| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "eval_steps": 500, | |
| "global_step": 66, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.4678362573099415, | |
| "grad_norm": 2.0982398986816406, | |
| "learning_rate": 9e-05, | |
| "loss": 1.4244, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.935672514619883, | |
| "grad_norm": 2.533539056777954, | |
| "learning_rate": 0.00019, | |
| "loss": 1.337, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 1.2067078351974487, | |
| "eval_runtime": 0.8256, | |
| "eval_samples_per_second": 23.013, | |
| "eval_steps_per_second": 3.634, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 1.3742690058479532, | |
| "grad_norm": 2.3733065128326416, | |
| "learning_rate": 0.00016086956521739132, | |
| "loss": 1.1269, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 1.8421052631578947, | |
| "grad_norm": 1.8323912620544434, | |
| "learning_rate": 0.0001173913043478261, | |
| "loss": 0.9533, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 1.1962652206420898, | |
| "eval_runtime": 0.8259, | |
| "eval_samples_per_second": 23.005, | |
| "eval_steps_per_second": 3.632, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 2.280701754385965, | |
| "grad_norm": 2.286313533782959, | |
| "learning_rate": 7.391304347826086e-05, | |
| "loss": 0.8912, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 2.7485380116959064, | |
| "grad_norm": 2.591411590576172, | |
| "learning_rate": 3.0434782608695656e-05, | |
| "loss": 0.7545, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 1.2603905200958252, | |
| "eval_runtime": 0.8232, | |
| "eval_samples_per_second": 23.079, | |
| "eval_steps_per_second": 3.644, | |
| "step": 66 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 66, | |
| "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": 1.1248909934395392e+16, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |