| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.0, | |
| "global_step": 21720, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 2.930939226519337e-05, | |
| "loss": 0.6989, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 2.861878453038674e-05, | |
| "loss": 0.6341, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 2.7928176795580113e-05, | |
| "loss": 0.5836, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 2.7237569060773482e-05, | |
| "loss": 0.5758, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 2.6546961325966852e-05, | |
| "loss": 0.5956, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 2.585635359116022e-05, | |
| "loss": 0.584, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 2.516574585635359e-05, | |
| "loss": 0.5533, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 2.4475138121546964e-05, | |
| "loss": 0.5688, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 2.3784530386740334e-05, | |
| "loss": 0.5749, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 2.3093922651933703e-05, | |
| "loss": 0.5635, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 2.2403314917127073e-05, | |
| "loss": 0.554, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 2.1712707182320442e-05, | |
| "loss": 0.5563, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 2.1022099447513815e-05, | |
| "loss": 0.5607, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 2.0331491712707185e-05, | |
| "loss": 0.5657, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.9640883977900554e-05, | |
| "loss": 0.5345, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 1.8950276243093924e-05, | |
| "loss": 0.5502, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 1.825966850828729e-05, | |
| "loss": 0.5417, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 1.7569060773480663e-05, | |
| "loss": 0.5506, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 1.6878453038674033e-05, | |
| "loss": 0.5352, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.6187845303867402e-05, | |
| "loss": 0.5388, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.5497237569060772e-05, | |
| "loss": 0.5312, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.4806629834254145e-05, | |
| "loss": 0.5222, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 1.4116022099447514e-05, | |
| "loss": 0.4981, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.3425414364640886e-05, | |
| "loss": 0.5235, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 1.2734806629834255e-05, | |
| "loss": 0.5144, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2044198895027623e-05, | |
| "loss": 0.5037, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.1353591160220994e-05, | |
| "loss": 0.4856, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.0662983425414364e-05, | |
| "loss": 0.5175, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 9.972375690607735e-06, | |
| "loss": 0.5023, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 9.281767955801105e-06, | |
| "loss": 0.4884, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 8.591160220994474e-06, | |
| "loss": 0.5, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 7.900552486187846e-06, | |
| "loss": 0.5106, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 7.209944751381215e-06, | |
| "loss": 0.52, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 6.519337016574586e-06, | |
| "loss": 0.4942, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 5.828729281767956e-06, | |
| "loss": 0.4999, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 5.1381215469613265e-06, | |
| "loss": 0.4934, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 4.447513812154697e-06, | |
| "loss": 0.5301, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.756906077348067e-06, | |
| "loss": 0.4763, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.0662983425414365e-06, | |
| "loss": 0.4909, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 2.375690607734807e-06, | |
| "loss": 0.506, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.6850828729281769e-06, | |
| "loss": 0.5036, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 9.944751381215469e-07, | |
| "loss": 0.4891, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.0386740331491715e-07, | |
| "loss": 0.4841, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "step": 21720, | |
| "total_flos": 2.645641233904435e+16, | |
| "train_loss": 0.5347459301325058, | |
| "train_runtime": 6415.5456, | |
| "train_samples_per_second": 40.626, | |
| "train_steps_per_second": 3.386 | |
| } | |
| ], | |
| "max_steps": 21720, | |
| "num_train_epochs": 2, | |
| "total_flos": 2.645641233904435e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |