| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 37500, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.4355555555555555e-06, | |
| "loss": 0.0959, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 8.880000000000001e-06, | |
| "loss": 0.0055, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 1.3324444444444446e-05, | |
| "loss": 0.0042, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 1.776888888888889e-05, | |
| "loss": 0.0038, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 1.985872340425532e-05, | |
| "loss": 0.0033, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.9575035460992908e-05, | |
| "loss": 0.0033, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.92913475177305e-05, | |
| "loss": 0.0031, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.9007659574468088e-05, | |
| "loss": 0.0032, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 1.8723971631205676e-05, | |
| "loss": 0.0031, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 1.8440283687943264e-05, | |
| "loss": 0.003, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 1.815659574468085e-05, | |
| "loss": 0.0029, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 1.7872907801418443e-05, | |
| "loss": 0.003, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.758921985815603e-05, | |
| "loss": 0.0029, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.730553191489362e-05, | |
| "loss": 0.0029, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.7021843971631207e-05, | |
| "loss": 0.0029, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.6738156028368795e-05, | |
| "loss": 0.0029, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 1.6454468085106386e-05, | |
| "loss": 0.0028, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 1.6170780141843974e-05, | |
| "loss": 0.0029, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 1.5887092198581562e-05, | |
| "loss": 0.0028, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 1.560340425531915e-05, | |
| "loss": 0.0028, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.5320283687943263e-05, | |
| "loss": 0.0028, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.5036595744680853e-05, | |
| "loss": 0.0028, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.4753475177304965e-05, | |
| "loss": 0.0028, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 1.4469787234042553e-05, | |
| "loss": 0.0028, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 1.4186099290780141e-05, | |
| "loss": 0.0028, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 1.3902411347517733e-05, | |
| "loss": 0.0026, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 1.361872340425532e-05, | |
| "loss": 0.0026, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.3335035460992909e-05, | |
| "loss": 0.0027, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.3051347517730497e-05, | |
| "loss": 0.0026, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.2767659574468085e-05, | |
| "loss": 0.0026, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 1.24845390070922e-05, | |
| "loss": 0.0026, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 1.2201418439716313e-05, | |
| "loss": 0.0025, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 1.1917730496453901e-05, | |
| "loss": 0.0026, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 1.1634042553191491e-05, | |
| "loss": 0.0026, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.1350354609929079e-05, | |
| "loss": 0.0027, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.1066666666666669e-05, | |
| "loss": 0.0026, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0782978723404257e-05, | |
| "loss": 0.0027, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 1.0499290780141845e-05, | |
| "loss": 0.0026, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 1.0215602836879433e-05, | |
| "loss": 0.0026, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 9.932482269503547e-06, | |
| "loss": 0.0026, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 9.648794326241135e-06, | |
| "loss": 0.0026, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 9.365106382978723e-06, | |
| "loss": 0.0026, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 9.081418439716313e-06, | |
| "loss": 0.0025, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.798297872340427e-06, | |
| "loss": 0.0025, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 8.514609929078015e-06, | |
| "loss": 0.0025, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 8.230921985815603e-06, | |
| "loss": 0.0025, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 7.947234042553192e-06, | |
| "loss": 0.0025, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 7.664113475177307e-06, | |
| "loss": 0.0026, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 7.380425531914894e-06, | |
| "loss": 0.0026, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 7.096737588652483e-06, | |
| "loss": 0.0025, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.813049645390072e-06, | |
| "loss": 0.0024, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.529929078014184e-06, | |
| "loss": 0.0024, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 6.246241134751774e-06, | |
| "loss": 0.0024, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 5.962553191489362e-06, | |
| "loss": 0.0024, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 5.6788652482269516e-06, | |
| "loss": 0.0024, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 5.395744680851064e-06, | |
| "loss": 0.0024, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 5.112056737588653e-06, | |
| "loss": 0.0024, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.828368794326241e-06, | |
| "loss": 0.0024, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.54468085106383e-06, | |
| "loss": 0.0024, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 4.261560283687944e-06, | |
| "loss": 0.0024, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 3.977872340425532e-06, | |
| "loss": 0.0024, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 3.694184397163121e-06, | |
| "loss": 0.0025, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 3.41049645390071e-06, | |
| "loss": 0.0024, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 3.1273758865248228e-06, | |
| "loss": 0.0024, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.8436879432624116e-06, | |
| "loss": 0.0024, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.56e-06, | |
| "loss": 0.0025, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.276312056737589e-06, | |
| "loss": 0.0024, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 1.9931914893617023e-06, | |
| "loss": 0.0024, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 1.709503546099291e-06, | |
| "loss": 0.0024, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 1.4258156028368794e-06, | |
| "loss": 0.0024, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 1.1421276595744682e-06, | |
| "loss": 0.0023, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.590070921985816e-07, | |
| "loss": 0.0024, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.753191489361703e-07, | |
| "loss": 0.0023, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.916312056737589e-07, | |
| "loss": 0.0023, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 7.943262411347518e-09, | |
| "loss": 0.0024, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 37500, | |
| "total_flos": 538957701120000.0, | |
| "train_runtime": 865.2735, | |
| "train_samples_per_second": 43.339 | |
| } | |
| ], | |
| "max_steps": 37500, | |
| "num_train_epochs": 3, | |
| "total_flos": 538957701120000.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |