| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 2850, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_bleu": 0.07566256044426542, | |
| "eval_loss": 4.71478796005249, | |
| "eval_runtime": 94.3673, | |
| "eval_samples_per_second": 12.049, | |
| "eval_steps_per_second": 0.763, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 1.7543859649122808, | |
| "grad_norm": 0.7608898878097534, | |
| "learning_rate": 1.9840000000000003e-05, | |
| "loss": 5.6113, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_bleu": 2.043915291200077e-09, | |
| "eval_loss": 1.6729369163513184, | |
| "eval_runtime": 334.9077, | |
| "eval_samples_per_second": 3.395, | |
| "eval_steps_per_second": 0.215, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_bleu": 0.013493219939051442, | |
| "eval_loss": 1.6423040628433228, | |
| "eval_runtime": 326.9519, | |
| "eval_samples_per_second": 3.478, | |
| "eval_steps_per_second": 0.22, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 3.5087719298245617, | |
| "grad_norm": 0.5546914935112, | |
| "learning_rate": 1.5778723404255322e-05, | |
| "loss": 1.6436, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_bleu": 26.712968804132256, | |
| "eval_loss": 1.6389009952545166, | |
| "eval_runtime": 129.4881, | |
| "eval_samples_per_second": 8.781, | |
| "eval_steps_per_second": 0.556, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_bleu": 27.593238887604137, | |
| "eval_loss": 1.6381840705871582, | |
| "eval_runtime": 92.3845, | |
| "eval_samples_per_second": 12.307, | |
| "eval_steps_per_second": 0.779, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 5.2631578947368425, | |
| "grad_norm": 0.4653625190258026, | |
| "learning_rate": 1.152340425531915e-05, | |
| "loss": 1.5933, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_bleu": 27.85219103460007, | |
| "eval_loss": 1.639172077178955, | |
| "eval_runtime": 94.3628, | |
| "eval_samples_per_second": 12.049, | |
| "eval_steps_per_second": 0.763, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_bleu": 27.69478534477315, | |
| "eval_loss": 1.6408443450927734, | |
| "eval_runtime": 97.4176, | |
| "eval_samples_per_second": 11.671, | |
| "eval_steps_per_second": 0.739, | |
| "step": 1995 | |
| }, | |
| { | |
| "epoch": 7.017543859649122, | |
| "grad_norm": 0.38791486620903015, | |
| "learning_rate": 7.268085106382979e-06, | |
| "loss": 1.5729, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_bleu": 28.056102822986603, | |
| "eval_loss": 1.6421233415603638, | |
| "eval_runtime": 88.6352, | |
| "eval_samples_per_second": 12.828, | |
| "eval_steps_per_second": 0.812, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 8.771929824561404, | |
| "grad_norm": 0.3788717985153198, | |
| "learning_rate": 3.012765957446809e-06, | |
| "loss": 1.5597, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_bleu": 27.829370325397207, | |
| "eval_loss": 1.6420503854751587, | |
| "eval_runtime": 88.6233, | |
| "eval_samples_per_second": 12.83, | |
| "eval_steps_per_second": 0.812, | |
| "step": 2565 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_bleu": 28.125503003363978, | |
| "eval_loss": 1.6436502933502197, | |
| "eval_runtime": 91.7782, | |
| "eval_samples_per_second": 12.389, | |
| "eval_steps_per_second": 0.785, | |
| "step": 2850 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 2850, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 10, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 2.463997932601344e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |