num_updates=48200, lr=0.000288076, gnorm=0.193, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=44587 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1087 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=557040, ups=1.12, wpb=496382, bsz=16395.8, num_updates=48300, lr=0.000287777, gnorm=0.178, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.9, wall=44676 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1187 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=554015, ups=1.12, wpb=496354, bsz=16757.9, num_updates=48400, lr=0.00028748, gnorm=0.172, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=44766 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1287 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=557942, ups=1.12, wpb=496042, bsz=16553.2, num_updates=48500, lr=0.000287183, gnorm=0.178, clip=0, loss_scale=1, train_wall=87, gb_free=21.9, wall=44855 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1387 / 1689 loss=3.573, nll_loss=2.028, ppl=4.08, wps=553391, ups=1.11, wpb=496637, bsz=16498.3, num_updates=48600, lr=0.000286888, gnorm=0.177, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=44945 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1487 / 1689 loss=3.562, nll_loss=2.016, ppl=4.04, wps=551637, ups=1.12, wpb=494586, bsz=16412.4, num_updates=48700, lr=0.000286593, gnorm=0.186, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=45034 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1587 / 1689 loss=3.569, nll_loss=2.022, ppl=4.06, wps=548426, ups=1.11, wpb=494307, bsz=16295.2, num_updates=48800, lr=0.000286299, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=45124 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 epoch 029: 1687 / 1689 loss=3.569, nll_loss=2.023, ppl=4.06, wps=556797, ups=1.12, wpb=496706, bsz=16284.3, num_updates=48900, lr=0.000286006, gnorm=0.178, clip=0, loss_scale=2, train_wall=88, gb_free=22, wall=45214 end of epoch 29 (average epoch stats below) epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 epoch 029 | loss 3.56 | nll_loss 2.013 | ppl 4.04 | wps 523765 | ups 1.06 | wpb 495107 | bsz 16502.1 | num_updates 48902 | lr 0.000286 | gnorm 0.183 | clip 0 | loss_scale 2 | train_wall 1505 | gb_free 25.8 | wall 45215 Start iterating over samples epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 epoch 030: 99 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=541705, ups=1.1, wpb=491038, bsz=16432.2, num_updates=49000, lr=0.000285714, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=45304 begin validation on "valid" subset epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.701 | nll_loss 2.146 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 49000 | best_loss 3.69 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 199 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=401305, ups=0.81, wpb=495452, bsz=16212.6, num_updates=49100, lr=0.000285423, gnorm=0.183, clip=0, loss_scale=1, train_wall=96, gb_free=22.1, wall=45428 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 299 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=566013, ups=1.14, wpb=495851, bsz=16249, num_updates=49200, lr=0.000285133, gnorm=0.191, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=45515 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 399 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=557975, ups=1.13, wpb=494905, bsz=16506, num_updates=49300, lr=0.000284844, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45604 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 499 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=557862, ups=1.12, wpb=496335, bsz=16781.5, num_updates=49400, lr=0.000284555, gnorm=0.187, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=45693 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 600 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=549401, ups=1.11, wpb=496345, bsz=16506.6, num_updates=49500, lr=0.000284268, gnorm=0.188, clip=0, loss_scale=1, train_wall=90, gb_free=21.6, wall=45783 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 700 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556510, ups=1.12, wpb=495294, bsz=16461.4, num_updates=49600, lr=0.000283981, gnorm=0.184, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=45872 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 800 / 1689 loss=3.561, nll_loss=2.014, ppl=4.04, wps=556289, ups=1.12, wpb=496257, bsz=16346.6, num_updates=49700, lr=0.000283695, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=45962 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 900 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=555319, ups=1.12, wpb=494923, bsz=16430.3, num_updates=49800, lr=0.00028341, gnorm=0.186, clip=0, loss_scale=1, train_wall=87, gb_free=22, wall=46051 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1000 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=557843, ups=1.13, wpb=494874, bsz=16581.8, num_updates=49900, lr=0.000283126, gnorm=0.18, clip=0, loss_scale=1, train_wall=87, gb_free=21.5, wall=46139 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 epoch 030: 1101 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=548105, ups=1.11, wpb=494873, bsz=16577, num_updates=50000, lr=0.000282843, gnorm=0.177, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=46230 begin validation on "valid" subset epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030 | valid on 'valid' subset | loss 3.711 | nll_loss 2.16 | ppl 4.47 | wps 0 | wpb 44526 | bsz 2008 | num_updates 50000 | best_loss 3.69 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1201 / 1689 loss=3.567, nll_loss=2.021, ppl=4.06, wps=401920, ups=0.81, wpb=494441, bsz=16925.5, num_updates=50100, lr=0.00028256, gnorm=0.195, clip=0, loss_scale=1, train_wall=106, gb_free=21.3, wall=46353 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1301 / 1689 loss=3.563, nll_loss=2.016, ppl=4.05, wps=554279, ups=1.12, wpb=493980, bsz=16257.7, num_updates=50200, lr=0.000282279, gnorm=0.185, clip=0, loss_scale=1, train_wall=87, gb_free=22.1, wall=46442 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1401 / 1689 loss=3.562, nll_loss=2.015, ppl=4.04, wps=553345, ups=1.12, wpb=495715, bsz=16682.4, num_updates=50300, lr=0.000281998, gnorm=0.198, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=46531 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1501 / 1689 loss=3.557, nll_loss=2.01, ppl=4.03, wps=552318, ups=1.11, wpb=495914, bsz=16674.2, num_updates=50400, lr=0.000281718, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=46621 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 epoch 030: 1602 / 1689 loss=3.559, nll_loss=2.012, ppl=4.03, wps=548281, ups=1.1, wpb=496230, bsz=16485, num_updates=50500, lr=0.000281439, gnorm=0.179, clip=0, loss_scale=1, train_wall=89, gb_free=22, wall=46712 end of epoch 30 (average epoch stats below) epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 epoch 030 | loss 3.557 | nll_loss 2.009 | ppl 4.02 | wps 529927 | ups 1.07 | wpb 495142 | bsz 16504.3 | num_updates 50587 | lr 0.000281197 | gnorm 0.187 | clip 0 | loss_scale 1 | train_wall 1509 | gb_free 23.8 | wall 46789 Start iterating over samples epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 13 / 1689 loss=3.563, nll_loss=2.017, ppl=4.05, wps=546674, ups=1.11, wpb=490940, bsz=16352.8, num_updates=50600, lr=0.000281161, gnorm=0.199, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=46802 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 113 / 1689 loss=3.543, nll_loss=1.993, ppl=3.98, wps=554853, ups=1.12, wpb=494863, bsz=16403.4, num_updates=50700, lr=0.000280883, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=46891 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 213 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=550391, ups=1.11, wpb=495008, bsz=16661.3, num_updates=50800, lr=0.000280607, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=46981 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 313 / 1689 loss=3.541, nll_loss=1.991, ppl=3.97, wps=556261, ups=1.12, wpb=495583, bsz=16250.6, num_updates=50900, lr=0.000280331, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=47070 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 epoch 031: 413 / 1689 loss=3.555, nll_loss=2.006, ppl=4.02, wps=544498, ups=1.1, wpb=493470, bsz=16421.3, num_updates=51000, lr=0.000280056, gnorm=0.193, clip=0, loss_scale=1, train_wall=89, gb_free=22.1, wall=47160 begin validation on "valid" subset epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.722 | nll_loss 2.173 | ppl 4.51 | wps 0 | wpb 44526 | bsz 2008 | num_updates 51000 | best_loss 3.69 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 514 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=478125, ups=0.97, wpb=494696, bsz=16803.4, num_updates=51100, lr=0.000279782, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=22.2, wall=47264 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 614 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=552496, ups=1.12, wpb=495376, bsz=16505.4, num_updates=51200, lr=0.000279508, gnorm=0.193, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=47353 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 714 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=551643, ups=1.11, wpb=495997, bsz=16694, num_updates=51300, lr=0.000279236, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=21.6, wall=47443 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 814 / 1689 loss=3.552, nll_loss=2.003, ppl=4.01, wps=556976, ups=1.12, wpb=496338, bsz=16337.3, num_updates=51400, lr=0.000278964, gnorm=0.176, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=47533 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 914 / 1689 loss=3.559, nll_loss=2.011, ppl=4.03, wps=561872, ups=1.13, wpb=495739, bsz=16429.5, num_updates=51500, lr=0.000278693, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=47621 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1014 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=561162, ups=1.13, wpb=495035, bsz=16707.4, num_updates=51600, lr=0.000278423, gnorm=0.187, clip=0, loss_scale=2, train_wall=87, gb_free=22.3, wall=47709 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1115 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=557970, ups=1.13, wpb=495864, bsz=15974, num_updates=51700, lr=0.000278154, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=21.9, wall=47798 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1215 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558536, ups=1.13, wpb=493577, bsz=16703.8, num_updates=51800, lr=0.000277885, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=20.2, wall=47886 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1315 / 1689 loss=3.563, nll_loss=2.016, ppl=4.04, wps=557452, ups=1.12, wpb=495600, bsz=16308.6, num_updates=51900, lr=0.000277617, gnorm=0.188, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=47975 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 epoch 031: 1415 / 1689 loss=3.558, nll_loss=2.01, ppl=4.03, wps=552570, ups=1.12, wpb=495014, bsz=16912.9, num_updates=52000, lr=0.00027735, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=48065 begin validation on "valid" subset epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031 | valid on 'valid' subset | loss 3.708 | nll_loss 2.156 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 52000 | best_loss 3.69 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1515 / 1689 loss=3.564, nll_loss=2.018, ppl=4.05, wps=485338, ups=0.98, wpb=496827, bsz=16807.2, num_updates=52100, lr=0.000277084, gnorm=0.189, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=48167 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 epoch 031: 1616 / 1689 loss=3.558, nll_loss=2.011, ppl=4.03, wps=552210, ups=1.11, wpb=495412, bsz=16348.7, num_updates=52200, lr=0.000276818, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48257 end of epoch 31 (average epoch stats below) epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 epoch 031 | loss 3.553 | nll_loss 2.005 | ppl 4.01 | wps 544787 | ups 1.1 | wpb 495114 | bsz 16508.6 | num_updates 52273 | lr 0.000276625 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1484 | gb_free 22.9 | wall 48321 Start iterating over samples epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 27 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=545738, ups=1.11, wpb=492847, bsz=16583.7, num_updates=52300, lr=0.000276553, gnorm=0.183, clip=0, loss_scale=1, train_wall=87, gb_free=21.6, wall=48347 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 127 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=555665, ups=1.12, wpb=494354, bsz=16750.3, num_updates=52400, lr=0.000276289, gnorm=0.194, clip=0, loss_scale=1, train_wall=87, gb_free=21, wall=48436 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 227 / 1689 loss=3.539, nll_loss=1.988, ppl=3.97, wps=553658, ups=1.12, wpb=494644, bsz=16585.4, num_updates=52500, lr=0.000276026, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48525 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 327 / 1689 loss=3.534, nll_loss=1.983, ppl=3.95, wps=553161, ups=1.11, wpb=496175, bsz=16535.8, num_updates=52600, lr=0.000275764, gnorm=0.181, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=48615 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 428 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551985, ups=1.11, wpb=496603, bsz=16594.2, num_updates=52700, lr=0.000275502, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=20.3, wall=48705 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 528 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=551500, ups=1.11, wpb=494969, bsz=16808.4, num_updates=52800, lr=0.000275241, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.5, wall=48795 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 628 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=547085, ups=1.1, wpb=495846, bsz=16461.6, num_updates=52900, lr=0.000274981, gnorm=0.184, clip=0, loss_scale=1, train_wall=89, gb_free=21.4, wall=48885 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 epoch 032: 728 / 1689 loss=3.55, nll_loss=2.001, ppl=4, wps=554011, ups=1.12, wpb=493682, bsz=16559.2, num_updates=53000, lr=0.000274721, gnorm=0.19, clip=0, loss_scale=1, train_wall=87, gb_free=22.2, wall=48975 begin validation on "valid" subset epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032 | valid on 'valid' subset | loss 3.69 | nll_loss 2.137 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 53000 | best_loss 3.69 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 828 / 1689 loss=3.551, nll_loss=2.002, ppl=4.01, wps=462303, ups=0.93, wpb=495254, bsz=16085, num_updates=53100, lr=0.000274462, gnorm=0.176, clip=0, loss_scale=1, train_wall=87, gb_free=22.5, wall=49082 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 929 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=545150, ups=1.1, wpb=495788, bsz=16505.8, num_updates=53200, lr=0.000274204, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.8, wall=49173 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1029 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=554035, ups=1.12, wpb=496270, bsz=16096.6, num_updates=53300, lr=0.000273947, gnorm=0.174, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=49262 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1129 / 1689 loss=3.557, nll_loss=2.009, ppl=4.03, wps=550981, ups=1.12, wpb=494122, bsz=16493.6, num_updates=53400, lr=0.00027369, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.4, wall=49352 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1229 / 1689 loss=3.555, nll_loss=2.008, ppl=4.02, wps=556077, ups=1.12, wpb=495046, bsz=16122.9, num_updates=53500, lr=0.000273434, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.4, wall=49441 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1329 / 1689 loss=3.549, nll_loss=2.001, ppl=4, wps=558258, ups=1.13, wpb=496177, bsz=16236.6, num_updates=53600, lr=0.000273179, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.9, wall=49530 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1429 / 1689 loss=3.56, nll_loss=2.014, ppl=4.04, wps=556211, ups=1.13, wpb=494180, bsz=16758.4, num_updates=53700, lr=0.000272925, gnorm=0.192, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=49619 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1529 / 1689 loss=3.556, nll_loss=2.008, ppl=4.02, wps=558602, ups=1.12, wpb=497380, bsz=16660.2, num_updates=53800, lr=0.000272671, gnorm=0.191, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=49708 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 epoch 032: 1629 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=554608, ups=1.12, wpb=494432, bsz=16592.4, num_updates=53900, lr=0.000272418, gnorm=0.183, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=49797 end of epoch 32 (average epoch stats below) epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 epoch 032 | loss 3.55 | nll_loss 2.001 | ppl 4 | wps 546513 | ups 1.1 | wpb 495106 | bsz 16508.4 | num_updates 53960 | lr 0.000272266 | gnorm 0.184 | clip 0 | loss_scale 1 | train_wall 1486 | gb_free 22.8 | wall 49850 Start iterating over samples epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 epoch 033: 40 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=550432, ups=1.12, wpb=491688, bsz=16434.4, num_updates=54000, lr=0.000272166, gnorm=0.179, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=49886 begin validation on "valid" subset epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033 | valid on 'valid' subset | loss 3.695 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 54000 | best_loss 3.69 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 140 / 1689 loss=3.537, nll_loss=1.986, ppl=3.96, wps=478763, ups=0.97, wpb=495064, bsz=16574.6, num_updates=54100, lr=0.000271914, gnorm=0.186, clip=0, loss_scale=1, train_wall=89, gb_free=21.8, wall=49990 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 