{ "best_metric": null, "best_model_checkpoint": null, "epoch": 10.0, "global_step": 9710, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.1, "learning_rate": 4.9485066941297637e-05, "loss": 2.9147, "step": 100 }, { "epoch": 0.21, "learning_rate": 4.8970133882595264e-05, "loss": 2.4978, "step": 200 }, { "epoch": 0.31, "learning_rate": 4.84552008238929e-05, "loss": 2.274, "step": 300 }, { "epoch": 0.41, "learning_rate": 4.7940267765190525e-05, "loss": 2.136, "step": 400 }, { "epoch": 0.51, "learning_rate": 4.742533470648816e-05, "loss": 2.0295, "step": 500 }, { "epoch": 0.62, "learning_rate": 4.691040164778579e-05, "loss": 1.9494, "step": 600 }, { "epoch": 0.72, "learning_rate": 4.639546858908342e-05, "loss": 1.8878, "step": 700 }, { "epoch": 0.82, "learning_rate": 4.5880535530381055e-05, "loss": 1.8246, "step": 800 }, { "epoch": 0.93, "learning_rate": 4.536560247167869e-05, "loss": 1.7813, "step": 900 }, { "epoch": 1.03, "learning_rate": 4.485066941297632e-05, "loss": 1.7285, "step": 1000 }, { "epoch": 1.13, "learning_rate": 4.433573635427395e-05, "loss": 1.6964, "step": 1100 }, { "epoch": 1.24, "learning_rate": 4.382080329557158e-05, "loss": 1.6462, "step": 1200 }, { "epoch": 1.34, "learning_rate": 4.330587023686921e-05, "loss": 1.6142, "step": 1300 }, { "epoch": 1.44, "learning_rate": 4.279093717816684e-05, "loss": 1.5939, "step": 1400 }, { "epoch": 1.54, "learning_rate": 4.2276004119464474e-05, "loss": 1.5741, "step": 1500 }, { "epoch": 1.65, "learning_rate": 4.17610710607621e-05, "loss": 1.5514, "step": 1600 }, { "epoch": 1.75, "learning_rate": 4.1246138002059735e-05, "loss": 1.5269, "step": 1700 }, { "epoch": 1.85, "learning_rate": 4.073120494335736e-05, "loss": 1.5055, "step": 1800 }, { "epoch": 1.96, "learning_rate": 4.0216271884655e-05, "loss": 1.4796, "step": 1900 }, { "epoch": 2.06, "learning_rate": 3.9701338825952624e-05, "loss": 1.479, "step": 2000 }, { "epoch": 2.16, "learning_rate": 3.918640576725026e-05, "loss": 1.4328, "step": 2100 }, { "epoch": 2.27, "learning_rate": 3.867147270854789e-05, "loss": 1.4311, "step": 2200 }, { "epoch": 2.37, "learning_rate": 3.815653964984553e-05, "loss": 1.4157, "step": 2300 }, { "epoch": 2.47, "learning_rate": 3.7641606591143154e-05, "loss": 1.3942, "step": 2400 }, { "epoch": 2.57, "learning_rate": 3.712667353244079e-05, "loss": 1.3872, "step": 2500 }, { "epoch": 2.68, "learning_rate": 3.6611740473738416e-05, "loss": 1.373, "step": 2600 }, { "epoch": 2.78, "learning_rate": 3.609680741503605e-05, "loss": 1.3508, "step": 2700 }, { "epoch": 2.88, "learning_rate": 3.558187435633368e-05, "loss": 1.3609, "step": 2800 }, { "epoch": 2.99, "learning_rate": 3.506694129763131e-05, "loss": 1.3498, "step": 2900 }, { "epoch": 3.09, "learning_rate": 3.455200823892894e-05, "loss": 1.3198, "step": 3000 }, { "epoch": 3.19, "learning_rate": 3.403707518022657e-05, "loss": 1.3052, "step": 3100 }, { "epoch": 3.3, "learning_rate": 3.35221421215242e-05, "loss": 1.3161, "step": 3200 }, { "epoch": 3.4, "learning_rate": 3.3007209062821834e-05, "loss": 1.2906, "step": 3300 }, { "epoch": 3.5, "learning_rate": 3.249227600411947e-05, "loss": 1.2864, "step": 3400 }, { "epoch": 3.6, "learning_rate": 3.1977342945417096e-05, "loss": 1.2838, "step": 3500 }, { "epoch": 3.71, "learning_rate": 3.146240988671473e-05, "loss": 1.2766, "step": 3600 }, { "epoch": 3.81, "learning_rate": 3.0947476828012364e-05, "loss": 1.2654, "step": 3700 }, { "epoch": 3.91, "learning_rate": 3.043254376930999e-05, "loss": 1.2532, "step": 3800 }, { "epoch": 4.02, "learning_rate": 2.9917610710607626e-05, "loss": 1.2465, "step": 3900 }, { "epoch": 4.12, "learning_rate": 2.9402677651905253e-05, "loss": 1.2295, "step": 4000 }, { "epoch": 4.22, "learning_rate": 2.8887744593202887e-05, "loss": 1.2284, "step": 4100 }, { "epoch": 4.33, "learning_rate": 2.8372811534500514e-05, "loss": 1.2304, "step": 4200 }, { "epoch": 4.43, "learning_rate": 2.785787847579815e-05, "loss": 1.2268, "step": 4300 }, { "epoch": 4.53, "learning_rate": 2.734294541709578e-05, "loss": 1.2098, "step": 4400 }, { "epoch": 4.63, "learning_rate": 2.682801235839341e-05, "loss": 1.2154, "step": 4500 }, { "epoch": 4.74, "learning_rate": 2.631307929969104e-05, "loss": 1.2007, "step": 4600 }, { "epoch": 4.84, "learning_rate": 2.5798146240988675e-05, "loss": 1.1932, "step": 4700 }, { "epoch": 4.94, "learning_rate": 2.5283213182286302e-05, "loss": 1.2034, "step": 4800 }, { "epoch": 5.05, "learning_rate": 2.4768280123583933e-05, "loss": 1.1826, "step": 4900 }, { "epoch": 5.15, "learning_rate": 2.4253347064881564e-05, "loss": 1.1724, "step": 5000 }, { "epoch": 5.25, "learning_rate": 2.3738414006179198e-05, "loss": 1.1758, "step": 5100 }, { "epoch": 5.36, "learning_rate": 2.322348094747683e-05, "loss": 1.1893, "step": 5200 }, { "epoch": 5.46, "learning_rate": 2.270854788877446e-05, "loss": 1.1681, "step": 5300 }, { "epoch": 5.56, "learning_rate": 2.2193614830072094e-05, "loss": 1.167, "step": 5400 }, { "epoch": 5.66, "learning_rate": 2.1678681771369724e-05, "loss": 1.1539, "step": 5500 }, { "epoch": 5.77, "learning_rate": 2.1163748712667355e-05, "loss": 1.1447, "step": 5600 }, { "epoch": 5.87, "learning_rate": 2.0648815653964986e-05, "loss": 1.1511, "step": 5700 }, { "epoch": 5.97, "learning_rate": 2.0133882595262617e-05, "loss": 1.1562, "step": 5800 }, { "epoch": 6.08, "learning_rate": 1.961894953656025e-05, "loss": 1.1269, "step": 5900 }, { "epoch": 6.18, "learning_rate": 1.910401647785788e-05, "loss": 1.1357, "step": 6000 }, { "epoch": 6.28, "learning_rate": 1.8589083419155512e-05, "loss": 1.1408, "step": 6100 }, { "epoch": 6.39, "learning_rate": 1.8074150360453143e-05, "loss": 1.1308, "step": 6200 }, { "epoch": 6.49, "learning_rate": 1.7559217301750774e-05, "loss": 1.1182, "step": 6300 }, { "epoch": 6.59, "learning_rate": 1.7044284243048405e-05, "loss": 1.1268, "step": 6400 }, { "epoch": 6.69, "learning_rate": 1.6529351184346035e-05, "loss": 1.1281, "step": 6500 }, { "epoch": 6.8, "learning_rate": 1.601441812564367e-05, "loss": 1.1391, "step": 6600 }, { "epoch": 6.9, "learning_rate": 1.54994850669413e-05, "loss": 1.1204, "step": 6700 }, { "epoch": 7.0, "learning_rate": 1.4984552008238931e-05, "loss": 1.1198, "step": 6800 }, { "epoch": 7.11, "learning_rate": 1.4469618949536562e-05, "loss": 1.1201, "step": 6900 }, { "epoch": 7.21, "learning_rate": 1.3954685890834193e-05, "loss": 1.1088, "step": 7000 }, { "epoch": 7.31, "learning_rate": 1.3439752832131825e-05, "loss": 1.109, "step": 7100 }, { "epoch": 7.42, "learning_rate": 1.2924819773429456e-05, "loss": 1.1025, "step": 7200 }, { "epoch": 7.52, "learning_rate": 1.2409886714727085e-05, "loss": 1.1016, "step": 7300 }, { "epoch": 7.62, "learning_rate": 1.1894953656024717e-05, "loss": 1.0987, "step": 7400 }, { "epoch": 7.72, "learning_rate": 1.1380020597322348e-05, "loss": 1.096, "step": 7500 }, { "epoch": 7.83, "learning_rate": 1.0865087538619979e-05, "loss": 1.0994, "step": 7600 }, { "epoch": 7.93, "learning_rate": 1.0350154479917611e-05, "loss": 1.0929, "step": 7700 }, { "epoch": 8.03, "learning_rate": 9.835221421215242e-06, "loss": 1.0961, "step": 7800 }, { "epoch": 8.14, "learning_rate": 9.320288362512873e-06, "loss": 1.0888, "step": 7900 }, { "epoch": 8.24, "learning_rate": 8.805355303810504e-06, "loss": 1.0947, "step": 8000 }, { "epoch": 8.34, "learning_rate": 8.290422245108138e-06, "loss": 1.0819, "step": 8100 }, { "epoch": 8.44, "learning_rate": 7.775489186405768e-06, "loss": 1.0882, "step": 8200 }, { "epoch": 8.55, "learning_rate": 7.260556127703399e-06, "loss": 1.0903, "step": 8300 }, { "epoch": 8.65, "learning_rate": 6.745623069001031e-06, "loss": 1.0891, "step": 8400 }, { "epoch": 8.75, "learning_rate": 6.230690010298662e-06, "loss": 1.0889, "step": 8500 }, { "epoch": 8.86, "learning_rate": 5.715756951596292e-06, "loss": 1.0865, "step": 8600 }, { "epoch": 8.96, "learning_rate": 5.200823892893924e-06, "loss": 1.0819, "step": 8700 }, { "epoch": 9.06, "learning_rate": 4.6858908341915555e-06, "loss": 1.0824, "step": 8800 }, { "epoch": 9.17, "learning_rate": 4.170957775489186e-06, "loss": 1.0729, "step": 8900 }, { "epoch": 9.27, "learning_rate": 3.6560247167868175e-06, "loss": 1.0824, "step": 9000 }, { "epoch": 9.37, "learning_rate": 3.141091658084449e-06, "loss": 1.0812, "step": 9100 }, { "epoch": 9.47, "learning_rate": 2.6261585993820807e-06, "loss": 1.0828, "step": 9200 }, { "epoch": 9.58, "learning_rate": 2.111225540679712e-06, "loss": 1.0772, "step": 9300 }, { "epoch": 9.68, "learning_rate": 1.596292481977343e-06, "loss": 1.0773, "step": 9400 }, { "epoch": 9.78, "learning_rate": 1.0813594232749742e-06, "loss": 1.0833, "step": 9500 }, { "epoch": 9.89, "learning_rate": 5.664263645726055e-07, "loss": 1.0771, "step": 9600 }, { "epoch": 9.99, "learning_rate": 5.149330587023687e-08, "loss": 1.0738, "step": 9700 }, { "epoch": 10.0, "step": 9710, "total_flos": 7.347441844550246e+16, "train_loss": 1.314396687811116, "train_runtime": 11608.2196, "train_samples_per_second": 53.503, "train_steps_per_second": 0.836 } ], "max_steps": 9710, "num_train_epochs": 10, "total_flos": 7.347441844550246e+16, "trial_name": null, "trial_params": null }