| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 9.0, | |
| "global_step": 2475, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3e-05, | |
| "loss": 1.8194, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_exact_match": 53.317010609770314, | |
| "eval_f1": 61.60892163355557, | |
| "eval_runtime": 3.5138, | |
| "eval_samples_per_second": 2440.967, | |
| "eval_steps_per_second": 7.684, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 3e-05, | |
| "loss": 1.0001, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_exact_match": 53.585169639734175, | |
| "eval_f1": 61.93356641465034, | |
| "eval_runtime": 2.9917, | |
| "eval_samples_per_second": 2866.944, | |
| "eval_steps_per_second": 9.025, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.7994, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_exact_match": 54.249737670514165, | |
| "eval_f1": 62.43774864185708, | |
| "eval_runtime": 2.987, | |
| "eval_samples_per_second": 2871.487, | |
| "eval_steps_per_second": 9.039, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.6546, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_exact_match": 55.40398740818468, | |
| "eval_f1": 63.027221748816174, | |
| "eval_runtime": 2.9863, | |
| "eval_samples_per_second": 2872.108, | |
| "eval_steps_per_second": 9.041, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.5371, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_exact_match": 55.49726011425906, | |
| "eval_f1": 62.521799916881605, | |
| "eval_runtime": 2.9886, | |
| "eval_samples_per_second": 2869.878, | |
| "eval_steps_per_second": 9.034, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.4433, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "eval_exact_match": 56.58155532237379, | |
| "eval_f1": 63.38356690410977, | |
| "eval_runtime": 3.0108, | |
| "eval_samples_per_second": 2848.707, | |
| "eval_steps_per_second": 8.968, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.3643, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 7.0, | |
| "eval_exact_match": 55.57887373207415, | |
| "eval_f1": 63.013080189247944, | |
| "eval_runtime": 2.9636, | |
| "eval_samples_per_second": 2894.147, | |
| "eval_steps_per_second": 9.111, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.301, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 8.0, | |
| "eval_exact_match": 56.61653258715168, | |
| "eval_f1": 63.33968306601299, | |
| "eval_runtime": 3.0118, | |
| "eval_samples_per_second": 2847.846, | |
| "eval_steps_per_second": 8.965, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 3e-05, | |
| "loss": 0.25, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "eval_exact_match": 57.234464264894484, | |
| "eval_f1": 63.485590393372654, | |
| "eval_runtime": 2.9902, | |
| "eval_samples_per_second": 2868.341, | |
| "eval_steps_per_second": 9.029, | |
| "step": 2475 | |
| } | |
| ], | |
| "max_steps": 2750, | |
| "num_train_epochs": 10, | |
| "total_flos": 298038669082624.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |