| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1500.0, | |
| "global_step": 27000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 27.78, | |
| "learning_rate": 9.814814814814815e-06, | |
| "loss": 3.2486, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 55.56, | |
| "learning_rate": 9.62962962962963e-06, | |
| "loss": 1.3764, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 83.33, | |
| "learning_rate": 9.444444444444445e-06, | |
| "loss": 0.5771, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 111.11, | |
| "learning_rate": 9.25925925925926e-06, | |
| "loss": 0.2647, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 138.89, | |
| "learning_rate": 9.074074074074075e-06, | |
| "loss": 0.1349, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 166.67, | |
| "learning_rate": 8.888888888888888e-06, | |
| "loss": 0.0852, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 194.44, | |
| "learning_rate": 8.703703703703705e-06, | |
| "loss": 0.0669, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 222.22, | |
| "learning_rate": 8.518518518518519e-06, | |
| "loss": 0.0575, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 250.0, | |
| "learning_rate": 8.333333333333334e-06, | |
| "loss": 0.0514, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 277.78, | |
| "learning_rate": 8.148148148148148e-06, | |
| "loss": 0.0477, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 305.56, | |
| "learning_rate": 7.962962962962963e-06, | |
| "loss": 0.0446, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 333.33, | |
| "learning_rate": 7.77777777777778e-06, | |
| "loss": 0.0428, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 361.11, | |
| "learning_rate": 7.592592592592594e-06, | |
| "loss": 0.0412, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 388.89, | |
| "learning_rate": 7.4074074074074075e-06, | |
| "loss": 0.0401, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 416.67, | |
| "learning_rate": 7.222222222222223e-06, | |
| "loss": 0.0393, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 444.44, | |
| "learning_rate": 7.0370370370370375e-06, | |
| "loss": 0.0384, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 472.22, | |
| "learning_rate": 6.851851851851853e-06, | |
| "loss": 0.0376, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 500.0, | |
| "learning_rate": 6.666666666666667e-06, | |
| "loss": 0.037, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 527.78, | |
| "learning_rate": 6.481481481481482e-06, | |
| "loss": 0.0366, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 555.56, | |
| "learning_rate": 6.296296296296297e-06, | |
| "loss": 0.0362, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 583.33, | |
| "learning_rate": 6.111111111111112e-06, | |
| "loss": 0.0356, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 611.11, | |
| "learning_rate": 5.925925925925926e-06, | |
| "loss": 0.0353, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 638.89, | |
| "learning_rate": 5.740740740740741e-06, | |
| "loss": 0.0348, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 666.67, | |
| "learning_rate": 5.555555555555557e-06, | |
| "loss": 0.0346, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 694.44, | |
| "learning_rate": 5.370370370370371e-06, | |
| "loss": 0.0343, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 722.22, | |
| "learning_rate": 5.185185185185185e-06, | |
| "loss": 0.0339, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 750.0, | |
| "learning_rate": 5e-06, | |
| "loss": 0.0338, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 777.78, | |
| "learning_rate": 4.814814814814815e-06, | |
| "loss": 0.0335, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 805.56, | |
| "learning_rate": 4.62962962962963e-06, | |
| "loss": 0.0335, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 833.33, | |
| "learning_rate": 4.444444444444444e-06, | |
| "loss": 0.0331, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 861.11, | |
| "learning_rate": 4.2592592592592596e-06, | |
| "loss": 0.0329, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 888.89, | |
| "learning_rate": 4.074074074074074e-06, | |
| "loss": 0.0327, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 916.67, | |
| "learning_rate": 3.88888888888889e-06, | |
| "loss": 0.0325, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 944.44, | |
| "learning_rate": 3.7037037037037037e-06, | |
| "loss": 0.0322, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 972.22, | |
| "learning_rate": 3.5185185185185187e-06, | |
| "loss": 0.0323, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1000.0, | |
| "learning_rate": 3.3333333333333333e-06, | |
| "loss": 0.032, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1027.78, | |
| "learning_rate": 3.1481481481481483e-06, | |
| "loss": 0.0319, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1055.56, | |
| "learning_rate": 2.962962962962963e-06, | |
| "loss": 0.0316, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1083.33, | |
| "learning_rate": 2.7777777777777783e-06, | |
| "loss": 0.0317, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1111.11, | |
| "learning_rate": 2.5925925925925925e-06, | |
| "loss": 0.0315, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1138.89, | |
| "learning_rate": 2.4074074074074075e-06, | |
| "loss": 0.0313, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1166.67, | |
| "learning_rate": 2.222222222222222e-06, | |
| "loss": 0.0312, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1194.44, | |
| "learning_rate": 2.037037037037037e-06, | |
| "loss": 0.0312, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1222.22, | |
| "learning_rate": 1.8518518518518519e-06, | |
| "loss": 0.031, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1250.0, | |
| "learning_rate": 1.6666666666666667e-06, | |
| "loss": 0.031, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1277.78, | |
| "learning_rate": 1.4814814814814815e-06, | |
| "loss": 0.031, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1305.56, | |
| "learning_rate": 1.2962962962962962e-06, | |
| "loss": 0.0308, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1333.33, | |
| "learning_rate": 1.111111111111111e-06, | |
| "loss": 0.0306, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1361.11, | |
| "learning_rate": 9.259259259259259e-07, | |
| "loss": 0.0307, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1388.89, | |
| "learning_rate": 7.407407407407407e-07, | |
| "loss": 0.0306, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1416.67, | |
| "learning_rate": 5.555555555555555e-07, | |
| "loss": 0.0306, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1444.44, | |
| "learning_rate": 3.7037037037037036e-07, | |
| "loss": 0.0305, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1472.22, | |
| "learning_rate": 1.8518518518518518e-07, | |
| "loss": 0.0305, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1500.0, | |
| "learning_rate": 0.0, | |
| "loss": 0.0305, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1500.0, | |
| "step": 27000, | |
| "total_flos": 1.8317092847616e+16, | |
| "train_runtime": 3225.4397, | |
| "train_samples_per_second": 8.371 | |
| } | |
| ], | |
| "max_steps": 27000, | |
| "num_train_epochs": 1500, | |
| "total_flos": 1.8317092847616e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |