{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.6913836314925245, "global_step": 40000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "learning_rate": 4.9913577046063434e-05, "loss": 2.9531, "step": 500 }, { "epoch": 0.02, "learning_rate": 4.982715409212687e-05, "loss": 2.7985, "step": 1000 }, { "epoch": 0.03, "learning_rate": 4.97407311381903e-05, "loss": 2.7573, "step": 1500 }, { "epoch": 0.03, "learning_rate": 4.965430818425374e-05, "loss": 2.7173, "step": 2000 }, { "epoch": 0.04, "learning_rate": 4.956788523031717e-05, "loss": 2.7035, "step": 2500 }, { "epoch": 0.05, "learning_rate": 4.948146227638061e-05, "loss": 2.6966, "step": 3000 }, { "epoch": 0.06, "learning_rate": 4.939503932244404e-05, "loss": 2.6763, "step": 3500 }, { "epoch": 0.07, "learning_rate": 4.930861636850748e-05, "loss": 2.6809, "step": 4000 }, { "epoch": 0.08, "learning_rate": 4.922219341457091e-05, "loss": 2.6415, "step": 4500 }, { "epoch": 0.09, "learning_rate": 4.913577046063435e-05, "loss": 2.6747, "step": 5000 }, { "epoch": 0.1, "learning_rate": 4.904934750669778e-05, "loss": 2.6637, "step": 5500 }, { "epoch": 0.1, "learning_rate": 4.896292455276122e-05, "loss": 2.6527, "step": 6000 }, { "epoch": 0.11, "learning_rate": 4.887650159882465e-05, "loss": 2.6423, "step": 6500 }, { "epoch": 0.12, "learning_rate": 4.879007864488808e-05, "loss": 2.6331, "step": 7000 }, { "epoch": 0.13, "learning_rate": 4.8703655690951524e-05, "loss": 2.5868, "step": 7500 }, { "epoch": 0.14, "learning_rate": 4.8617232737014955e-05, "loss": 2.5901, "step": 8000 }, { "epoch": 0.15, "learning_rate": 4.853080978307839e-05, "loss": 2.5846, "step": 8500 }, { "epoch": 0.16, "learning_rate": 4.8444386829141824e-05, "loss": 2.6172, "step": 9000 }, { "epoch": 0.16, "learning_rate": 4.835796387520526e-05, "loss": 2.6035, "step": 9500 }, { "epoch": 0.17, "learning_rate": 4.827154092126869e-05, "loss": 2.6213, "step": 10000 }, { "epoch": 0.18, "learning_rate": 4.8185117967332124e-05, "loss": 2.5827, "step": 10500 }, { "epoch": 0.19, "learning_rate": 4.809869501339556e-05, "loss": 2.606, "step": 11000 }, { "epoch": 0.2, "learning_rate": 4.801227205945899e-05, "loss": 2.5827, "step": 11500 }, { "epoch": 0.21, "learning_rate": 4.792584910552243e-05, "loss": 2.5601, "step": 12000 }, { "epoch": 0.22, "learning_rate": 4.783942615158586e-05, "loss": 2.554, "step": 12500 }, { "epoch": 0.22, "learning_rate": 4.77530031976493e-05, "loss": 2.5501, "step": 13000 }, { "epoch": 0.23, "learning_rate": 4.766658024371273e-05, "loss": 2.5841, "step": 13500 }, { "epoch": 0.24, "learning_rate": 4.758015728977617e-05, "loss": 2.547, "step": 14000 }, { "epoch": 0.25, "learning_rate": 4.74937343358396e-05, "loss": 2.5823, "step": 14500 }, { "epoch": 0.26, "learning_rate": 4.740731138190304e-05, "loss": 2.5408, "step": 15000 }, { "epoch": 0.27, "learning_rate": 4.732088842796647e-05, "loss": 2.5678, "step": 15500 }, { "epoch": 0.28, "learning_rate": 4.72344654740299e-05, "loss": 2.5628, "step": 16000 }, { "epoch": 0.29, "learning_rate": 4.714804252009334e-05, "loss": 2.5549, "step": 16500 }, { "epoch": 0.29, "learning_rate": 4.706161956615677e-05, "loss": 2.5335, "step": 17000 }, { "epoch": 0.3, "learning_rate": 4.697519661222021e-05, "loss": 2.5287, "step": 17500 }, { "epoch": 0.31, "learning_rate": 4.688877365828364e-05, "loss": 2.5478, "step": 18000 }, { "epoch": 0.32, "learning_rate": 4.680235070434708e-05, "loss": 2.5501, "step": 18500 }, { "epoch": 0.33, "learning_rate": 4.671592775041051e-05, "loss": 2.5335, "step": 19000 }, { "epoch": 0.34, "learning_rate": 4.6629504796473946e-05, "loss": 2.5057, "step": 19500 }, { "epoch": 0.35, "learning_rate": 4.654308184253738e-05, "loss": 2.5316, "step": 20000 }, { "epoch": 0.35, "learning_rate": 4.6456658888600815e-05, "loss": 2.5584, "step": 20500 }, { "epoch": 0.36, "learning_rate": 4.6370235934664246e-05, "loss": 2.5184, "step": 21000 }, { "epoch": 0.37, "learning_rate": 4.6283812980727684e-05, "loss": 2.5301, "step": 21500 }, { "epoch": 0.38, "learning_rate": 4.6197390026791115e-05, "loss": 2.5337, "step": 22000 }, { "epoch": 0.39, "learning_rate": 4.611096707285455e-05, "loss": 2.5136, "step": 22500 }, { "epoch": 0.4, "learning_rate": 4.602454411891799e-05, "loss": 2.4641, "step": 23000 }, { "epoch": 0.41, "learning_rate": 4.593812116498142e-05, "loss": 2.5473, "step": 23500 }, { "epoch": 0.41, "learning_rate": 4.585169821104486e-05, "loss": 2.5219, "step": 24000 }, { "epoch": 0.42, "learning_rate": 4.576527525710829e-05, "loss": 2.5111, "step": 24500 }, { "epoch": 0.43, "learning_rate": 4.567885230317173e-05, "loss": 2.5045, "step": 25000 }, { "epoch": 0.44, "learning_rate": 4.559242934923516e-05, "loss": 2.4803, "step": 25500 }, { "epoch": 0.45, "learning_rate": 4.550600639529859e-05, "loss": 2.478, "step": 26000 }, { "epoch": 0.46, "learning_rate": 4.541958344136203e-05, "loss": 2.502, "step": 26500 }, { "epoch": 0.47, "learning_rate": 4.533316048742546e-05, "loss": 2.5121, "step": 27000 }, { "epoch": 0.48, "learning_rate": 4.52467375334889e-05, "loss": 2.4963, "step": 27500 }, { "epoch": 0.48, "learning_rate": 4.516031457955233e-05, "loss": 2.4793, "step": 28000 }, { "epoch": 0.49, "learning_rate": 4.507389162561577e-05, "loss": 2.5068, "step": 28500 }, { "epoch": 0.5, "learning_rate": 4.49874686716792e-05, "loss": 2.5034, "step": 29000 }, { "epoch": 0.51, "learning_rate": 4.490104571774264e-05, "loss": 2.4744, "step": 29500 }, { "epoch": 0.52, "learning_rate": 4.481462276380607e-05, "loss": 2.4565, "step": 30000 }, { "epoch": 0.53, "learning_rate": 4.4728199809869506e-05, "loss": 2.491, "step": 30500 }, { "epoch": 0.54, "learning_rate": 4.464177685593294e-05, "loss": 2.5165, "step": 31000 }, { "epoch": 0.54, "learning_rate": 4.4555353901996375e-05, "loss": 2.4936, "step": 31500 }, { "epoch": 0.55, "learning_rate": 4.4468930948059806e-05, "loss": 2.4991, "step": 32000 }, { "epoch": 0.56, "learning_rate": 4.438250799412324e-05, "loss": 2.454, "step": 32500 }, { "epoch": 0.57, "learning_rate": 4.4296085040186675e-05, "loss": 2.4699, "step": 33000 }, { "epoch": 0.58, "learning_rate": 4.4209662086250106e-05, "loss": 2.4951, "step": 33500 }, { "epoch": 0.59, "learning_rate": 4.4123239132313544e-05, "loss": 2.4536, "step": 34000 }, { "epoch": 0.6, "learning_rate": 4.4036816178376975e-05, "loss": 2.4745, "step": 34500 }, { "epoch": 0.6, "learning_rate": 4.395039322444041e-05, "loss": 2.4752, "step": 35000 }, { "epoch": 0.61, "learning_rate": 4.3863970270503844e-05, "loss": 2.4757, "step": 35500 }, { "epoch": 0.62, "learning_rate": 4.377754731656728e-05, "loss": 2.5043, "step": 36000 }, { "epoch": 0.63, "learning_rate": 4.3691124362630714e-05, "loss": 2.4513, "step": 36500 }, { "epoch": 0.64, "learning_rate": 4.360470140869415e-05, "loss": 2.4683, "step": 37000 }, { "epoch": 0.65, "learning_rate": 4.351827845475759e-05, "loss": 2.4723, "step": 37500 }, { "epoch": 0.66, "learning_rate": 4.343185550082102e-05, "loss": 2.4607, "step": 38000 }, { "epoch": 0.67, "learning_rate": 4.334543254688446e-05, "loss": 2.4937, "step": 38500 }, { "epoch": 0.67, "learning_rate": 4.325900959294789e-05, "loss": 2.4491, "step": 39000 }, { "epoch": 0.68, "learning_rate": 4.317258663901133e-05, "loss": 2.4476, "step": 39500 }, { "epoch": 0.69, "learning_rate": 4.308616368507476e-05, "loss": 2.479, "step": 40000 } ], "max_steps": 289275, "num_train_epochs": 5, "total_flos": 3.2856743048380416e+16, "trial_name": null, "trial_params": null }