| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 10.0, | |
| "global_step": 31110, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.919639987142398e-05, | |
| "loss": 3.736, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.839279974284796e-05, | |
| "loss": 2.694, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.758919961427194e-05, | |
| "loss": 2.3793, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.678559948569592e-05, | |
| "loss": 2.2157, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.59819993571199e-05, | |
| "loss": 2.0865, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.517839922854388e-05, | |
| "loss": 1.9888, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 4.4374799099967856e-05, | |
| "loss": 1.7791, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 4.357119897139183e-05, | |
| "loss": 1.7181, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 4.2767598842815816e-05, | |
| "loss": 1.6788, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 4.196399871423979e-05, | |
| "loss": 1.6565, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 4.1160398585663775e-05, | |
| "loss": 1.6128, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 4.035679845708776e-05, | |
| "loss": 1.5839, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 3.9553198328511734e-05, | |
| "loss": 1.4628, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 3.874959819993572e-05, | |
| "loss": 1.3677, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 3.7945998071359693e-05, | |
| "loss": 1.3458, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 3.7142397942783676e-05, | |
| "loss": 1.3436, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 3.633879781420765e-05, | |
| "loss": 1.3476, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 3.5535197685631636e-05, | |
| "loss": 1.3265, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 3.473159755705561e-05, | |
| "loss": 1.2552, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 3.392799742847959e-05, | |
| "loss": 1.1139, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 3.312439729990357e-05, | |
| "loss": 1.1205, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 3.232079717132755e-05, | |
| "loss": 1.1202, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 3.151719704275153e-05, | |
| "loss": 1.1305, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 3.071359691417551e-05, | |
| "loss": 1.1291, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 2.9909996785599486e-05, | |
| "loss": 1.1071, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 2.9106396657023466e-05, | |
| "loss": 0.9143, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 2.8302796528447446e-05, | |
| "loss": 0.9409, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 2.7499196399871425e-05, | |
| "loss": 0.9381, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 2.6695596271295405e-05, | |
| "loss": 0.96, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 2.589199614271938e-05, | |
| "loss": 0.9616, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 2.508839601414336e-05, | |
| "loss": 0.948, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 5.14, | |
| "learning_rate": 2.4284795885567344e-05, | |
| "loss": 0.7787, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 5.3, | |
| "learning_rate": 2.3481195756991323e-05, | |
| "loss": 0.7821, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 5.46, | |
| "learning_rate": 2.2677595628415303e-05, | |
| "loss": 0.7927, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 5.63, | |
| "learning_rate": 2.1873995499839283e-05, | |
| "loss": 0.7907, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 5.79, | |
| "learning_rate": 2.107039537126326e-05, | |
| "loss": 0.7904, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 5.95, | |
| "learning_rate": 2.026679524268724e-05, | |
| "loss": 0.804, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 6.11, | |
| "learning_rate": 1.9463195114111218e-05, | |
| "loss": 0.6899, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 6.27, | |
| "learning_rate": 1.8659594985535198e-05, | |
| "loss": 0.6512, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 6.43, | |
| "learning_rate": 1.7855994856959177e-05, | |
| "loss": 0.6662, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 6.59, | |
| "learning_rate": 1.7052394728383157e-05, | |
| "loss": 0.6688, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 1.6248794599807137e-05, | |
| "loss": 0.6663, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 6.91, | |
| "learning_rate": 1.5445194471231116e-05, | |
| "loss": 0.6584, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 7.07, | |
| "learning_rate": 1.4641594342655096e-05, | |
| "loss": 0.6115, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 7.23, | |
| "learning_rate": 1.3837994214079075e-05, | |
| "loss": 0.5433, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 1.3034394085503055e-05, | |
| "loss": 0.5653, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 7.55, | |
| "learning_rate": 1.2230793956927033e-05, | |
| "loss": 0.5509, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 7.71, | |
| "learning_rate": 1.1427193828351013e-05, | |
| "loss": 0.553, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 7.88, | |
| "learning_rate": 1.0623593699774992e-05, | |
| "loss": 0.5659, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 9.819993571198972e-06, | |
| "loss": 0.5369, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 8.2, | |
| "learning_rate": 9.016393442622952e-06, | |
| "loss": 0.4594, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 8.36, | |
| "learning_rate": 8.212793314046931e-06, | |
| "loss": 0.4692, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 8.52, | |
| "learning_rate": 7.40919318547091e-06, | |
| "loss": 0.4773, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 6.605593056894889e-06, | |
| "loss": 0.479, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 8.84, | |
| "learning_rate": 5.801992928318869e-06, | |
| "loss": 0.4784, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 4.998392799742848e-06, | |
| "loss": 0.4781, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 9.16, | |
| "learning_rate": 4.194792671166828e-06, | |
| "loss": 0.423, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 9.32, | |
| "learning_rate": 3.391192542590807e-06, | |
| "loss": 0.4236, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 9.48, | |
| "learning_rate": 2.5875924140147865e-06, | |
| "loss": 0.4197, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 9.64, | |
| "learning_rate": 1.7839922854387657e-06, | |
| "loss": 0.4091, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 9.8, | |
| "learning_rate": 9.80392156862745e-07, | |
| "loss": 0.4076, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 9.96, | |
| "learning_rate": 1.7679202828672455e-07, | |
| "loss": 0.4182, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 10.0, | |
| "step": 31110, | |
| "total_flos": 4.532670747485798e+16, | |
| "train_loss": 1.0457585005237526, | |
| "train_runtime": 9612.1472, | |
| "train_samples_per_second": 32.359, | |
| "train_steps_per_second": 3.237 | |
| } | |
| ], | |
| "max_steps": 31110, | |
| "num_train_epochs": 10, | |
| "total_flos": 4.532670747485798e+16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |