| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.052534804307853955, | |
| "eval_steps": 13, | |
| "global_step": 50, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010506960861570791, | |
| "eval_loss": 0.3191992938518524, | |
| "eval_runtime": 25.634, | |
| "eval_samples_per_second": 15.643, | |
| "eval_steps_per_second": 7.841, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.010506960861570791, | |
| "grad_norm": 0.15685106813907623, | |
| "learning_rate": 0.00019396926207859084, | |
| "loss": 0.2342, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013659049120042028, | |
| "eval_loss": 0.14280569553375244, | |
| "eval_runtime": 24.0643, | |
| "eval_samples_per_second": 16.664, | |
| "eval_steps_per_second": 8.353, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.021013921723141583, | |
| "grad_norm": 0.12126398086547852, | |
| "learning_rate": 0.00015000000000000001, | |
| "loss": 0.1313, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.027318098240084056, | |
| "eval_loss": 0.10690547525882721, | |
| "eval_runtime": 24.0709, | |
| "eval_samples_per_second": 16.659, | |
| "eval_steps_per_second": 8.35, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03152088258471237, | |
| "grad_norm": 0.1156439557671547, | |
| "learning_rate": 8.263518223330697e-05, | |
| "loss": 0.115, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.04097714736012608, | |
| "eval_loss": 0.10069109499454498, | |
| "eval_runtime": 24.1029, | |
| "eval_samples_per_second": 16.637, | |
| "eval_steps_per_second": 8.339, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.042027843446283165, | |
| "grad_norm": 0.0701930969953537, | |
| "learning_rate": 2.339555568810221e-05, | |
| "loss": 0.1017, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.052534804307853955, | |
| "grad_norm": 0.10752475261688232, | |
| "learning_rate": 0.0, | |
| "loss": 0.0807, | |
| "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": 1.931385312903168e+16, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |