| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 4.685408299866131, | |
| "global_step": 3500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.7322623828647926e-05, | |
| "loss": 0.0412, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.993359375, | |
| "eval_loss": 0.03145532310009003, | |
| "eval_runtime": 22.6963, | |
| "eval_samples_per_second": 225.587, | |
| "eval_steps_per_second": 14.099, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.4645247657295851e-05, | |
| "loss": 0.0047, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.998828125, | |
| "eval_loss": 0.007673000451177359, | |
| "eval_runtime": 22.4025, | |
| "eval_samples_per_second": 228.546, | |
| "eval_steps_per_second": 14.284, | |
| "step": 1494 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.1967871485943776e-05, | |
| "loss": 0.0057, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 9.2904953145917e-06, | |
| "loss": 0.0015, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.998046875, | |
| "eval_loss": 0.010105514898896217, | |
| "eval_runtime": 22.3654, | |
| "eval_samples_per_second": 228.925, | |
| "eval_steps_per_second": 14.308, | |
| "step": 2241 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 6.6131191432396255e-06, | |
| "loss": 0.0001, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_accuracy": 0.9978515625, | |
| "eval_loss": 0.012340093031525612, | |
| "eval_runtime": 22.3303, | |
| "eval_samples_per_second": 229.285, | |
| "eval_steps_per_second": 14.33, | |
| "step": 2988 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 3.93574297188755e-06, | |
| "loss": 0.0, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 1.2583668005354755e-06, | |
| "loss": 0.0, | |
| "step": 3500 | |
| } | |
| ], | |
| "max_steps": 3735, | |
| "num_train_epochs": 5, | |
| "total_flos": 746705064001440.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |