| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 7.998864926220204, | |
| "global_step": 1760, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.4856970986003244, | |
| "eval_loss": 3.856384515762329, | |
| "eval_runtime": 5.67, | |
| "eval_samples_per_second": 16.049, | |
| "eval_steps_per_second": 8.113, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.6095839644226742, | |
| "eval_loss": 2.7515063285827637, | |
| "eval_runtime": 5.6811, | |
| "eval_samples_per_second": 16.018, | |
| "eval_steps_per_second": 8.097, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 7.181818181818182e-06, | |
| "loss": 4.1568, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.6779671940962263, | |
| "eval_loss": 2.2462902069091797, | |
| "eval_runtime": 5.6846, | |
| "eval_samples_per_second": 16.008, | |
| "eval_steps_per_second": 8.092, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.7152310055535862, | |
| "eval_loss": 1.981738567352295, | |
| "eval_runtime": 5.6899, | |
| "eval_samples_per_second": 15.993, | |
| "eval_steps_per_second": 8.084, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.340909090909091e-06, | |
| "loss": 2.2818, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_accuracy": 0.7353184449958643, | |
| "eval_loss": 1.8277500867843628, | |
| "eval_runtime": 5.6901, | |
| "eval_samples_per_second": 15.993, | |
| "eval_steps_per_second": 8.084, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_accuracy": 0.7485740066385228, | |
| "eval_loss": 1.7312932014465332, | |
| "eval_runtime": 5.683, | |
| "eval_samples_per_second": 16.013, | |
| "eval_steps_per_second": 8.094, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 1.5e-06, | |
| "loss": 1.8444, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_accuracy": 0.7553092069221101, | |
| "eval_loss": 1.6847076416015625, | |
| "eval_runtime": 5.6791, | |
| "eval_samples_per_second": 16.024, | |
| "eval_steps_per_second": 8.1, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_accuracy": 0.75708162804937, | |
| "eval_loss": 1.669904112815857, | |
| "eval_runtime": 5.9939, | |
| "eval_samples_per_second": 15.182, | |
| "eval_steps_per_second": 7.674, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "step": 1760, | |
| "total_flos": 6.134202859152998e+16, | |
| "train_loss": 2.6170630194924094, | |
| "train_runtime": 2594.3652, | |
| "train_samples_per_second": 5.433, | |
| "train_steps_per_second": 0.678 | |
| } | |
| ], | |
| "max_steps": 1760, | |
| "num_train_epochs": 8, | |
| "total_flos": 6.134202859152998e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |