| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.0032508167677128877, | |
| "eval_steps": 13, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 6.501633535425775e-05, | |
| "eval_loss": 2.526299476623535, | |
| "eval_runtime": 238.6737, | |
| "eval_samples_per_second": 27.137, | |
| "eval_steps_per_second": 13.571, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0006501633535425775, | |
| "grad_norm": 3.365540027618408, | |
| "learning_rate": 0.00019396926207859084, | |
| "loss": 9.4926, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0008452123596053508, | |
| "eval_loss": 2.0112905502319336, | |
| "eval_runtime": 239.0698, | |
| "eval_samples_per_second": 27.093, | |
| "eval_steps_per_second": 13.548, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.001300326707085155, | |
| "grad_norm": 4.012279033660889, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 8.0936, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0016904247192107017, | |
| "eval_loss": 1.9174778461456299, | |
| "eval_runtime": 239.2067, | |
| "eval_samples_per_second": 27.077, | |
| "eval_steps_per_second": 13.541, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0019504900606277327, | |
| "grad_norm": 3.9299874305725098, | |
| "learning_rate": 8.263518223330697e-05, | |
| "loss": 7.7523, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0025356370788160525, | |
| "eval_loss": 1.8927605152130127, | |
| "eval_runtime": 238.4964, | |
| "eval_samples_per_second": 27.158, | |
| "eval_steps_per_second": 13.581, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.00260065341417031, | |
| "grad_norm": 3.0746917724609375, | |
| "learning_rate": 2.339555568810221e-05, | |
| "loss": 7.6451, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.0032508167677128877, | |
| "grad_norm": 3.5105154514312744, | |
| "learning_rate": 0.0, | |
| "loss": 7.5913, | |
| "step": 50 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 50, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 13, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 5783106279702528.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |