| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.567377698301188, | |
| "global_step": 20100, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9576751117734728e-05, | |
| "loss": 4.2394, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.9150947413242498e-05, | |
| "loss": 3.1532, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8725143708750267e-05, | |
| "loss": 2.8126, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.829934000425804e-05, | |
| "loss": 2.5858, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.787353629976581e-05, | |
| "loss": 2.4035, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 1.744773259527358e-05, | |
| "loss": 2.2331, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.7021928890781352e-05, | |
| "loss": 2.0692, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 1.6596125186289122e-05, | |
| "loss": 1.9111, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 1.6170321481796895e-05, | |
| "loss": 1.7653, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5744517777304664e-05, | |
| "loss": 1.669, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 1.5320417287630403e-05, | |
| "loss": 1.588, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.4895465190547159e-05, | |
| "loss": 1.5321, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.4469661486054928e-05, | |
| "loss": 1.4851, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 1.4044709388971684e-05, | |
| "loss": 1.4505, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.3618905684479455e-05, | |
| "loss": 1.4109, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 1.3193101979987227e-05, | |
| "loss": 1.3851, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 1.2767298275494996e-05, | |
| "loss": 1.3566, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2341494571002767e-05, | |
| "loss": 1.3336, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 1.1915690866510539e-05, | |
| "loss": 1.316, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.1491590376836279e-05, | |
| "loss": 1.3038, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 1.106578667234405e-05, | |
| "loss": 1.2931, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 1.064168618266979e-05, | |
| "loss": 1.2743, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 1.0215882478177562e-05, | |
| "loss": 1.27, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 9.792633595912285e-06, | |
| "loss": 1.2495, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.366829891420056e-06, | |
| "loss": 1.239, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 8.941026186927828e-06, | |
| "loss": 1.2301, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.515222482435597e-06, | |
| "loss": 1.2189, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 8.091121992761337e-06, | |
| "loss": 1.2179, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.665318288269109e-06, | |
| "loss": 1.2099, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.241217798594848e-06, | |
| "loss": 1.1948, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.815414094102619e-06, | |
| "loss": 1.1981, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.391313604428359e-06, | |
| "loss": 1.1782, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 5.965509899936129e-06, | |
| "loss": 1.1714, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.54140941026187e-06, | |
| "loss": 1.1706, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.115605705769641e-06, | |
| "loss": 1.161, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.69150521609538e-06, | |
| "loss": 1.1585, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.265701511603152e-06, | |
| "loss": 1.1535, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.841601021928891e-06, | |
| "loss": 1.146, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.4157973174366623e-06, | |
| "loss": 1.1474, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.991696827762402e-06, | |
| "loss": 1.1486, | |
| "step": 20000 | |
| } | |
| ], | |
| "max_steps": 23487, | |
| "num_train_epochs": 3, | |
| "total_flos": 1436562323668992.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |