Upload folder using huggingface_hub
Browse files- added_tokens.json +5 -0
- gliner_config.json +144 -0
- pytorch_model.bin +3 -0
- special_tokens_map.json +15 -0
- spm.model +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +75 -0
- trainer_state.json +1465 -0
added_tokens.json
ADDED
|
@@ -0,0 +1,5 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"<<ENT>>": 128001,
|
| 3 |
+
"<<SEP>>": 128002,
|
| 4 |
+
"[MASK]": 128000
|
| 5 |
+
}
|
gliner_config.json
ADDED
|
@@ -0,0 +1,144 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"class_token_index": 128001,
|
| 3 |
+
"dropout": 0.4,
|
| 4 |
+
"embed_ent_token": true,
|
| 5 |
+
"encoder_config": {
|
| 6 |
+
"_attn_implementation_autoset": false,
|
| 7 |
+
"_name_or_path": "microsoft/deberta-v3-base",
|
| 8 |
+
"add_cross_attention": false,
|
| 9 |
+
"architectures": null,
|
| 10 |
+
"attention_probs_dropout_prob": 0.1,
|
| 11 |
+
"bad_words_ids": null,
|
| 12 |
+
"begin_suppress_tokens": null,
|
| 13 |
+
"bos_token_id": null,
|
| 14 |
+
"chunk_size_feed_forward": 0,
|
| 15 |
+
"cross_attention_hidden_size": null,
|
| 16 |
+
"decoder_start_token_id": null,
|
| 17 |
+
"diversity_penalty": 0.0,
|
| 18 |
+
"do_sample": false,
|
| 19 |
+
"early_stopping": false,
|
| 20 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 21 |
+
"eos_token_id": null,
|
| 22 |
+
"exponential_decay_length_penalty": null,
|
| 23 |
+
"finetuning_task": null,
|
| 24 |
+
"forced_bos_token_id": null,
|
| 25 |
+
"forced_eos_token_id": null,
|
| 26 |
+
"hidden_act": "gelu",
|
| 27 |
+
"hidden_dropout_prob": 0.1,
|
| 28 |
+
"hidden_size": 768,
|
| 29 |
+
"id2label": {
|
| 30 |
+
"0": "LABEL_0",
|
| 31 |
+
"1": "LABEL_1"
|
| 32 |
+
},
|
| 33 |
+
"initializer_range": 0.02,
|
| 34 |
+
"intermediate_size": 3072,
|
| 35 |
+
"is_decoder": false,
|
| 36 |
+
"is_encoder_decoder": false,
|
| 37 |
+
"label2id": {
|
| 38 |
+
"LABEL_0": 0,
|
| 39 |
+
"LABEL_1": 1
|
| 40 |
+
},
|
| 41 |
+
"layer_norm_eps": 1e-07,
|
| 42 |
+
"legacy": true,
|
| 43 |
+
"length_penalty": 1.0,
|
| 44 |
+
"max_length": 20,
|
| 45 |
+
"max_position_embeddings": 512,
|
| 46 |
+
"max_relative_positions": -1,
|
| 47 |
+
"min_length": 0,
|
| 48 |
+
"model_type": "deberta-v2",
|
| 49 |
+
"no_repeat_ngram_size": 0,
|
| 50 |
+
"norm_rel_ebd": "layer_norm",
|
| 51 |
+
"num_attention_heads": 12,
|
| 52 |
+
"num_beam_groups": 1,
|
| 53 |
+
"num_beams": 1,
|
| 54 |
+
"num_hidden_layers": 12,
|
| 55 |
+
"num_return_sequences": 1,
|
| 56 |
+
"output_attentions": false,
|
| 57 |
+
"output_hidden_states": false,
|
| 58 |
+
"output_scores": false,
|
| 59 |
+
"pad_token_id": 0,
|
| 60 |
+
"pooler_dropout": 0,
|
| 61 |
+
"pooler_hidden_act": "gelu",
|
| 62 |
+
"pooler_hidden_size": 768,
|
| 63 |
+
"pos_att_type": [
|
| 64 |
+
"p2c",
|
| 65 |
+
"c2p"
|
| 66 |
+
],
|
| 67 |
+
"position_biased_input": false,
|
| 68 |
+
"position_buckets": 256,
|
| 69 |
+
"prefix": null,
|
| 70 |
+
"problem_type": null,
|
| 71 |
+
"pruned_heads": {},
|
| 72 |
+
"relative_attention": true,
|
| 73 |
+
"remove_invalid_values": false,
|
| 74 |
+
"repetition_penalty": 1.0,
|
| 75 |
+
"return_dict": true,
|
| 76 |
+
"return_dict_in_generate": false,
|
| 77 |
+
"sep_token_id": null,
|
| 78 |
+
"share_att_key": true,
|
| 79 |
+
"suppress_tokens": null,
|
| 80 |
+
"task_specific_params": null,
|
| 81 |
+
"temperature": 1.0,
|
| 82 |
+
"tf_legacy_loss": false,
|
| 83 |
+
"tie_encoder_decoder": false,
|
| 84 |
+
"tie_word_embeddings": true,
|
| 85 |
+
"tokenizer_class": null,
|
| 86 |
+
"top_k": 50,
|
| 87 |
+
"top_p": 1.0,
|
| 88 |
+
"torch_dtype": null,
|
| 89 |
+
"torchscript": false,
|
| 90 |
+
"type_vocab_size": 0,
|
| 91 |
+
"typical_p": 1.0,
|
| 92 |
+
"use_bfloat16": false,
|
| 93 |
+
"vocab_size": 128003
|
| 94 |
+
},
|
| 95 |
+
"ent_token": "<<ENT>>",
|
| 96 |
+
"eval_every": 10000,
|
| 97 |
+
"fine_tune": true,
|
| 98 |
+
"freeze_token_rep": false,
|
| 99 |
+
"fuse_layers": false,
|
| 100 |
+
"has_rnn": true,
|
| 101 |
+
"hidden_size": 768,
|
| 102 |
+
"label_smoothing": 0.0,
|
| 103 |
+
"labels_encoder": null,
|
| 104 |
+
"labels_encoder_config": null,
|
| 105 |
+
"labels_fusion_schema": "",
|
| 106 |
+
"log_dir": "models/",
|
| 107 |
+
"loss_alpha": 0.75,
|
| 108 |
+
"loss_gamma": -1,
|
| 109 |
+
"loss_reduction": "sum",
|
| 110 |
+
"lr_encoder": "1e-5",
|
| 111 |
+
"lr_others": "5e-5",
|
| 112 |
+
"max_grad_norm": 1.0,
|
| 113 |
+
"max_len": 1024,
|
| 114 |
+
"max_neg_type_ratio": 1,
|
| 115 |
+
"max_types": 30,
|
| 116 |
+
"max_width": 12,
|
| 117 |
+
"model_name": "microsoft/deberta-v3-base",
|
| 118 |
+
"model_type": "gliner",
|
| 119 |
+
"name": "span level gliner",
|
| 120 |
+
"num_post_fusion_layers": 1,
|
| 121 |
+
"num_rnns": 1,
|
| 122 |
+
"num_steps": 20000,
|
| 123 |
+
"post_fusion_schema": "",
|
| 124 |
+
"pre_fusion": true,
|
| 125 |
+
"prev_path": null,
|
| 126 |
+
"random_drop": true,
|
| 127 |
+
"root_dir": "gliner_logs",
|
| 128 |
+
"save_total_limit": 5,
|
| 129 |
+
"scheduler_type": "linear",
|
| 130 |
+
"sep_token": "<<SEP>>",
|
| 131 |
+
"shuffle_types": true,
|
| 132 |
+
"size_sup": -1,
|
| 133 |
+
"span_mode": "markerV0",
|
| 134 |
+
"subtoken_pooling": "first",
|
| 135 |
+
"train_batch_size": 4,
|
| 136 |
+
"train_data": "../data/post_ner.json",
|
| 137 |
+
"transformers_version": "4.49.0",
|
| 138 |
+
"val_data_dir": "none",
|
| 139 |
+
"vocab_size": 128003,
|
| 140 |
+
"warmup_ratio": 0.1,
|
| 141 |
+
"weight_decay_encoder": 0.01,
|
| 142 |
+
"weight_decay_other": 0.01,
|
| 143 |
+
"words_splitter_type": "whitespace"
|
| 144 |
+
}
|
pytorch_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3ca34c5c678838c3d10e957a54bfdd58e0831ed20097b542ac1e2e770de09644
|
| 3 |
+
size 834285094
|
special_tokens_map.json
ADDED
|
@@ -0,0 +1,15 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": "[CLS]",
|
| 3 |
+
"cls_token": "[CLS]",
|
| 4 |
+
"eos_token": "[SEP]",
|
| 5 |
+
"mask_token": "[MASK]",
|
| 6 |
+
"pad_token": "[PAD]",
|
| 7 |
+
"sep_token": "[SEP]",
|
| 8 |
+
"unk_token": {
|
| 9 |
+
"content": "[UNK]",
|
| 10 |
+
"lstrip": false,
|
| 11 |
+
"normalized": true,
|
| 12 |
+
"rstrip": false,
|
| 13 |
+
"single_word": false
|
| 14 |
+
}
|
| 15 |
+
}
|
spm.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
|
| 3 |
+
size 2464616
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"added_tokens_decoder": {
|
| 3 |
+
"0": {
|
| 4 |
+
"content": "[PAD]",
|
| 5 |
+
"lstrip": false,
|
| 6 |
+
"normalized": false,
|
| 7 |
+
"rstrip": false,
|
| 8 |
+
"single_word": false,
|
| 9 |
+
"special": true
|
| 10 |
+
},
|
| 11 |
+
"1": {
|
| 12 |
+
"content": "[CLS]",
|
| 13 |
+
"lstrip": false,
|
| 14 |
+
"normalized": false,
|
| 15 |
+
"rstrip": false,
|
| 16 |
+
"single_word": false,
|
| 17 |
+
"special": true
|
| 18 |
+
},
|
| 19 |
+
"2": {
|
| 20 |
+
"content": "[SEP]",
|
| 21 |
+
"lstrip": false,
|
| 22 |
+
"normalized": false,
|
| 23 |
+
"rstrip": false,
|
| 24 |
+
"single_word": false,
|
| 25 |
+
"special": true
|
| 26 |
+
},
|
| 27 |
+
"3": {
|
| 28 |
+
"content": "[UNK]",
|
| 29 |
+
"lstrip": false,
|
| 30 |
+
"normalized": true,
|
| 31 |
+
"rstrip": false,
|
| 32 |
+
"single_word": false,
|
| 33 |
+
"special": true
|
| 34 |
+
},
|
| 35 |
+
"128000": {
|
| 36 |
+
"content": "[MASK]",
|
| 37 |
+
"lstrip": false,
|
| 38 |
+
"normalized": false,
|
| 39 |
+
"rstrip": false,
|
| 40 |
+
"single_word": false,
|
| 41 |
+
"special": true
|
| 42 |
+
},
|
| 43 |
+
"128001": {
|
| 44 |
+
"content": "<<ENT>>",
|
| 45 |
+
"lstrip": false,
|
| 46 |
+
"normalized": false,
|
| 47 |
+
"rstrip": false,
|
| 48 |
+
"single_word": false,
|
| 49 |
+
"special": true
|
| 50 |
+
},
|
| 51 |
+
"128002": {
|
| 52 |
+
"content": "<<SEP>>",
|
| 53 |
+
"lstrip": false,
|
| 54 |
+
"normalized": false,
|
| 55 |
+
"rstrip": false,
|
| 56 |
+
"single_word": false,
|
| 57 |
+
"special": true
|
| 58 |
+
}
|
| 59 |
+
},
|
| 60 |
+
"bos_token": "[CLS]",
|
| 61 |
+
"clean_up_tokenization_spaces": false,
|
| 62 |
+
"cls_token": "[CLS]",
|
| 63 |
+
"do_lower_case": false,
|
| 64 |
+
"eos_token": "[SEP]",
|
| 65 |
+
"extra_special_tokens": {},
|
| 66 |
+
"mask_token": "[MASK]",
|
| 67 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 68 |
+
"pad_token": "[PAD]",
|
| 69 |
+
"sep_token": "[SEP]",
|
| 70 |
+
"sp_model_kwargs": {},
|
| 71 |
+
"split_by_punct": false,
|
| 72 |
+
"tokenizer_class": "DebertaV2Tokenizer",
|
| 73 |
+
"unk_token": "[UNK]",
|
| 74 |
+
"vocab_type": "spm"
|
| 75 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,1465 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 4.423800044238001,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 20000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.022119000221190004,
|
| 13 |
+
"grad_norm": 2561.85302734375,
|
| 14 |
+
"learning_rate": 2.5e-06,
|
| 15 |
+
"loss": 7244.9438,
|
| 16 |
+
"step": 100
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.04423800044238001,
|
| 20 |
+
"grad_norm": 1641.6300048828125,
|
| 21 |
+
"learning_rate": 5e-06,
|
| 22 |
+
"loss": 278.0641,
|
| 23 |
+
"step": 200
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.06635700066357,
|
| 27 |
+
"grad_norm": 413.9412841796875,
|
| 28 |
+
"learning_rate": 7.5e-06,
|
| 29 |
+
"loss": 305.1475,
|
| 30 |
+
"step": 300
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.08847600088476001,
|
| 34 |
+
"grad_norm": 298.65997314453125,
|
| 35 |
+
"learning_rate": 1e-05,
|
| 36 |
+
"loss": 296.6687,
|
| 37 |
+
"step": 400
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.11059500110595001,
|
| 41 |
+
"grad_norm": 612.1175537109375,
|
| 42 |
+
"learning_rate": 1.25e-05,
|
| 43 |
+
"loss": 247.2131,
|
| 44 |
+
"step": 500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.13271400132714,
|
| 48 |
+
"grad_norm": 1398.299560546875,
|
| 49 |
+
"learning_rate": 1.5e-05,
|
| 50 |
+
"loss": 273.615,
|
| 51 |
+
"step": 600
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.15483300154833002,
|
| 55 |
+
"grad_norm": 538.8013305664062,
|
| 56 |
+
"learning_rate": 1.75e-05,
|
| 57 |
+
"loss": 277.284,
|
| 58 |
+
"step": 700
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.17695200176952003,
|
| 62 |
+
"grad_norm": 834.0810546875,
|
| 63 |
+
"learning_rate": 2e-05,
|
| 64 |
+
"loss": 343.3682,
|
| 65 |
+
"step": 800
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.19907100199071,
|
| 69 |
+
"grad_norm": 514.625732421875,
|
| 70 |
+
"learning_rate": 2.25e-05,
|
| 71 |
+
"loss": 172.4172,
|
| 72 |
+
"step": 900
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.22119000221190002,
|
| 76 |
+
"grad_norm": 720.7930908203125,
|
| 77 |
+
"learning_rate": 2.5e-05,
|
| 78 |
+
"loss": 243.5434,
|
| 79 |
+
"step": 1000
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.24330900243309003,
|
| 83 |
+
"grad_norm": 538.9693603515625,
|
| 84 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 85 |
+
"loss": 227.322,
|
| 86 |
+
"step": 1100
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.26542800265428,
|
| 90 |
+
"grad_norm": 365.2975769042969,
|
| 91 |
+
"learning_rate": 3e-05,
|
| 92 |
+
"loss": 289.7278,
|
| 93 |
+
"step": 1200
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.28754700287547,
|
| 97 |
+
"grad_norm": 1308.2694091796875,
|
| 98 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 99 |
+
"loss": 191.7522,
|
| 100 |
+
"step": 1300
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.30966600309666004,
|
| 104 |
+
"grad_norm": 6047.15625,
|
| 105 |
+
"learning_rate": 3.5e-05,
|
| 106 |
+
"loss": 316.0845,
|
| 107 |
+
"step": 1400
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.33178500331785005,
|
| 111 |
+
"grad_norm": 482.5833740234375,
|
| 112 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 113 |
+
"loss": 275.0296,
|
| 114 |
+
"step": 1500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.35390400353904006,
|
| 118 |
+
"grad_norm": 811.97607421875,
|
| 119 |
+
"learning_rate": 4e-05,
|
| 120 |
+
"loss": 232.9227,
|
| 121 |
+
"step": 1600
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.37602300376023,
|
| 125 |
+
"grad_norm": 585.74365234375,
|
| 126 |
+
"learning_rate": 4.25e-05,
|
| 127 |
+
"loss": 157.1875,
|
| 128 |
+
"step": 1700
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.39814200398142,
|
| 132 |
+
"grad_norm": 2657.204833984375,
|
| 133 |
+
"learning_rate": 4.5e-05,
|
| 134 |
+
"loss": 202.9244,
|
| 135 |
+
"step": 1800
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.42026100420261003,
|
| 139 |
+
"grad_norm": 788.7911376953125,
|
| 140 |
+
"learning_rate": 4.75e-05,
|
| 141 |
+
"loss": 230.3789,
|
| 142 |
+
"step": 1900
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.44238000442380004,
|
| 146 |
+
"grad_norm": 1466.7137451171875,
|
| 147 |
+
"learning_rate": 5e-05,
|
| 148 |
+
"loss": 184.6603,
|
| 149 |
+
"step": 2000
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.46449900464499005,
|
| 153 |
+
"grad_norm": 717.4619750976562,
|
| 154 |
+
"learning_rate": 4.972222222222223e-05,
|
| 155 |
+
"loss": 160.0884,
|
| 156 |
+
"step": 2100
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.48661800486618007,
|
| 160 |
+
"grad_norm": 1592.6282958984375,
|
| 161 |
+
"learning_rate": 4.9444444444444446e-05,
|
| 162 |
+
"loss": 159.9456,
|
| 163 |
+
"step": 2200
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.5087370050873701,
|
| 167 |
+
"grad_norm": 258.3177795410156,
|
| 168 |
+
"learning_rate": 4.9166666666666665e-05,
|
| 169 |
+
"loss": 272.3899,
|
| 170 |
+
"step": 2300
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.53085600530856,
|
| 174 |
+
"grad_norm": 2940.7119140625,
|
| 175 |
+
"learning_rate": 4.888888888888889e-05,
|
| 176 |
+
"loss": 108.9058,
|
| 177 |
+
"step": 2400
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.5529750055297501,
|
| 181 |
+
"grad_norm": 1595.506103515625,
|
| 182 |
+
"learning_rate": 4.8611111111111115e-05,
|
| 183 |
+
"loss": 137.0249,
|
| 184 |
+
"step": 2500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.57509400575094,
|
| 188 |
+
"grad_norm": 476.0992126464844,
|
| 189 |
+
"learning_rate": 4.8333333333333334e-05,
|
| 190 |
+
"loss": 157.8203,
|
| 191 |
+
"step": 2600
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.59721300597213,
|
| 195 |
+
"grad_norm": 462.79119873046875,
|
| 196 |
+
"learning_rate": 4.805555555555556e-05,
|
| 197 |
+
"loss": 202.3364,
|
| 198 |
+
"step": 2700
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.6193320061933201,
|
| 202 |
+
"grad_norm": 7305.47900390625,
|
| 203 |
+
"learning_rate": 4.7777777777777784e-05,
|
| 204 |
+
"loss": 176.8393,
|
| 205 |
+
"step": 2800
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.64145100641451,
|
| 209 |
+
"grad_norm": 1471.9434814453125,
|
| 210 |
+
"learning_rate": 4.75e-05,
|
| 211 |
+
"loss": 192.2721,
|
| 212 |
+
"step": 2900
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.6635700066357001,
|
| 216 |
+
"grad_norm": 253.4272003173828,
|
| 217 |
+
"learning_rate": 4.722222222222222e-05,
|
| 218 |
+
"loss": 204.4471,
|
| 219 |
+
"step": 3000
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.68568900685689,
|
| 223 |
+
"grad_norm": 4707.13916015625,
|
| 224 |
+
"learning_rate": 4.6944444444444446e-05,
|
| 225 |
+
"loss": 154.4462,
|
| 226 |
+
"step": 3100
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.7078080070780801,
|
| 230 |
+
"grad_norm": 477.1028747558594,
|
| 231 |
+
"learning_rate": 4.666666666666667e-05,
|
| 232 |
+
"loss": 93.3721,
|
| 233 |
+
"step": 3200
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.7299270072992701,
|
| 237 |
+
"grad_norm": 492.2008361816406,
|
| 238 |
+
"learning_rate": 4.638888888888889e-05,
|
| 239 |
+
"loss": 129.002,
|
| 240 |
+
"step": 3300
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.75204600752046,
|
| 244 |
+
"grad_norm": 414.3881530761719,
|
| 245 |
+
"learning_rate": 4.6111111111111115e-05,
|
| 246 |
+
"loss": 88.6518,
|
| 247 |
+
"step": 3400
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.7741650077416501,
|
| 251 |
+
"grad_norm": 170.11138916015625,
|
| 252 |
+
"learning_rate": 4.5833333333333334e-05,
|
| 253 |
+
"loss": 193.1568,
|
| 254 |
+
"step": 3500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.79628400796284,
|
| 258 |
+
"grad_norm": 787.8067016601562,
|
| 259 |
+
"learning_rate": 4.555555555555556e-05,
|
| 260 |
+
"loss": 153.1195,
|
| 261 |
+
"step": 3600
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.8184030081840301,
|
| 265 |
+
"grad_norm": 940.7820434570312,
|
| 266 |
+
"learning_rate": 4.527777777777778e-05,
|
| 267 |
+
"loss": 312.0311,
|
| 268 |
+
"step": 3700
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.8405220084052201,
|
| 272 |
+
"grad_norm": 719.3717041015625,
|
| 273 |
+
"learning_rate": 4.5e-05,
|
| 274 |
+
"loss": 160.2157,
|
| 275 |
+
"step": 3800
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.8626410086264101,
|
| 279 |
+
"grad_norm": 710.9347534179688,
|
| 280 |
+
"learning_rate": 4.472222222222223e-05,
|
| 281 |
+
"loss": 149.8423,
|
| 282 |
+
"step": 3900
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.8847600088476001,
|
| 286 |
+
"grad_norm": 812.0198364257812,
|
| 287 |
+
"learning_rate": 4.4444444444444447e-05,
|
| 288 |
+
"loss": 171.4201,
|
| 289 |
+
"step": 4000
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.90687900906879,
|
| 293 |
+
"grad_norm": 347.59246826171875,
|
| 294 |
+
"learning_rate": 4.4166666666666665e-05,
|
| 295 |
+
"loss": 85.4954,
|
| 296 |
+
"step": 4100
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.9289980092899801,
|
| 300 |
+
"grad_norm": 494.03570556640625,
|
| 301 |
+
"learning_rate": 4.388888888888889e-05,
|
| 302 |
+
"loss": 122.7032,
|
| 303 |
+
"step": 4200
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.9511170095111701,
|
| 307 |
+
"grad_norm": 1107.2685546875,
|
| 308 |
+
"learning_rate": 4.3611111111111116e-05,
|
| 309 |
+
"loss": 110.0679,
|
| 310 |
+
"step": 4300
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.9732360097323601,
|
| 314 |
+
"grad_norm": 554.3077392578125,
|
| 315 |
+
"learning_rate": 4.3333333333333334e-05,
|
| 316 |
+
"loss": 162.5302,
|
| 317 |
+
"step": 4400
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.9953550099535501,
|
| 321 |
+
"grad_norm": 1022.9378662109375,
|
| 322 |
+
"learning_rate": 4.305555555555556e-05,
|
| 323 |
+
"loss": 123.3146,
|
| 324 |
+
"step": 4500
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.0,
|
| 328 |
+
"eval_loss": 317.1508483886719,
|
| 329 |
+
"eval_runtime": 20.7601,
|
| 330 |
+
"eval_samples_per_second": 96.82,
|
| 331 |
+
"eval_steps_per_second": 24.229,
|
| 332 |
+
"step": 4521
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.0174740101747402,
|
| 336 |
+
"grad_norm": 604.2803344726562,
|
| 337 |
+
"learning_rate": 4.277777777777778e-05,
|
| 338 |
+
"loss": 87.0649,
|
| 339 |
+
"step": 4600
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 1.03959301039593,
|
| 343 |
+
"grad_norm": 655.1663208007812,
|
| 344 |
+
"learning_rate": 4.25e-05,
|
| 345 |
+
"loss": 60.0608,
|
| 346 |
+
"step": 4700
|
| 347 |
+
},
|
| 348 |
+
{
|
| 349 |
+
"epoch": 1.06171201061712,
|
| 350 |
+
"grad_norm": 613.4644165039062,
|
| 351 |
+
"learning_rate": 4.222222222222222e-05,
|
| 352 |
+
"loss": 152.5058,
|
| 353 |
+
"step": 4800
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 1.08383101083831,
|
| 357 |
+
"grad_norm": 943.52392578125,
|
| 358 |
+
"learning_rate": 4.194444444444445e-05,
|
| 359 |
+
"loss": 112.7452,
|
| 360 |
+
"step": 4900
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"epoch": 1.1059500110595002,
|
| 364 |
+
"grad_norm": 741.4888916015625,
|
| 365 |
+
"learning_rate": 4.166666666666667e-05,
|
| 366 |
+
"loss": 180.4975,
|
| 367 |
+
"step": 5000
|
| 368 |
+
},
|
| 369 |
+
{
|
| 370 |
+
"epoch": 1.1280690112806901,
|
| 371 |
+
"grad_norm": 984.3648071289062,
|
| 372 |
+
"learning_rate": 4.138888888888889e-05,
|
| 373 |
+
"loss": 109.8211,
|
| 374 |
+
"step": 5100
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.15018801150188,
|
| 378 |
+
"grad_norm": 4328.1767578125,
|
| 379 |
+
"learning_rate": 4.111111111111111e-05,
|
| 380 |
+
"loss": 86.7452,
|
| 381 |
+
"step": 5200
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 1.17230701172307,
|
| 385 |
+
"grad_norm": 440.3537902832031,
|
| 386 |
+
"learning_rate": 4.0833333333333334e-05,
|
| 387 |
+
"loss": 145.0485,
|
| 388 |
+
"step": 5300
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 1.1944260119442602,
|
| 392 |
+
"grad_norm": 1904.060302734375,
|
| 393 |
+
"learning_rate": 4.055555555555556e-05,
|
| 394 |
+
"loss": 137.598,
|
| 395 |
+
"step": 5400
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 1.2165450121654502,
|
| 399 |
+
"grad_norm": 325.7207336425781,
|
| 400 |
+
"learning_rate": 4.027777777777778e-05,
|
| 401 |
+
"loss": 133.5432,
|
| 402 |
+
"step": 5500
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 1.2386640123866401,
|
| 406 |
+
"grad_norm": 602.099609375,
|
| 407 |
+
"learning_rate": 4e-05,
|
| 408 |
+
"loss": 90.936,
|
| 409 |
+
"step": 5600
|
| 410 |
+
},
|
| 411 |
+
{
|
| 412 |
+
"epoch": 1.26078301260783,
|
| 413 |
+
"grad_norm": 373.9248962402344,
|
| 414 |
+
"learning_rate": 3.972222222222222e-05,
|
| 415 |
+
"loss": 126.1215,
|
| 416 |
+
"step": 5700
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.28290201282902,
|
| 420 |
+
"grad_norm": 1740.2845458984375,
|
| 421 |
+
"learning_rate": 3.944444444444445e-05,
|
| 422 |
+
"loss": 106.4278,
|
| 423 |
+
"step": 5800
|
| 424 |
+
},
|
| 425 |
+
{
|
| 426 |
+
"epoch": 1.3050210130502102,
|
| 427 |
+
"grad_norm": 619.2306518554688,
|
| 428 |
+
"learning_rate": 3.9166666666666665e-05,
|
| 429 |
+
"loss": 69.9521,
|
| 430 |
+
"step": 5900
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"epoch": 1.3271400132714002,
|
| 434 |
+
"grad_norm": 397.7071838378906,
|
| 435 |
+
"learning_rate": 3.888888888888889e-05,
|
| 436 |
+
"loss": 92.6775,
|
| 437 |
+
"step": 6000
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 1.3492590134925901,
|
| 441 |
+
"grad_norm": 261.9856872558594,
|
| 442 |
+
"learning_rate": 3.8611111111111116e-05,
|
| 443 |
+
"loss": 139.3951,
|
| 444 |
+
"step": 6100
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 1.37137801371378,
|
| 448 |
+
"grad_norm": 326.20806884765625,
|
| 449 |
+
"learning_rate": 3.8333333333333334e-05,
|
| 450 |
+
"loss": 102.4056,
|
| 451 |
+
"step": 6200
|
| 452 |
+
},
|
| 453 |
+
{
|
| 454 |
+
"epoch": 1.39349701393497,
|
| 455 |
+
"grad_norm": 5344.56005859375,
|
| 456 |
+
"learning_rate": 3.805555555555555e-05,
|
| 457 |
+
"loss": 108.2192,
|
| 458 |
+
"step": 6300
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.4156160141561602,
|
| 462 |
+
"grad_norm": 1083.1983642578125,
|
| 463 |
+
"learning_rate": 3.777777777777778e-05,
|
| 464 |
+
"loss": 121.1852,
|
| 465 |
+
"step": 6400
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.4377350143773502,
|
| 469 |
+
"grad_norm": 285.3344421386719,
|
| 470 |
+
"learning_rate": 3.7500000000000003e-05,
|
| 471 |
+
"loss": 113.1963,
|
| 472 |
+
"step": 6500
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.4598540145985401,
|
| 476 |
+
"grad_norm": 1590.1190185546875,
|
| 477 |
+
"learning_rate": 3.722222222222222e-05,
|
| 478 |
+
"loss": 101.6212,
|
| 479 |
+
"step": 6600
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.48197301481973,
|
| 483 |
+
"grad_norm": 242.43016052246094,
|
| 484 |
+
"learning_rate": 3.694444444444445e-05,
|
| 485 |
+
"loss": 100.9942,
|
| 486 |
+
"step": 6700
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.50409201504092,
|
| 490 |
+
"grad_norm": 419.2840576171875,
|
| 491 |
+
"learning_rate": 3.6666666666666666e-05,
|
| 492 |
+
"loss": 91.4742,
|
| 493 |
+
"step": 6800
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.5262110152621102,
|
| 497 |
+
"grad_norm": 145.44712829589844,
|
| 498 |
+
"learning_rate": 3.638888888888889e-05,
|
| 499 |
+
"loss": 108.68,
|
| 500 |
+
"step": 6900
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.5483300154833002,
|
| 504 |
+
"grad_norm": 866.0416259765625,
|
| 505 |
+
"learning_rate": 3.611111111111111e-05,
|
| 506 |
+
"loss": 139.873,
|
| 507 |
+
"step": 7000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.5704490157044901,
|
| 511 |
+
"grad_norm": 543.1744995117188,
|
| 512 |
+
"learning_rate": 3.5833333333333335e-05,
|
| 513 |
+
"loss": 84.9976,
|
| 514 |
+
"step": 7100
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.5925680159256803,
|
| 518 |
+
"grad_norm": 961.9137573242188,
|
| 519 |
+
"learning_rate": 3.555555555555556e-05,
|
| 520 |
+
"loss": 91.9599,
|
| 521 |
+
"step": 7200
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.61468701614687,
|
| 525 |
+
"grad_norm": 1136.900146484375,
|
| 526 |
+
"learning_rate": 3.527777777777778e-05,
|
| 527 |
+
"loss": 100.5749,
|
| 528 |
+
"step": 7300
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.6368060163680602,
|
| 532 |
+
"grad_norm": 845.6568603515625,
|
| 533 |
+
"learning_rate": 3.5e-05,
|
| 534 |
+
"loss": 128.4079,
|
| 535 |
+
"step": 7400
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.6589250165892502,
|
| 539 |
+
"grad_norm": 1455.4808349609375,
|
| 540 |
+
"learning_rate": 3.472222222222222e-05,
|
| 541 |
+
"loss": 124.5921,
|
| 542 |
+
"step": 7500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.6810440168104401,
|
| 546 |
+
"grad_norm": 370.6483459472656,
|
| 547 |
+
"learning_rate": 3.444444444444445e-05,
|
| 548 |
+
"loss": 103.7647,
|
| 549 |
+
"step": 7600
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 1.7031630170316303,
|
| 553 |
+
"grad_norm": 9296.732421875,
|
| 554 |
+
"learning_rate": 3.4166666666666666e-05,
|
| 555 |
+
"loss": 104.9438,
|
| 556 |
+
"step": 7700
|
| 557 |
+
},
|
| 558 |
+
{
|
| 559 |
+
"epoch": 1.72528201725282,
|
| 560 |
+
"grad_norm": 2471.158447265625,
|
| 561 |
+
"learning_rate": 3.388888888888889e-05,
|
| 562 |
+
"loss": 109.1481,
|
| 563 |
+
"step": 7800
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 1.7474010174740102,
|
| 567 |
+
"grad_norm": 230.1200408935547,
|
| 568 |
+
"learning_rate": 3.3611111111111116e-05,
|
| 569 |
+
"loss": 87.4035,
|
| 570 |
+
"step": 7900
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 1.7695200176952002,
|
| 574 |
+
"grad_norm": 875.7386474609375,
|
| 575 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 576 |
+
"loss": 87.8452,
|
| 577 |
+
"step": 8000
|
| 578 |
+
},
|
| 579 |
+
{
|
| 580 |
+
"epoch": 1.7916390179163901,
|
| 581 |
+
"grad_norm": 1371.0994873046875,
|
| 582 |
+
"learning_rate": 3.3055555555555553e-05,
|
| 583 |
+
"loss": 98.3798,
|
| 584 |
+
"step": 8100
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 1.8137580181375803,
|
| 588 |
+
"grad_norm": 682.475341796875,
|
| 589 |
+
"learning_rate": 3.277777777777778e-05,
|
| 590 |
+
"loss": 104.5044,
|
| 591 |
+
"step": 8200
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 1.83587701835877,
|
| 595 |
+
"grad_norm": 512.0708618164062,
|
| 596 |
+
"learning_rate": 3.2500000000000004e-05,
|
| 597 |
+
"loss": 64.1792,
|
| 598 |
+
"step": 8300
|
| 599 |
+
},
|
| 600 |
+
{
|
| 601 |
+
"epoch": 1.8579960185799602,
|
| 602 |
+
"grad_norm": 1042.0284423828125,
|
| 603 |
+
"learning_rate": 3.222222222222223e-05,
|
| 604 |
+
"loss": 74.1599,
|
| 605 |
+
"step": 8400
|
| 606 |
+
},
|
| 607 |
+
{
|
| 608 |
+
"epoch": 1.8801150188011502,
|
| 609 |
+
"grad_norm": 358.9875183105469,
|
| 610 |
+
"learning_rate": 3.194444444444444e-05,
|
| 611 |
+
"loss": 93.9574,
|
| 612 |
+
"step": 8500
|
| 613 |
+
},
|
| 614 |
+
{
|
| 615 |
+
"epoch": 1.9022340190223401,
|
| 616 |
+
"grad_norm": 4613.166015625,
|
| 617 |
+
"learning_rate": 3.1666666666666666e-05,
|
| 618 |
+
"loss": 103.6234,
|
| 619 |
+
"step": 8600
|
| 620 |
+
},
|
| 621 |
+
{
|
| 622 |
+
"epoch": 1.9243530192435303,
|
| 623 |
+
"grad_norm": 436.9840393066406,
|
| 624 |
+
"learning_rate": 3.138888888888889e-05,
|
| 625 |
+
"loss": 98.4847,
|
| 626 |
+
"step": 8700
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.94647201946472,
|
| 630 |
+
"grad_norm": 29151.51171875,
|
| 631 |
+
"learning_rate": 3.111111111111111e-05,
|
| 632 |
+
"loss": 96.9473,
|
| 633 |
+
"step": 8800
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.9685910196859102,
|
| 637 |
+
"grad_norm": 189.884033203125,
|
| 638 |
+
"learning_rate": 3.0833333333333335e-05,
|
| 639 |
+
"loss": 65.5306,
|
| 640 |
+
"step": 8900
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.9907100199071002,
|
| 644 |
+
"grad_norm": 416.74969482421875,
|
| 645 |
+
"learning_rate": 3.055555555555556e-05,
|
| 646 |
+
"loss": 104.1101,
|
| 647 |
+
"step": 9000
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 2.0,
|
| 651 |
+
"eval_loss": 227.4776611328125,
|
| 652 |
+
"eval_runtime": 20.702,
|
| 653 |
+
"eval_samples_per_second": 97.092,
|
| 654 |
+
"eval_steps_per_second": 24.297,
|
| 655 |
+
"step": 9042
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 2.01282902012829,
|
| 659 |
+
"grad_norm": 345.6188049316406,
|
| 660 |
+
"learning_rate": 3.0277777777777776e-05,
|
| 661 |
+
"loss": 103.7359,
|
| 662 |
+
"step": 9100
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 2.0349480203494803,
|
| 666 |
+
"grad_norm": 2839.026611328125,
|
| 667 |
+
"learning_rate": 3e-05,
|
| 668 |
+
"loss": 83.4189,
|
| 669 |
+
"step": 9200
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 2.05706702057067,
|
| 673 |
+
"grad_norm": 493.89093017578125,
|
| 674 |
+
"learning_rate": 2.9722222222222223e-05,
|
| 675 |
+
"loss": 93.5555,
|
| 676 |
+
"step": 9300
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 2.07918602079186,
|
| 680 |
+
"grad_norm": 239.73707580566406,
|
| 681 |
+
"learning_rate": 2.9444444444444448e-05,
|
| 682 |
+
"loss": 102.3052,
|
| 683 |
+
"step": 9400
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 2.1013050210130504,
|
| 687 |
+
"grad_norm": 293.8266296386719,
|
| 688 |
+
"learning_rate": 2.916666666666667e-05,
|
| 689 |
+
"loss": 80.1726,
|
| 690 |
+
"step": 9500
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 2.12342402123424,
|
| 694 |
+
"grad_norm": 251.94790649414062,
|
| 695 |
+
"learning_rate": 2.8888888888888888e-05,
|
| 696 |
+
"loss": 57.1753,
|
| 697 |
+
"step": 9600
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 2.1455430214554303,
|
| 701 |
+
"grad_norm": 1789.1011962890625,
|
| 702 |
+
"learning_rate": 2.861111111111111e-05,
|
| 703 |
+
"loss": 68.5661,
|
| 704 |
+
"step": 9700
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 2.16766202167662,
|
| 708 |
+
"grad_norm": 1088.1776123046875,
|
| 709 |
+
"learning_rate": 2.8333333333333335e-05,
|
| 710 |
+
"loss": 67.3807,
|
| 711 |
+
"step": 9800
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 2.18978102189781,
|
| 715 |
+
"grad_norm": 511.173828125,
|
| 716 |
+
"learning_rate": 2.8055555555555557e-05,
|
| 717 |
+
"loss": 71.9351,
|
| 718 |
+
"step": 9900
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 2.2119000221190004,
|
| 722 |
+
"grad_norm": 372.8301696777344,
|
| 723 |
+
"learning_rate": 2.777777777777778e-05,
|
| 724 |
+
"loss": 116.7822,
|
| 725 |
+
"step": 10000
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 2.23401902234019,
|
| 729 |
+
"grad_norm": 673.3489379882812,
|
| 730 |
+
"learning_rate": 2.7500000000000004e-05,
|
| 731 |
+
"loss": 119.5753,
|
| 732 |
+
"step": 10100
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 2.2561380225613803,
|
| 736 |
+
"grad_norm": 188.46298217773438,
|
| 737 |
+
"learning_rate": 2.7222222222222223e-05,
|
| 738 |
+
"loss": 77.327,
|
| 739 |
+
"step": 10200
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 2.2782570227825705,
|
| 743 |
+
"grad_norm": 1344.5301513671875,
|
| 744 |
+
"learning_rate": 2.6944444444444445e-05,
|
| 745 |
+
"loss": 81.0025,
|
| 746 |
+
"step": 10300
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 2.30037602300376,
|
| 750 |
+
"grad_norm": 950.5435791015625,
|
| 751 |
+
"learning_rate": 2.6666666666666667e-05,
|
| 752 |
+
"loss": 77.1479,
|
| 753 |
+
"step": 10400
|
| 754 |
+
},
|
| 755 |
+
{
|
| 756 |
+
"epoch": 2.3224950232249504,
|
| 757 |
+
"grad_norm": 253.42047119140625,
|
| 758 |
+
"learning_rate": 2.6388888888888892e-05,
|
| 759 |
+
"loss": 78.325,
|
| 760 |
+
"step": 10500
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 2.34461402344614,
|
| 764 |
+
"grad_norm": 537.2485961914062,
|
| 765 |
+
"learning_rate": 2.6111111111111114e-05,
|
| 766 |
+
"loss": 114.1938,
|
| 767 |
+
"step": 10600
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 2.3667330236673303,
|
| 771 |
+
"grad_norm": 391.37054443359375,
|
| 772 |
+
"learning_rate": 2.5833333333333336e-05,
|
| 773 |
+
"loss": 74.7277,
|
| 774 |
+
"step": 10700
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 2.3888520238885205,
|
| 778 |
+
"grad_norm": 777.1848754882812,
|
| 779 |
+
"learning_rate": 2.5555555555555554e-05,
|
| 780 |
+
"loss": 87.6657,
|
| 781 |
+
"step": 10800
|
| 782 |
+
},
|
| 783 |
+
{
|
| 784 |
+
"epoch": 2.41097102410971,
|
| 785 |
+
"grad_norm": 233.57789611816406,
|
| 786 |
+
"learning_rate": 2.527777777777778e-05,
|
| 787 |
+
"loss": 74.0286,
|
| 788 |
+
"step": 10900
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 2.4330900243309004,
|
| 792 |
+
"grad_norm": 683.5059814453125,
|
| 793 |
+
"learning_rate": 2.5e-05,
|
| 794 |
+
"loss": 120.9458,
|
| 795 |
+
"step": 11000
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 2.45520902455209,
|
| 799 |
+
"grad_norm": 495.2901916503906,
|
| 800 |
+
"learning_rate": 2.4722222222222223e-05,
|
| 801 |
+
"loss": 86.7892,
|
| 802 |
+
"step": 11100
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 2.4773280247732803,
|
| 806 |
+
"grad_norm": 550.5516967773438,
|
| 807 |
+
"learning_rate": 2.4444444444444445e-05,
|
| 808 |
+
"loss": 89.1761,
|
| 809 |
+
"step": 11200
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 2.4994470249944705,
|
| 813 |
+
"grad_norm": 403.9693298339844,
|
| 814 |
+
"learning_rate": 2.4166666666666667e-05,
|
| 815 |
+
"loss": 68.0416,
|
| 816 |
+
"step": 11300
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 2.52156602521566,
|
| 820 |
+
"grad_norm": 156.51942443847656,
|
| 821 |
+
"learning_rate": 2.3888888888888892e-05,
|
| 822 |
+
"loss": 102.727,
|
| 823 |
+
"step": 11400
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 2.5436850254368504,
|
| 827 |
+
"grad_norm": 1048.75439453125,
|
| 828 |
+
"learning_rate": 2.361111111111111e-05,
|
| 829 |
+
"loss": 101.3001,
|
| 830 |
+
"step": 11500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 2.56580402565804,
|
| 834 |
+
"grad_norm": 1137.298828125,
|
| 835 |
+
"learning_rate": 2.3333333333333336e-05,
|
| 836 |
+
"loss": 80.9905,
|
| 837 |
+
"step": 11600
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 2.5879230258792303,
|
| 841 |
+
"grad_norm": 600.23681640625,
|
| 842 |
+
"learning_rate": 2.3055555555555558e-05,
|
| 843 |
+
"loss": 68.7155,
|
| 844 |
+
"step": 11700
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 2.6100420261004205,
|
| 848 |
+
"grad_norm": 840.565185546875,
|
| 849 |
+
"learning_rate": 2.277777777777778e-05,
|
| 850 |
+
"loss": 97.0005,
|
| 851 |
+
"step": 11800
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 2.63216102632161,
|
| 855 |
+
"grad_norm": 722.6634521484375,
|
| 856 |
+
"learning_rate": 2.25e-05,
|
| 857 |
+
"loss": 43.6906,
|
| 858 |
+
"step": 11900
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 2.6542800265428004,
|
| 862 |
+
"grad_norm": 399.0193176269531,
|
| 863 |
+
"learning_rate": 2.2222222222222223e-05,
|
| 864 |
+
"loss": 87.6618,
|
| 865 |
+
"step": 12000
|
| 866 |
+
},
|
| 867 |
+
{
|
| 868 |
+
"epoch": 2.67639902676399,
|
| 869 |
+
"grad_norm": 1078.0657958984375,
|
| 870 |
+
"learning_rate": 2.1944444444444445e-05,
|
| 871 |
+
"loss": 90.9304,
|
| 872 |
+
"step": 12100
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 2.6985180269851803,
|
| 876 |
+
"grad_norm": 719.8499145507812,
|
| 877 |
+
"learning_rate": 2.1666666666666667e-05,
|
| 878 |
+
"loss": 60.1098,
|
| 879 |
+
"step": 12200
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 2.7206370272063705,
|
| 883 |
+
"grad_norm": 2423.7578125,
|
| 884 |
+
"learning_rate": 2.138888888888889e-05,
|
| 885 |
+
"loss": 80.944,
|
| 886 |
+
"step": 12300
|
| 887 |
+
},
|
| 888 |
+
{
|
| 889 |
+
"epoch": 2.74275602742756,
|
| 890 |
+
"grad_norm": 1538.58447265625,
|
| 891 |
+
"learning_rate": 2.111111111111111e-05,
|
| 892 |
+
"loss": 84.306,
|
| 893 |
+
"step": 12400
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 2.7648750276487504,
|
| 897 |
+
"grad_norm": 1219.7099609375,
|
| 898 |
+
"learning_rate": 2.0833333333333336e-05,
|
| 899 |
+
"loss": 61.9064,
|
| 900 |
+
"step": 12500
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 2.78699402786994,
|
| 904 |
+
"grad_norm": 419.0851135253906,
|
| 905 |
+
"learning_rate": 2.0555555555555555e-05,
|
| 906 |
+
"loss": 58.863,
|
| 907 |
+
"step": 12600
|
| 908 |
+
},
|
| 909 |
+
{
|
| 910 |
+
"epoch": 2.8091130280911303,
|
| 911 |
+
"grad_norm": 10715.943359375,
|
| 912 |
+
"learning_rate": 2.027777777777778e-05,
|
| 913 |
+
"loss": 74.7694,
|
| 914 |
+
"step": 12700
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 2.8312320283123205,
|
| 918 |
+
"grad_norm": 201.3819122314453,
|
| 919 |
+
"learning_rate": 2e-05,
|
| 920 |
+
"loss": 83.1654,
|
| 921 |
+
"step": 12800
|
| 922 |
+
},
|
| 923 |
+
{
|
| 924 |
+
"epoch": 2.85335102853351,
|
| 925 |
+
"grad_norm": 3191.744873046875,
|
| 926 |
+
"learning_rate": 1.9722222222222224e-05,
|
| 927 |
+
"loss": 79.8109,
|
| 928 |
+
"step": 12900
|
| 929 |
+
},
|
| 930 |
+
{
|
| 931 |
+
"epoch": 2.8754700287547004,
|
| 932 |
+
"grad_norm": 284.8264465332031,
|
| 933 |
+
"learning_rate": 1.9444444444444445e-05,
|
| 934 |
+
"loss": 91.2385,
|
| 935 |
+
"step": 13000
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 2.89758902897589,
|
| 939 |
+
"grad_norm": 355.34674072265625,
|
| 940 |
+
"learning_rate": 1.9166666666666667e-05,
|
| 941 |
+
"loss": 53.8721,
|
| 942 |
+
"step": 13100
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 2.9197080291970803,
|
| 946 |
+
"grad_norm": 760.095703125,
|
| 947 |
+
"learning_rate": 1.888888888888889e-05,
|
| 948 |
+
"loss": 70.6495,
|
| 949 |
+
"step": 13200
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 2.9418270294182705,
|
| 953 |
+
"grad_norm": 163.60043334960938,
|
| 954 |
+
"learning_rate": 1.861111111111111e-05,
|
| 955 |
+
"loss": 61.0286,
|
| 956 |
+
"step": 13300
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 2.96394602963946,
|
| 960 |
+
"grad_norm": 694.377197265625,
|
| 961 |
+
"learning_rate": 1.8333333333333333e-05,
|
| 962 |
+
"loss": 70.8661,
|
| 963 |
+
"step": 13400
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 2.9860650298606504,
|
| 967 |
+
"grad_norm": 789.5768432617188,
|
| 968 |
+
"learning_rate": 1.8055555555555555e-05,
|
| 969 |
+
"loss": 51.5301,
|
| 970 |
+
"step": 13500
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 3.0,
|
| 974 |
+
"eval_loss": 166.22483825683594,
|
| 975 |
+
"eval_runtime": 20.711,
|
| 976 |
+
"eval_samples_per_second": 97.05,
|
| 977 |
+
"eval_steps_per_second": 24.287,
|
| 978 |
+
"step": 13563
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 3.00818403008184,
|
| 982 |
+
"grad_norm": 588.47265625,
|
| 983 |
+
"learning_rate": 1.777777777777778e-05,
|
| 984 |
+
"loss": 60.6116,
|
| 985 |
+
"step": 13600
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 3.0303030303030303,
|
| 989 |
+
"grad_norm": 637.5720825195312,
|
| 990 |
+
"learning_rate": 1.75e-05,
|
| 991 |
+
"loss": 75.5694,
|
| 992 |
+
"step": 13700
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 3.0524220305242205,
|
| 996 |
+
"grad_norm": 271.2422180175781,
|
| 997 |
+
"learning_rate": 1.7222222222222224e-05,
|
| 998 |
+
"loss": 67.7535,
|
| 999 |
+
"step": 13800
|
| 1000 |
+
},
|
| 1001 |
+
{
|
| 1002 |
+
"epoch": 3.07454103074541,
|
| 1003 |
+
"grad_norm": 1230.089111328125,
|
| 1004 |
+
"learning_rate": 1.6944444444444446e-05,
|
| 1005 |
+
"loss": 84.6008,
|
| 1006 |
+
"step": 13900
|
| 1007 |
+
},
|
| 1008 |
+
{
|
| 1009 |
+
"epoch": 3.0966600309666004,
|
| 1010 |
+
"grad_norm": 493.8188781738281,
|
| 1011 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1012 |
+
"loss": 57.6848,
|
| 1013 |
+
"step": 14000
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 3.11877903118779,
|
| 1017 |
+
"grad_norm": 927.6559448242188,
|
| 1018 |
+
"learning_rate": 1.638888888888889e-05,
|
| 1019 |
+
"loss": 77.8659,
|
| 1020 |
+
"step": 14100
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 3.1408980314089803,
|
| 1024 |
+
"grad_norm": 1547.560302734375,
|
| 1025 |
+
"learning_rate": 1.6111111111111115e-05,
|
| 1026 |
+
"loss": 56.6124,
|
| 1027 |
+
"step": 14200
|
| 1028 |
+
},
|
| 1029 |
+
{
|
| 1030 |
+
"epoch": 3.1630170316301705,
|
| 1031 |
+
"grad_norm": 1353.518798828125,
|
| 1032 |
+
"learning_rate": 1.5833333333333333e-05,
|
| 1033 |
+
"loss": 52.2098,
|
| 1034 |
+
"step": 14300
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 3.18513603185136,
|
| 1038 |
+
"grad_norm": 20034.76171875,
|
| 1039 |
+
"learning_rate": 1.5555555555555555e-05,
|
| 1040 |
+
"loss": 47.6291,
|
| 1041 |
+
"step": 14400
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 3.2072550320725504,
|
| 1045 |
+
"grad_norm": 295.83868408203125,
|
| 1046 |
+
"learning_rate": 1.527777777777778e-05,
|
| 1047 |
+
"loss": 76.8827,
|
| 1048 |
+
"step": 14500
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 3.22937403229374,
|
| 1052 |
+
"grad_norm": 1203.1685791015625,
|
| 1053 |
+
"learning_rate": 1.5e-05,
|
| 1054 |
+
"loss": 75.1167,
|
| 1055 |
+
"step": 14600
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 3.2514930325149303,
|
| 1059 |
+
"grad_norm": 514.1155395507812,
|
| 1060 |
+
"learning_rate": 1.4722222222222224e-05,
|
| 1061 |
+
"loss": 56.7415,
|
| 1062 |
+
"step": 14700
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 3.2736120327361204,
|
| 1066 |
+
"grad_norm": 254.90188598632812,
|
| 1067 |
+
"learning_rate": 1.4444444444444444e-05,
|
| 1068 |
+
"loss": 65.6232,
|
| 1069 |
+
"step": 14800
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 3.29573103295731,
|
| 1073 |
+
"grad_norm": 2475.375,
|
| 1074 |
+
"learning_rate": 1.4166666666666668e-05,
|
| 1075 |
+
"loss": 82.0212,
|
| 1076 |
+
"step": 14900
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 3.3178500331785004,
|
| 1080 |
+
"grad_norm": 695.0991821289062,
|
| 1081 |
+
"learning_rate": 1.388888888888889e-05,
|
| 1082 |
+
"loss": 70.9937,
|
| 1083 |
+
"step": 15000
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 3.33996903339969,
|
| 1087 |
+
"grad_norm": 3679.786376953125,
|
| 1088 |
+
"learning_rate": 1.3611111111111111e-05,
|
| 1089 |
+
"loss": 87.9886,
|
| 1090 |
+
"step": 15100
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 3.3620880336208803,
|
| 1094 |
+
"grad_norm": 335.57244873046875,
|
| 1095 |
+
"learning_rate": 1.3333333333333333e-05,
|
| 1096 |
+
"loss": 81.2132,
|
| 1097 |
+
"step": 15200
|
| 1098 |
+
},
|
| 1099 |
+
{
|
| 1100 |
+
"epoch": 3.3842070338420704,
|
| 1101 |
+
"grad_norm": 3056.2080078125,
|
| 1102 |
+
"learning_rate": 1.3055555555555557e-05,
|
| 1103 |
+
"loss": 84.2817,
|
| 1104 |
+
"step": 15300
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 3.40632603406326,
|
| 1108 |
+
"grad_norm": 633.110107421875,
|
| 1109 |
+
"learning_rate": 1.2777777777777777e-05,
|
| 1110 |
+
"loss": 91.3568,
|
| 1111 |
+
"step": 15400
|
| 1112 |
+
},
|
| 1113 |
+
{
|
| 1114 |
+
"epoch": 3.4284450342844504,
|
| 1115 |
+
"grad_norm": 1400.1513671875,
|
| 1116 |
+
"learning_rate": 1.25e-05,
|
| 1117 |
+
"loss": 64.4974,
|
| 1118 |
+
"step": 15500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 3.4505640345056405,
|
| 1122 |
+
"grad_norm": 366.1932678222656,
|
| 1123 |
+
"learning_rate": 1.2222222222222222e-05,
|
| 1124 |
+
"loss": 65.3457,
|
| 1125 |
+
"step": 15600
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 3.4726830347268303,
|
| 1129 |
+
"grad_norm": 77.0812759399414,
|
| 1130 |
+
"learning_rate": 1.1944444444444446e-05,
|
| 1131 |
+
"loss": 94.2989,
|
| 1132 |
+
"step": 15700
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 3.4948020349480204,
|
| 1136 |
+
"grad_norm": 517.035888671875,
|
| 1137 |
+
"learning_rate": 1.1666666666666668e-05,
|
| 1138 |
+
"loss": 54.1213,
|
| 1139 |
+
"step": 15800
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 3.5169210351692106,
|
| 1143 |
+
"grad_norm": 582.3343505859375,
|
| 1144 |
+
"learning_rate": 1.138888888888889e-05,
|
| 1145 |
+
"loss": 80.1348,
|
| 1146 |
+
"step": 15900
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 3.5390400353904004,
|
| 1150 |
+
"grad_norm": 134.23880004882812,
|
| 1151 |
+
"learning_rate": 1.1111111111111112e-05,
|
| 1152 |
+
"loss": 76.6385,
|
| 1153 |
+
"step": 16000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 3.56115903561159,
|
| 1157 |
+
"grad_norm": 341.63604736328125,
|
| 1158 |
+
"learning_rate": 1.0833333333333334e-05,
|
| 1159 |
+
"loss": 78.0171,
|
| 1160 |
+
"step": 16100
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 3.5832780358327803,
|
| 1164 |
+
"grad_norm": 1244.135009765625,
|
| 1165 |
+
"learning_rate": 1.0555555555555555e-05,
|
| 1166 |
+
"loss": 51.3064,
|
| 1167 |
+
"step": 16200
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 3.6053970360539704,
|
| 1171 |
+
"grad_norm": 916.8475952148438,
|
| 1172 |
+
"learning_rate": 1.0277777777777777e-05,
|
| 1173 |
+
"loss": 79.2429,
|
| 1174 |
+
"step": 16300
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 3.6275160362751606,
|
| 1178 |
+
"grad_norm": 269.9622802734375,
|
| 1179 |
+
"learning_rate": 1e-05,
|
| 1180 |
+
"loss": 68.2658,
|
| 1181 |
+
"step": 16400
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 3.6496350364963503,
|
| 1185 |
+
"grad_norm": 272.8192138671875,
|
| 1186 |
+
"learning_rate": 9.722222222222223e-06,
|
| 1187 |
+
"loss": 54.9957,
|
| 1188 |
+
"step": 16500
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 3.6717540367175405,
|
| 1192 |
+
"grad_norm": 529.631591796875,
|
| 1193 |
+
"learning_rate": 9.444444444444445e-06,
|
| 1194 |
+
"loss": 55.4425,
|
| 1195 |
+
"step": 16600
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 3.6938730369387303,
|
| 1199 |
+
"grad_norm": 478.92327880859375,
|
| 1200 |
+
"learning_rate": 9.166666666666666e-06,
|
| 1201 |
+
"loss": 49.7692,
|
| 1202 |
+
"step": 16700
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 3.7159920371599204,
|
| 1206 |
+
"grad_norm": 499.35406494140625,
|
| 1207 |
+
"learning_rate": 8.88888888888889e-06,
|
| 1208 |
+
"loss": 46.7543,
|
| 1209 |
+
"step": 16800
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 3.7381110373811106,
|
| 1213 |
+
"grad_norm": 179.26344299316406,
|
| 1214 |
+
"learning_rate": 8.611111111111112e-06,
|
| 1215 |
+
"loss": 64.0389,
|
| 1216 |
+
"step": 16900
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 3.7602300376023003,
|
| 1220 |
+
"grad_norm": 1287.7215576171875,
|
| 1221 |
+
"learning_rate": 8.333333333333334e-06,
|
| 1222 |
+
"loss": 59.5381,
|
| 1223 |
+
"step": 17000
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 3.7823490378234905,
|
| 1227 |
+
"grad_norm": 916.1105346679688,
|
| 1228 |
+
"learning_rate": 8.055555555555557e-06,
|
| 1229 |
+
"loss": 60.6414,
|
| 1230 |
+
"step": 17100
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 3.8044680380446803,
|
| 1234 |
+
"grad_norm": 398.42724609375,
|
| 1235 |
+
"learning_rate": 7.777777777777777e-06,
|
| 1236 |
+
"loss": 56.4996,
|
| 1237 |
+
"step": 17200
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 3.8265870382658704,
|
| 1241 |
+
"grad_norm": 466.2370910644531,
|
| 1242 |
+
"learning_rate": 7.5e-06,
|
| 1243 |
+
"loss": 55.1325,
|
| 1244 |
+
"step": 17300
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 3.8487060384870606,
|
| 1248 |
+
"grad_norm": 426.53271484375,
|
| 1249 |
+
"learning_rate": 7.222222222222222e-06,
|
| 1250 |
+
"loss": 74.4013,
|
| 1251 |
+
"step": 17400
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 3.8708250387082503,
|
| 1255 |
+
"grad_norm": 595.925537109375,
|
| 1256 |
+
"learning_rate": 6.944444444444445e-06,
|
| 1257 |
+
"loss": 65.174,
|
| 1258 |
+
"step": 17500
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 3.8929440389294405,
|
| 1262 |
+
"grad_norm": 960.3365478515625,
|
| 1263 |
+
"learning_rate": 6.666666666666667e-06,
|
| 1264 |
+
"loss": 61.204,
|
| 1265 |
+
"step": 17600
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 3.9150630391506303,
|
| 1269 |
+
"grad_norm": 488.77978515625,
|
| 1270 |
+
"learning_rate": 6.3888888888888885e-06,
|
| 1271 |
+
"loss": 64.5583,
|
| 1272 |
+
"step": 17700
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 3.9371820393718204,
|
| 1276 |
+
"grad_norm": 386.27911376953125,
|
| 1277 |
+
"learning_rate": 6.111111111111111e-06,
|
| 1278 |
+
"loss": 53.6574,
|
| 1279 |
+
"step": 17800
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 3.9593010395930106,
|
| 1283 |
+
"grad_norm": 741.3096313476562,
|
| 1284 |
+
"learning_rate": 5.833333333333334e-06,
|
| 1285 |
+
"loss": 56.0408,
|
| 1286 |
+
"step": 17900
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 3.9814200398142003,
|
| 1290 |
+
"grad_norm": 125.60317993164062,
|
| 1291 |
+
"learning_rate": 5.555555555555556e-06,
|
| 1292 |
+
"loss": 51.7872,
|
| 1293 |
+
"step": 18000
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 4.0,
|
| 1297 |
+
"eval_loss": 174.72474670410156,
|
| 1298 |
+
"eval_runtime": 20.9561,
|
| 1299 |
+
"eval_samples_per_second": 95.915,
|
| 1300 |
+
"eval_steps_per_second": 24.003,
|
| 1301 |
+
"step": 18084
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 4.00353904003539,
|
| 1305 |
+
"grad_norm": 206.36895751953125,
|
| 1306 |
+
"learning_rate": 5.277777777777778e-06,
|
| 1307 |
+
"loss": 40.9699,
|
| 1308 |
+
"step": 18100
|
| 1309 |
+
},
|
| 1310 |
+
{
|
| 1311 |
+
"epoch": 4.02565804025658,
|
| 1312 |
+
"grad_norm": 399.98455810546875,
|
| 1313 |
+
"learning_rate": 5e-06,
|
| 1314 |
+
"loss": 56.6913,
|
| 1315 |
+
"step": 18200
|
| 1316 |
+
},
|
| 1317 |
+
{
|
| 1318 |
+
"epoch": 4.04777704047777,
|
| 1319 |
+
"grad_norm": 1047.0919189453125,
|
| 1320 |
+
"learning_rate": 4.722222222222222e-06,
|
| 1321 |
+
"loss": 48.5654,
|
| 1322 |
+
"step": 18300
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 4.069896040698961,
|
| 1326 |
+
"grad_norm": 882.0901489257812,
|
| 1327 |
+
"learning_rate": 4.444444444444445e-06,
|
| 1328 |
+
"loss": 47.7457,
|
| 1329 |
+
"step": 18400
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 4.092015040920151,
|
| 1333 |
+
"grad_norm": 515.3323364257812,
|
| 1334 |
+
"learning_rate": 4.166666666666667e-06,
|
| 1335 |
+
"loss": 66.938,
|
| 1336 |
+
"step": 18500
|
| 1337 |
+
},
|
| 1338 |
+
{
|
| 1339 |
+
"epoch": 4.11413404114134,
|
| 1340 |
+
"grad_norm": 1115.885009765625,
|
| 1341 |
+
"learning_rate": 3.888888888888889e-06,
|
| 1342 |
+
"loss": 58.0705,
|
| 1343 |
+
"step": 18600
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 4.13625304136253,
|
| 1347 |
+
"grad_norm": 672.346435546875,
|
| 1348 |
+
"learning_rate": 3.611111111111111e-06,
|
| 1349 |
+
"loss": 68.3134,
|
| 1350 |
+
"step": 18700
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 4.15837204158372,
|
| 1354 |
+
"grad_norm": 592.4734497070312,
|
| 1355 |
+
"learning_rate": 3.3333333333333333e-06,
|
| 1356 |
+
"loss": 55.2237,
|
| 1357 |
+
"step": 18800
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 4.180491041804911,
|
| 1361 |
+
"grad_norm": 642.23583984375,
|
| 1362 |
+
"learning_rate": 3.0555555555555556e-06,
|
| 1363 |
+
"loss": 77.2789,
|
| 1364 |
+
"step": 18900
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 4.202610042026101,
|
| 1368 |
+
"grad_norm": 595.8189086914062,
|
| 1369 |
+
"learning_rate": 2.777777777777778e-06,
|
| 1370 |
+
"loss": 54.4822,
|
| 1371 |
+
"step": 19000
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 4.22472904224729,
|
| 1375 |
+
"grad_norm": 440.1129150390625,
|
| 1376 |
+
"learning_rate": 2.5e-06,
|
| 1377 |
+
"loss": 50.1236,
|
| 1378 |
+
"step": 19100
|
| 1379 |
+
},
|
| 1380 |
+
{
|
| 1381 |
+
"epoch": 4.24684804246848,
|
| 1382 |
+
"grad_norm": 8661.6494140625,
|
| 1383 |
+
"learning_rate": 2.2222222222222225e-06,
|
| 1384 |
+
"loss": 81.7971,
|
| 1385 |
+
"step": 19200
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 4.26896704268967,
|
| 1389 |
+
"grad_norm": 214.70748901367188,
|
| 1390 |
+
"learning_rate": 1.9444444444444444e-06,
|
| 1391 |
+
"loss": 51.6579,
|
| 1392 |
+
"step": 19300
|
| 1393 |
+
},
|
| 1394 |
+
{
|
| 1395 |
+
"epoch": 4.291086042910861,
|
| 1396 |
+
"grad_norm": 724.3253173828125,
|
| 1397 |
+
"learning_rate": 1.6666666666666667e-06,
|
| 1398 |
+
"loss": 68.406,
|
| 1399 |
+
"step": 19400
|
| 1400 |
+
},
|
| 1401 |
+
{
|
| 1402 |
+
"epoch": 4.313205043132051,
|
| 1403 |
+
"grad_norm": 1145.943115234375,
|
| 1404 |
+
"learning_rate": 1.388888888888889e-06,
|
| 1405 |
+
"loss": 62.2609,
|
| 1406 |
+
"step": 19500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 4.33532404335324,
|
| 1410 |
+
"grad_norm": 1727.2354736328125,
|
| 1411 |
+
"learning_rate": 1.1111111111111112e-06,
|
| 1412 |
+
"loss": 67.2163,
|
| 1413 |
+
"step": 19600
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 4.35744304357443,
|
| 1417 |
+
"grad_norm": 734.8032836914062,
|
| 1418 |
+
"learning_rate": 8.333333333333333e-07,
|
| 1419 |
+
"loss": 57.1625,
|
| 1420 |
+
"step": 19700
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 4.37956204379562,
|
| 1424 |
+
"grad_norm": 1056.7657470703125,
|
| 1425 |
+
"learning_rate": 5.555555555555556e-07,
|
| 1426 |
+
"loss": 61.3986,
|
| 1427 |
+
"step": 19800
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 4.401681044016811,
|
| 1431 |
+
"grad_norm": 550.4881591796875,
|
| 1432 |
+
"learning_rate": 2.777777777777778e-07,
|
| 1433 |
+
"loss": 60.5436,
|
| 1434 |
+
"step": 19900
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 4.423800044238001,
|
| 1438 |
+
"grad_norm": 435.5983581542969,
|
| 1439 |
+
"learning_rate": 0.0,
|
| 1440 |
+
"loss": 60.0424,
|
| 1441 |
+
"step": 20000
|
| 1442 |
+
}
|
| 1443 |
+
],
|
| 1444 |
+
"logging_steps": 100,
|
| 1445 |
+
"max_steps": 20000,
|
| 1446 |
+
"num_input_tokens_seen": 0,
|
| 1447 |
+
"num_train_epochs": 5,
|
| 1448 |
+
"save_steps": 10000,
|
| 1449 |
+
"stateful_callbacks": {
|
| 1450 |
+
"TrainerControl": {
|
| 1451 |
+
"args": {
|
| 1452 |
+
"should_epoch_stop": false,
|
| 1453 |
+
"should_evaluate": false,
|
| 1454 |
+
"should_log": false,
|
| 1455 |
+
"should_save": true,
|
| 1456 |
+
"should_training_stop": true
|
| 1457 |
+
},
|
| 1458 |
+
"attributes": {}
|
| 1459 |
+
}
|
| 1460 |
+
},
|
| 1461 |
+
"total_flos": 0.0,
|
| 1462 |
+
"train_batch_size": 4,
|
| 1463 |
+
"trial_name": null,
|
| 1464 |
+
"trial_params": null
|
| 1465 |
+
}
|