240 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=547494, ups=1.1, wpb=496839, bsz=16990.7, num_updates=54200, lr=0.000271663, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=21.6, wall=50080 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 341 / 1689 loss=3.533, nll_loss=1.982, ppl=3.95, wps=545982, ups=1.11, wpb=493974, bsz=16275.3, num_updates=54300, lr=0.000271413, gnorm=0.181, clip=0, loss_scale=1, train_wall=89, gb_free=22.2, wall=50171 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 442 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=544559, ups=1.1, wpb=494787, bsz=16432.6, num_updates=54400, lr=0.000271163, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=50262 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 542 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=554672, ups=1.12, wpb=495758, bsz=16627.8, num_updates=54500, lr=0.000270914, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=50351 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 642 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=548970, ups=1.11, wpb=495228, bsz=16698.1, num_updates=54600, lr=0.000270666, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.7, wall=50441 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 742 / 1689 loss=3.546, nll_loss=1.998, ppl=3.99, wps=554963, ups=1.12, wpb=494149, bsz=16336.9, num_updates=54700, lr=0.000270418, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=50530 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 842 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=550261, ups=1.11, wpb=494664, bsz=16663.4, num_updates=54800, lr=0.000270172, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=50620 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 942 / 1689 loss=3.54, nll_loss=1.99, ppl=3.97, wps=551732, ups=1.11, wpb=496063, bsz=16007.1, num_updates=54900, lr=0.000269925, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=50710 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 epoch 033: 1042 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=553573, ups=1.12, wpb=495350, bsz=16614.3, num_updates=55000, lr=0.00026968, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=21.8, wall=50800 begin validation on "valid" subset epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033 | valid on 'valid' subset | loss 3.689 | nll_loss 2.133 | ppl 4.39 | wps 0 | wpb 44526 | bsz 2008 | num_updates 55000 | best_loss 3.689 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1142 / 1689 loss=3.555, nll_loss=2.007, ppl=4.02, wps=360468, ups=0.73, wpb=494294, bsz=16423.4, num_updates=55100, lr=0.000269435, gnorm=0.179, clip=0, loss_scale=1, train_wall=112, gb_free=21.7, wall=50937 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1242 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=555180, ups=1.12, wpb=495753, bsz=17058.3, num_updates=55200, lr=0.000269191, gnorm=0.182, clip=0, loss_scale=1, train_wall=87, gb_free=21.1, wall=51026 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1342 / 1689 loss=3.553, nll_loss=2.005, ppl=4.01, wps=555350, ups=1.12, wpb=496147, bsz=16307.9, num_updates=55300, lr=0.000268947, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=20.9, wall=51115 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1443 / 1689 loss=3.56, nll_loss=2.013, ppl=4.04, wps=543082, ups=1.09, wpb=496486, bsz=16838.6, num_updates=55400, lr=0.000268705, gnorm=0.18, clip=0, loss_scale=1, train_wall=90, gb_free=21.5, wall=51207 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1544 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=542143, ups=1.09, wpb=495234, bsz=16624, num_updates=55500, lr=0.000268462, gnorm=0.184, clip=0, loss_scale=0.5, train_wall=89, gb_free=22, wall=51298 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 epoch 033: 1644 / 1689 loss=3.552, nll_loss=2.004, ppl=4.01, wps=545820, ups=1.1, wpb=495510, bsz=16066.7, num_updates=55600, lr=0.000268221, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.2, wall=51389 end of epoch 33 (average epoch stats below) epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 epoch 033 | loss 3.546 | nll_loss 1.997 | ppl 3.99 | wps 528413 | ups 1.07 | wpb 495111 | bsz 16503.5 | num_updates 55645 | lr 0.000268112 | gnorm 0.181 | clip 0 | loss_scale 0.5 | train_wall 1512 | gb_free 24.1 | wall 51428 Start iterating over samples epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 55 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=548653, ups=1.12, wpb=490902, bsz=16384.6, num_updates=55700, lr=0.00026798, gnorm=0.187, clip=0, loss_scale=0.5, train_wall=87, gb_free=21.5, wall=51478 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 155 / 1689 loss=3.529, nll_loss=1.977, ppl=3.94, wps=553458, ups=1.12, wpb=496135, bsz=16707, num_updates=55800, lr=0.00026774, gnorm=0.179, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=51568 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 255 / 1689 loss=3.531, nll_loss=1.98, ppl=3.94, wps=551350, ups=1.11, wpb=496491, bsz=16090.3, num_updates=55900, lr=0.0002675, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=51658 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 epoch 034: 355 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=547899, ups=1.11, wpb=495335, bsz=16383.1, num_updates=56000, lr=0.000267261, gnorm=0.17, clip=0, loss_scale=1, train_wall=89, gb_free=22.3, wall=51749 begin validation on "valid" subset epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034 | valid on 'valid' subset | loss 3.693 | nll_loss 2.143 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 56000 | best_loss 3.689 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 455 / 1689 loss=3.54, nll_loss=1.991, ppl=3.97, wps=479126, ups=0.97, wpb=494082, bsz=16589.7, num_updates=56100, lr=0.000267023, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=51852 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 555 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=551692, ups=1.11, wpb=496337, bsz=16394.4, num_updates=56200, lr=0.000266785, gnorm=0.194, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=51942 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 655 / 1689 loss=3.549, nll_loss=2, ppl=4, wps=550199, ups=1.11, wpb=494777, bsz=16764.9, num_updates=56300, lr=0.000266548, gnorm=0.196, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52032 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 755 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553312, ups=1.11, wpb=496565, bsz=16608.6, num_updates=56400, lr=0.000266312, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.1, wall=52121 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 856 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=548983, ups=1.11, wpb=494504, bsz=16448.1, num_updates=56500, lr=0.000266076, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=20.7, wall=52211 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 956 / 1689 loss=3.546, nll_loss=1.997, ppl=3.99, wps=550057, ups=1.11, wpb=493801, bsz=16613.4, num_updates=56600, lr=0.000265841, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.2, wall=52301 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1056 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=552162, ups=1.12, wpb=495204, bsz=16203.3, num_updates=56700, lr=0.000265606, gnorm=0.185, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=52391 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1156 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=552080, ups=1.12, wpb=495017, bsz=16429.7, num_updates=56800, lr=0.000265372, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.4, wall=52481 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1257 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=543545, ups=1.1, wpb=495352, bsz=16560.5, num_updates=56900, lr=0.000265139, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.8, wall=52572 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 epoch 034: 1357 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=549543, ups=1.11, wpb=496910, bsz=16707, num_updates=57000, lr=0.000264906, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.7, wall=52662 begin validation on "valid" subset epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034 | valid on 'valid' subset | loss 3.688 | nll_loss 2.138 | ppl 4.4 | wps 0 | wpb 44526 | bsz 2008 | num_updates 57000 | best_loss 3.688 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1457 / 1689 loss=3.554, nll_loss=2.006, ppl=4.02, wps=436616, ups=0.88, wpb=495611, bsz=16493.7, num_updates=57100, lr=0.000264674, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=90, gb_free=22.4, wall=52776 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1557 / 1689 loss=3.548, nll_loss=2, ppl=4, wps=550565, ups=1.11, wpb=494126, bsz=16457.7, num_updates=57200, lr=0.000264443, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.8, wall=52865 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 epoch 034: 1657 / 1689 loss=3.545, nll_loss=1.996, ppl=3.99, wps=551535, ups=1.11, wpb=495252, bsz=16456.7, num_updates=57300, lr=0.000264212, gnorm=0.176, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.4, wall=52955 end of epoch 34 (average epoch stats below) epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 epoch 034 | loss 3.543 | nll_loss 1.994 | ppl 3.98 | wps 537186 | ups 1.08 | wpb 495117 | bsz 16501.5 | num_updates 57332 | lr 0.000264138 | gnorm 0.182 | clip 0 | loss_scale 0.5 | train_wall 1492 | gb_free 22.4 | wall 52983 Start iterating over samples epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 68 / 1689 loss=3.534, nll_loss=1.984, ppl=3.96, wps=547697, ups=1.11, wpb=492049, bsz=16341, num_updates=57400, lr=0.000263982, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=87, gb_free=22, wall=53045 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 168 / 1689 loss=3.528, nll_loss=1.976, ppl=3.94, wps=554107, ups=1.12, wpb=493996, bsz=16361.9, num_updates=57500, lr=0.000263752, gnorm=0.181, clip=0, loss_scale=1, train_wall=87, gb_free=21.4, wall=53134 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 269 / 1689 loss=3.53, nll_loss=1.979, ppl=3.94, wps=550066, ups=1.11, wpb=497002, bsz=16431.3, num_updates=57600, lr=0.000263523, gnorm=0.181, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.5, wall=53225 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 370 / 1689 loss=3.531, nll_loss=1.981, ppl=3.95, wps=549254, ups=1.11, wpb=495984, bsz=16586.6, num_updates=57700, lr=0.000263295, gnorm=0.184, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.8, wall=53315 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 470 / 1689 loss=3.542, nll_loss=1.992, ppl=3.98, wps=551104, ups=1.11, wpb=494294, bsz=16447.9, num_updates=57800, lr=0.000263067, gnorm=0.178, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.9, wall=53405 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 570 / 1689 loss=3.538, nll_loss=1.988, ppl=3.97, wps=552939, ups=1.12, wpb=495679, bsz=16170.3, num_updates=57900, lr=0.00026284, gnorm=0.19, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.6, wall=53494 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 epoch 035: 670 / 1689 loss=3.547, nll_loss=1.998, ppl=4, wps=553247, ups=1.12, wpb=494544, bsz=16496.1, num_updates=58000, lr=0.000262613, gnorm=0.193, clip=0, loss_scale=0.25, train_wall=88, gb_free=21.7, wall=53584 begin validation on "valid" subset epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.708 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 58000 | best_loss 3.688 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 770 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=486284, ups=0.98, wpb=494634, bsz=16834.7, num_updates=58100, lr=0.000262387, gnorm=0.18, clip=0, loss_scale=0.25, train_wall=87, gb_free=22, wall=53685 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 870 / 1689 loss=3.541, nll_loss=1.991, ppl=3.98, wps=555953, ups=1.12, wpb=496599, bsz=16496.6, num_updates=58200, lr=0.000262161, gnorm=0.186, clip=0, loss_scale=0.5, train_wall=88, gb_free=21, wall=53775 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 970 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=553868, ups=1.12, wpb=495341, bsz=16693, num_updates=58300, lr=0.000261936, gnorm=0.189, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=53864 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1070 / 1689 loss=3.541, nll_loss=1.992, ppl=3.98, wps=556590, ups=1.12, wpb=495861, bsz=16321.4, num_updates=58400, lr=0.000261712, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.1, wall=53953 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1170 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550152, ups=1.11, wpb=495873, bsz=16418.2, num_updates=58500, lr=0.000261488, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=89, gb_free=20.3, wall=54043 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1270 / 1689 loss=3.539, nll_loss=1.989, ppl=3.97, wps=552863, ups=1.12, wpb=494416, bsz=16246.8, num_updates=58600, lr=0.000261265, gnorm=0.196, clip=0, loss_scale=0.5, train_wall=88, gb_free=22.2, wall=54133 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1370 / 1689 loss=3.547, nll_loss=1.999, ppl=4, wps=550661, ups=1.12, wpb=493722, bsz=16520.4, num_updates=58700, lr=0.000261042, gnorm=0.178, clip=0, loss_scale=1, train_wall=88, gb_free=22.3, wall=54222 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1470 / 1689 loss=3.542, nll_loss=1.993, ppl=3.98, wps=551879, ups=1.11, wpb=496676, bsz=16533.4, num_updates=58800, lr=0.00026082, gnorm=0.174, clip=0, loss_scale=1, train_wall=88, gb_free=21.4, wall=54312 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1570 / 1689 loss=3.55, nll_loss=2.002, ppl=4.01, wps=551384, ups=1.11, wpb=495171, bsz=16860.2, num_updates=58900, lr=0.000260599, gnorm=0.18, clip=0, loss_scale=1, train_wall=88, gb_free=22.6, wall=54402 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 epoch 035: 1670 / 1689 loss=3.551, nll_loss=2.003, ppl=4.01, wps=546199, ups=1.1, wpb=495250, bsz=16799, num_updates=59000, lr=0.000260378, gnorm=0.174, clip=0, loss_scale=1, train_wall=89, gb_free=22.9, wall=54493 begin validation on "valid" subset epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 epoch 035 | valid on 'valid' subset | loss 3.696 | nll_loss 2.145 | ppl 4.42 | wps 0 | wpb 44526 | bsz 2008 | num_updates 59000 | best_loss 3.688 end of epoch 35 (average epoch stats below) epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 epoch 035 | loss 3.54 | nll_loss 1.991 | ppl 3.97 | wps 536031 | ups 1.08 | wpb 495117 | bsz 16503.8 | num_updates 59019 | lr 0.000260336 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 1490 | gb_free 23 | wall 54542 Start iterating over samples epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 81 / 1689 loss=3.524, nll_loss=1.972, ppl=3.92, wps=385642, ups=0.78, wpb=491390, bsz=16769.8, num_updates=59100, lr=0.000260157, gnorm=0.18, clip=0, loss_scale=1, train_wall=95, gb_free=22, wall=54620 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 183 / 1689 loss=3.521, nll_loss=1.969, ppl=3.91, wps=546663, ups=1.11, wpb=493543, bsz=16204.3, num_updates=59200, lr=0.000259938, gnorm=0.208, clip=0, loss_scale=0.5, train_wall=90, gb_free=21.5, wall=54711 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 283 / 1689 loss=3.528, nll_loss=1.976, ppl=3.93, wps=554266, ups=1.12, wpb=495223, bsz=16353.4, num_updates=59300, lr=0.000259718, gnorm=0.183, clip=0, loss_scale=0.5, train_wall=89, gb_free=21.6, wall=54800 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 383 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=557877, ups=1.12, wpb=496367, bsz=16691.6, num_updates=59400, lr=0.0002595, gnorm=0.185, clip=0, loss_scale=0.5, train_wall=88, gb_free=21.6, wall=54889 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 483 / 1689 loss=3.526, nll_loss=1.974, ppl=3.93, wps=551870, ups=1.11, wpb=495783, bsz=16613, num_updates=59500, lr=0.000259281, gnorm=0.171, clip=0, loss_scale=0.5, train_wall=89, gb_free=22.4, wall=54979 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 583 / 1689 loss=3.536, nll_loss=1.986, ppl=3.96, wps=552101, ups=1.12, wpb=494868, bsz=16408.6, num_updates=59600, lr=0.000259064, gnorm=0.182, clip=0, loss_scale=0.5, train_wall=88, gb_free=22, wall=55068 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 683 / 1689 loss=3.537, nll_loss=1.987, ppl=3.96, wps=554754, ups=1.12, wpb=494536, bsz=16222, num_updates=59700, lr=0.000258847, gnorm=0.18, clip=0, loss_scale=0.5, train_wall=88, gb_free=23, wall=55158 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 783 / 1689 loss=3.543, nll_loss=1.994, ppl=3.98, wps=551193, ups=1.12, wpb=493767, bsz=16203.9, num_updates=59800, lr=0.00025863, gnorm=0.175, clip=0, loss_scale=1, train_wall=88, gb_free=22, wall=55247 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 883 / 1689 loss=3.537, nll_loss=1.987, ppl=3.97, wps=550362, ups=1.11, wpb=494438, bsz=16430.1, num_updates=59900, lr=0.000258414, gnorm=0.182, clip=0, loss_scale=1, train_wall=88, gb_free=21.7, wall=55337 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 epoch 036: 983 / 1689 loss=3.537, nll_loss=1.988, ppl=3.97, wps=554426, ups=1.12, wpb=495248, bsz=16470.1, num_updates=60000, lr=0.000258199, gnorm=0.179, clip=0, loss_scale=1, train_wall=88, gb_free=22.5, wall=55426 Stopping training due to num_updates: 60000 >= max_update: 60000 begin validation on "valid" subset epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 epoch 036 | valid on 'valid' subset | loss 3.706 | nll_loss 2.157 | ppl 4.46 | wps 0 | wpb 44526 | bsz 2008 | num_updates 60000 | best_loss 3.688 end of epoch 36 (average epoch stats below) epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 epoch 036 | loss 3.532 | nll_loss 1.982 | ppl 3.95 | wps 540793 | ups 1.09 | wpb 494830 | bsz 16447.5 | num_updates 60000 | lr 0.000258199 | gnorm 0.183 | clip 0 | loss_scale 1 | train_wall 871 | gb_free 22.5 | wall 55439 done training in 55425.5 seconds