| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 207, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.2898550724637681, | |
| "grad_norm": 0.7885370850563049, | |
| "learning_rate": 9.523809523809524e-05, | |
| "loss": 1.6677, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.5797101449275363, | |
| "grad_norm": 0.2611646056175232, | |
| "learning_rate": 9.744735364682346e-05, | |
| "loss": 0.9644, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.8695652173913043, | |
| "grad_norm": 0.3123443126678467, | |
| "learning_rate": 8.953878684688493e-05, | |
| "loss": 0.9091, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 1.1594202898550725, | |
| "grad_norm": 0.3900753855705261, | |
| "learning_rate": 7.716108602402094e-05, | |
| "loss": 0.8599, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 1.4492753623188406, | |
| "grad_norm": 0.3543001115322113, | |
| "learning_rate": 6.171332303360411e-05, | |
| "loss": 0.8032, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 1.7391304347826086, | |
| "grad_norm": 0.42595645785331726, | |
| "learning_rate": 4.49415839006284e-05, | |
| "loss": 0.8319, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 2.028985507246377, | |
| "grad_norm": 0.41535866260528564, | |
| "learning_rate": 2.874160584821798e-05, | |
| "loss": 0.8275, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 2.318840579710145, | |
| "grad_norm": 0.45032790303230286, | |
| "learning_rate": 1.4944499005397371e-05, | |
| "loss": 0.7486, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 2.608695652173913, | |
| "grad_norm": 0.4935797154903412, | |
| "learning_rate": 5.109773021462921e-06, | |
| "loss": 0.7324, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 2.898550724637681, | |
| "grad_norm": 0.4422053396701813, | |
| "learning_rate": 3.490631287033197e-07, | |
| "loss": 0.7303, | |
| "step": 200 | |
| } | |
| ], | |
| "logging_steps": 20, | |
| "max_steps": 207, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 9.449028359530906e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |