| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "eval_steps": 500, | |
| "global_step": 7860, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.15867776389314597, | |
| "eval_loss": 6.794844150543213, | |
| "eval_runtime": 6.6858, | |
| "eval_samples_per_second": 24.829, | |
| "eval_steps_per_second": 0.897, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 1.272264631043257, | |
| "grad_norm": 11287.4306640625, | |
| "learning_rate": 0.0005993999999999999, | |
| "loss": 1.6643, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.173225190389739, | |
| "eval_loss": 6.429712295532227, | |
| "eval_runtime": 5.2622, | |
| "eval_samples_per_second": 31.546, | |
| "eval_steps_per_second": 1.14, | |
| "step": 1572 | |
| }, | |
| { | |
| "epoch": 2.544529262086514, | |
| "grad_norm": 9641.5859375, | |
| "learning_rate": 0.0005126239067055393, | |
| "loss": 1.466, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.18816164855115178, | |
| "eval_loss": 6.129866123199463, | |
| "eval_runtime": 5.4576, | |
| "eval_samples_per_second": 30.416, | |
| "eval_steps_per_second": 1.099, | |
| "step": 2358 | |
| }, | |
| { | |
| "epoch": 3.816793893129771, | |
| "grad_norm": 9910.599609375, | |
| "learning_rate": 0.00042516034985422735, | |
| "loss": 1.3707, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.1985358262796784, | |
| "eval_loss": 5.971755027770996, | |
| "eval_runtime": 5.5163, | |
| "eval_samples_per_second": 30.093, | |
| "eval_steps_per_second": 1.088, | |
| "step": 3144 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.2067880131091882, | |
| "eval_loss": 5.8568878173828125, | |
| "eval_runtime": 5.2842, | |
| "eval_samples_per_second": 31.414, | |
| "eval_steps_per_second": 1.135, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 5.089058524173028, | |
| "grad_norm": 10377.1337890625, | |
| "learning_rate": 0.00033769679300291546, | |
| "loss": 1.308, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.21243486666823852, | |
| "eval_loss": 5.761693000793457, | |
| "eval_runtime": 5.5936, | |
| "eval_samples_per_second": 29.677, | |
| "eval_steps_per_second": 1.073, | |
| "step": 4716 | |
| }, | |
| { | |
| "epoch": 6.361323155216285, | |
| "grad_norm": 11166.55078125, | |
| "learning_rate": 0.00025023323615160347, | |
| "loss": 1.2627, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.2184118077004692, | |
| "eval_loss": 5.693126201629639, | |
| "eval_runtime": 5.4671, | |
| "eval_samples_per_second": 30.363, | |
| "eval_steps_per_second": 1.097, | |
| "step": 5502 | |
| }, | |
| { | |
| "epoch": 7.633587786259542, | |
| "grad_norm": 11332.716796875, | |
| "learning_rate": 0.00016276967930029153, | |
| "loss": 1.2279, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.2233277532831915, | |
| "eval_loss": 5.635900974273682, | |
| "eval_runtime": 5.6909, | |
| "eval_samples_per_second": 29.169, | |
| "eval_steps_per_second": 1.054, | |
| "step": 6288 | |
| }, | |
| { | |
| "epoch": 8.9058524173028, | |
| "grad_norm": 11919.126953125, | |
| "learning_rate": 7.530612244897959e-05, | |
| "loss": 1.1974, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_accuracy": 0.22648716195506094, | |
| "eval_loss": 5.598357677459717, | |
| "eval_runtime": 5.3666, | |
| "eval_samples_per_second": 30.932, | |
| "eval_steps_per_second": 1.118, | |
| "step": 7074 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "eval_accuracy": 0.2284676867941433, | |
| "eval_loss": 5.574704170227051, | |
| "eval_runtime": 5.4915, | |
| "eval_samples_per_second": 30.229, | |
| "eval_steps_per_second": 1.093, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 7860, | |
| "total_flos": 6.569404268544e+16, | |
| "train_loss": 1.3368007669618718, | |
| "train_runtime": 10231.0545, | |
| "train_samples_per_second": 24.574, | |
| "train_steps_per_second": 0.768 | |
| } | |
| ], | |
| "logging_steps": 1000, | |
| "max_steps": 7860, | |
| "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": 6.569404268544e+16, | |
| "train_batch_size": 32, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |