| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 42003, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9404804418732e-05, | |
| "loss": 1.2354, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.880960883746399e-05, | |
| "loss": 0.91, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.821441325619599e-05, | |
| "loss": 0.8225, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.761921767492799e-05, | |
| "loss": 0.769, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.702402209365998e-05, | |
| "loss": 0.7282, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.642882651239198e-05, | |
| "loss": 0.7009, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.583363093112397e-05, | |
| "loss": 0.6759, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.523843534985596e-05, | |
| "loss": 0.651, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.464323976858796e-05, | |
| "loss": 0.6379, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.404804418731996e-05, | |
| "loss": 0.6195, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.345284860605195e-05, | |
| "loss": 0.6079, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.2857653024783946e-05, | |
| "loss": 0.5897, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.226245744351594e-05, | |
| "loss": 0.5754, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1667261862247936e-05, | |
| "loss": 0.5669, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.1072066280979934e-05, | |
| "loss": 0.5562, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0476870699711925e-05, | |
| "loss": 0.5452, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.9881675118443923e-05, | |
| "loss": 0.5385, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.928647953717592e-05, | |
| "loss": 0.5336, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.869128395590791e-05, | |
| "loss": 0.5261, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.809608837463991e-05, | |
| "loss": 0.5156, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.75008927933719e-05, | |
| "loss": 0.5144, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.69056972121039e-05, | |
| "loss": 0.5038, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.63105016308359e-05, | |
| "loss": 0.4979, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.571530604956789e-05, | |
| "loss": 0.4946, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.512011046829988e-05, | |
| "loss": 0.486, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.452491488703188e-05, | |
| "loss": 0.4832, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.392971930576387e-05, | |
| "loss": 0.4798, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.333452372449587e-05, | |
| "loss": 0.4725, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.273932814322787e-05, | |
| "loss": 0.4623, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.214413256195986e-05, | |
| "loss": 0.4618, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.154893698069186e-05, | |
| "loss": 0.4546, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.0953741399423855e-05, | |
| "loss": 0.4488, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0358545818155846e-05, | |
| "loss": 0.4484, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.9763350236887845e-05, | |
| "loss": 0.4449, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.916815465561984e-05, | |
| "loss": 0.4394, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.8572959074351834e-05, | |
| "loss": 0.4408, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.7977763493083832e-05, | |
| "loss": 0.4358, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.7382567911815827e-05, | |
| "loss": 0.4379, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.678737233054782e-05, | |
| "loss": 0.429, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6192176749279813e-05, | |
| "loss": 0.4269, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5596981168011808e-05, | |
| "loss": 0.4276, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.5001785586743803e-05, | |
| "loss": 0.4183, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.44065900054758e-05, | |
| "loss": 0.4245, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3811394424207795e-05, | |
| "loss": 0.4195, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.321619884293979e-05, | |
| "loss": 0.4186, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2621003261671785e-05, | |
| "loss": 0.415, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2025807680403783e-05, | |
| "loss": 0.4157, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1430612099135778e-05, | |
| "loss": 0.4152, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0835416517867773e-05, | |
| "loss": 0.4117, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0240220936599767e-05, | |
| "loss": 0.4085, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9645025355331766e-05, | |
| "loss": 0.4033, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9049829774063757e-05, | |
| "loss": 0.4031, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.8454634192795752e-05, | |
| "loss": 0.4035, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.785943861152775e-05, | |
| "loss": 0.4015, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7264243030259745e-05, | |
| "loss": 0.4022, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.666904744899174e-05, | |
| "loss": 0.4007, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.6073851867723734e-05, | |
| "loss": 0.3913, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.5478656286455732e-05, | |
| "loss": 0.3864, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.4883460705187727e-05, | |
| "loss": 0.3889, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.428826512391972e-05, | |
| "loss": 0.387, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3693069542651715e-05, | |
| "loss": 0.3816, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3097873961383711e-05, | |
| "loss": 0.3889, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2502678380115706e-05, | |
| "loss": 0.3899, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.1907482798847703e-05, | |
| "loss": 0.3857, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1312287217579697e-05, | |
| "loss": 0.382, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0717091636311692e-05, | |
| "loss": 0.3841, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0121896055043688e-05, | |
| "loss": 0.3823, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.526700473775683e-06, | |
| "loss": 0.3826, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.93150489250768e-06, | |
| "loss": 0.3797, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.336309311239673e-06, | |
| "loss": 0.3752, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.74111372997167e-06, | |
| "loss": 0.374, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.145918148703665e-06, | |
| "loss": 0.3801, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.550722567435659e-06, | |
| "loss": 0.3762, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 5.955526986167654e-06, | |
| "loss": 0.373, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.36033140489965e-06, | |
| "loss": 0.3767, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.7651358236316455e-06, | |
| "loss": 0.375, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.169940242363641e-06, | |
| "loss": 0.3757, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.574744661095636e-06, | |
| "loss": 0.378, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.9795490798276315e-06, | |
| "loss": 0.3731, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.3843534985596267e-06, | |
| "loss": 0.3769, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.789157917291622e-06, | |
| "loss": 0.368, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1939623360236175e-06, | |
| "loss": 0.3779, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 5.987667547556127e-07, | |
| "loss": 0.3705, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 3.5711734876080284e-09, | |
| "loss": 0.3727, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 42003, | |
| "total_flos": 2.741262336802898e+17, | |
| "train_loss": 0.4736848745993613, | |
| "train_runtime": 22801.7195, | |
| "train_samples_per_second": 18.42, | |
| "train_steps_per_second": 1.842 | |
| } | |
| ], | |
| "max_steps": 42003, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.741262336802898e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |