{ "best_metric": null, "best_model_checkpoint": null, "epoch": 3.0, "global_step": 421875, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.0, "learning_rate": 5.902105637419509e-07, "loss": 1.046, "step": 500 }, { "epoch": 0.01, "learning_rate": 1.182791451033066e-06, "loss": 0.0314, "step": 1000 }, { "epoch": 0.01, "learning_rate": 1.7753723383241813e-06, "loss": 0.0192, "step": 1500 }, { "epoch": 0.01, "learning_rate": 2.3679532256152967e-06, "loss": 0.0164, "step": 2000 }, { "epoch": 0.02, "learning_rate": 2.9605341129064116e-06, "loss": 0.0152, "step": 2500 }, { "epoch": 0.02, "learning_rate": 3.553115000197527e-06, "loss": 0.0133, "step": 3000 }, { "epoch": 0.02, "learning_rate": 4.145695887488642e-06, "loss": 0.0121, "step": 3500 }, { "epoch": 0.03, "learning_rate": 4.738276774779758e-06, "loss": 0.0109, "step": 4000 }, { "epoch": 0.03, "learning_rate": 5.330857662070873e-06, "loss": 0.01, "step": 4500 }, { "epoch": 0.04, "learning_rate": 5.923438549361989e-06, "loss": 0.0086, "step": 5000 }, { "epoch": 0.04, "learning_rate": 6.516019436653103e-06, "loss": 0.0069, "step": 5500 }, { "epoch": 0.04, "learning_rate": 7.1086003239442186e-06, "loss": 0.0055, "step": 6000 }, { "epoch": 0.05, "learning_rate": 7.701181211235334e-06, "loss": 0.0047, "step": 6500 }, { "epoch": 0.05, "learning_rate": 8.293762098526448e-06, "loss": 0.0042, "step": 7000 }, { "epoch": 0.05, "learning_rate": 8.886342985817564e-06, "loss": 0.0033, "step": 7500 }, { "epoch": 0.06, "learning_rate": 9.47892387310868e-06, "loss": 0.0029, "step": 8000 }, { "epoch": 0.06, "learning_rate": 1.0071504760399794e-05, "loss": 0.0027, "step": 8500 }, { "epoch": 0.06, "learning_rate": 1.066408564769091e-05, "loss": 0.0025, "step": 9000 }, { "epoch": 0.07, "learning_rate": 1.1256666534982026e-05, "loss": 0.0022, "step": 9500 }, { "epoch": 0.07, "learning_rate": 1.1849247422273141e-05, "loss": 0.0021, "step": 10000 }, { "epoch": 0.07, "learning_rate": 1.2441828309564255e-05, "loss": 0.0021, "step": 10500 }, { "epoch": 0.08, "learning_rate": 1.3034409196855371e-05, "loss": 0.0019, "step": 11000 }, { "epoch": 0.08, "learning_rate": 1.3626990084146487e-05, "loss": 0.0018, "step": 11500 }, { "epoch": 0.09, "learning_rate": 1.4219570971437601e-05, "loss": 0.0018, "step": 12000 }, { "epoch": 0.09, "learning_rate": 1.4810966696954135e-05, "loss": 0.0017, "step": 12500 }, { "epoch": 0.09, "learning_rate": 1.540354758424525e-05, "loss": 0.0016, "step": 13000 }, { "epoch": 0.1, "learning_rate": 1.5996128471536368e-05, "loss": 0.0015, "step": 13500 }, { "epoch": 0.1, "learning_rate": 1.658870935882748e-05, "loss": 0.0016, "step": 14000 }, { "epoch": 0.1, "learning_rate": 1.7180105084344013e-05, "loss": 0.0015, "step": 14500 }, { "epoch": 0.11, "learning_rate": 1.777268597163513e-05, "loss": 0.0015, "step": 15000 }, { "epoch": 0.11, "learning_rate": 1.8365266858926244e-05, "loss": 0.0014, "step": 15500 }, { "epoch": 0.11, "learning_rate": 1.895784774621736e-05, "loss": 0.0013, "step": 16000 }, { "epoch": 0.12, "learning_rate": 1.9549243471733892e-05, "loss": 0.0013, "step": 16500 }, { "epoch": 0.12, "learning_rate": 2.0141824359025008e-05, "loss": 0.0013, "step": 17000 }, { "epoch": 0.12, "learning_rate": 2.0734405246316124e-05, "loss": 0.0013, "step": 17500 }, { "epoch": 0.13, "learning_rate": 2.132698613360724e-05, "loss": 0.0013, "step": 18000 }, { "epoch": 0.13, "learning_rate": 2.1918381859123768e-05, "loss": 0.0013, "step": 18500 }, { "epoch": 0.14, "learning_rate": 2.2510962746414887e-05, "loss": 0.0012, "step": 19000 }, { "epoch": 0.14, "learning_rate": 2.3103543633706003e-05, "loss": 0.0012, "step": 19500 }, { "epoch": 0.14, "learning_rate": 2.3696124520997115e-05, "loss": 0.0012, "step": 20000 }, { "epoch": 0.15, "learning_rate": 2.428752024651365e-05, "loss": 0.0011, "step": 20500 }, { "epoch": 0.15, "learning_rate": 2.4880101133804767e-05, "loss": 0.0011, "step": 21000 }, { "epoch": 0.15, "learning_rate": 2.547268202109588e-05, "loss": 0.0011, "step": 21500 }, { "epoch": 0.16, "learning_rate": 2.6065262908386995e-05, "loss": 0.0012, "step": 22000 }, { "epoch": 0.16, "learning_rate": 2.665784379567811e-05, "loss": 0.0011, "step": 22500 }, { "epoch": 0.16, "learning_rate": 2.7249239521194643e-05, "loss": 0.0011, "step": 23000 }, { "epoch": 0.17, "learning_rate": 2.784182040848576e-05, "loss": 0.0011, "step": 23500 }, { "epoch": 0.17, "learning_rate": 2.8434401295776874e-05, "loss": 0.0011, "step": 24000 }, { "epoch": 0.17, "learning_rate": 2.9026982183067987e-05, "loss": 0.001, "step": 24500 }, { "epoch": 0.18, "learning_rate": 2.9619563070359106e-05, "loss": 0.0011, "step": 25000 }, { "epoch": 0.18, "learning_rate": 2.9986534261981735e-05, "loss": 0.0011, "step": 25500 }, { "epoch": 0.18, "learning_rate": 2.9948709155188848e-05, "loss": 0.001, "step": 26000 }, { "epoch": 0.19, "learning_rate": 2.9910884048395964e-05, "loss": 0.001, "step": 26500 }, { "epoch": 0.19, "learning_rate": 2.987305894160308e-05, "loss": 0.001, "step": 27000 }, { "epoch": 0.2, "learning_rate": 2.983530948502378e-05, "loss": 0.001, "step": 27500 }, { "epoch": 0.2, "learning_rate": 2.9797484378230894e-05, "loss": 0.001, "step": 28000 }, { "epoch": 0.2, "learning_rate": 2.975965927143801e-05, "loss": 0.001, "step": 28500 }, { "epoch": 0.21, "learning_rate": 2.9721834164645126e-05, "loss": 0.001, "step": 29000 }, { "epoch": 0.21, "learning_rate": 2.968400905785224e-05, "loss": 0.0009, "step": 29500 }, { "epoch": 0.21, "learning_rate": 2.9646183951059355e-05, "loss": 0.0009, "step": 30000 }, { "epoch": 0.22, "learning_rate": 2.9608434494480056e-05, "loss": 0.001, "step": 30500 }, { "epoch": 0.22, "learning_rate": 2.9570609387687172e-05, "loss": 0.001, "step": 31000 }, { "epoch": 0.22, "learning_rate": 2.9532784280894288e-05, "loss": 0.0009, "step": 31500 }, { "epoch": 0.23, "learning_rate": 2.94949591741014e-05, "loss": 0.0009, "step": 32000 }, { "epoch": 0.23, "learning_rate": 2.945713406730852e-05, "loss": 0.0009, "step": 32500 }, { "epoch": 0.23, "learning_rate": 2.9419384610729218e-05, "loss": 0.0009, "step": 33000 }, { "epoch": 0.24, "learning_rate": 2.9381559503936334e-05, "loss": 0.0009, "step": 33500 }, { "epoch": 0.24, "learning_rate": 2.9343734397143447e-05, "loss": 0.0009, "step": 34000 }, { "epoch": 0.25, "learning_rate": 2.9305909290350566e-05, "loss": 0.0009, "step": 34500 }, { "epoch": 0.25, "learning_rate": 2.926808418355768e-05, "loss": 0.0009, "step": 35000 }, { "epoch": 0.25, "learning_rate": 2.9230259076764795e-05, "loss": 0.0009, "step": 35500 }, { "epoch": 0.26, "learning_rate": 2.9192509620185496e-05, "loss": 0.0008, "step": 36000 }, { "epoch": 0.26, "learning_rate": 2.915468451339261e-05, "loss": 0.0008, "step": 36500 }, { "epoch": 0.26, "learning_rate": 2.9116859406599725e-05, "loss": 0.0008, "step": 37000 }, { "epoch": 0.27, "learning_rate": 2.907903429980684e-05, "loss": 0.0009, "step": 37500 }, { "epoch": 0.27, "learning_rate": 2.9041209193013954e-05, "loss": 0.0008, "step": 38000 }, { "epoch": 0.27, "learning_rate": 2.9003384086221073e-05, "loss": 0.0009, "step": 38500 }, { "epoch": 0.28, "learning_rate": 2.8965558979428186e-05, "loss": 0.0008, "step": 39000 }, { "epoch": 0.28, "learning_rate": 2.8927733872635302e-05, "loss": 0.0008, "step": 39500 }, { "epoch": 0.28, "learning_rate": 2.8889984416056e-05, "loss": 0.0009, "step": 40000 }, { "epoch": 0.29, "learning_rate": 2.88522349594767e-05, "loss": 0.0008, "step": 40500 }, { "epoch": 0.29, "learning_rate": 2.881440985268382e-05, "loss": 0.0008, "step": 41000 }, { "epoch": 0.3, "learning_rate": 2.8776584745890933e-05, "loss": 0.0008, "step": 41500 }, { "epoch": 0.3, "learning_rate": 2.873875963909805e-05, "loss": 0.0008, "step": 42000 }, { "epoch": 0.3, "learning_rate": 2.870101018251875e-05, "loss": 0.0008, "step": 42500 }, { "epoch": 0.31, "learning_rate": 2.8663185075725863e-05, "loss": 0.0008, "step": 43000 }, { "epoch": 0.31, "learning_rate": 2.862535996893298e-05, "loss": 0.0008, "step": 43500 }, { "epoch": 0.31, "learning_rate": 2.8587534862140095e-05, "loss": 0.0008, "step": 44000 }, { "epoch": 0.32, "learning_rate": 2.8549709755347208e-05, "loss": 0.0008, "step": 44500 }, { "epoch": 0.32, "learning_rate": 2.8511884648554327e-05, "loss": 0.0008, "step": 45000 }, { "epoch": 0.32, "learning_rate": 2.847405954176144e-05, "loss": 0.0008, "step": 45500 }, { "epoch": 0.33, "learning_rate": 2.8436234434968556e-05, "loss": 0.0008, "step": 46000 }, { "epoch": 0.33, "learning_rate": 2.8398484978389257e-05, "loss": 0.0008, "step": 46500 }, { "epoch": 0.33, "learning_rate": 2.8360659871596373e-05, "loss": 0.0008, "step": 47000 }, { "epoch": 0.34, "learning_rate": 2.8322834764803486e-05, "loss": 0.0008, "step": 47500 }, { "epoch": 0.34, "learning_rate": 2.8285009658010602e-05, "loss": 0.0008, "step": 48000 }, { "epoch": 0.34, "learning_rate": 2.8247184551217718e-05, "loss": 0.0008, "step": 48500 }, { "epoch": 0.35, "learning_rate": 2.820943509463842e-05, "loss": 0.0008, "step": 49000 }, { "epoch": 0.35, "learning_rate": 2.8171609987845535e-05, "loss": 0.0008, "step": 49500 }, { "epoch": 0.36, "learning_rate": 2.8133784881052648e-05, "loss": 0.0007, "step": 50000 }, { "epoch": 0.36, "learning_rate": 2.809595977425976e-05, "loss": 0.0008, "step": 50500 }, { "epoch": 0.36, "learning_rate": 2.805821031768046e-05, "loss": 0.0008, "step": 51000 }, { "epoch": 0.37, "learning_rate": 2.802038521088758e-05, "loss": 0.0007, "step": 51500 }, { "epoch": 0.37, "learning_rate": 2.7982560104094694e-05, "loss": 0.0007, "step": 52000 }, { "epoch": 0.37, "learning_rate": 2.794473499730181e-05, "loss": 0.0008, "step": 52500 }, { "epoch": 0.38, "learning_rate": 2.790698554072251e-05, "loss": 0.0007, "step": 53000 }, { "epoch": 0.38, "learning_rate": 2.7869160433929627e-05, "loss": 0.0007, "step": 53500 }, { "epoch": 0.38, "learning_rate": 2.783133532713674e-05, "loss": 0.0007, "step": 54000 }, { "epoch": 0.39, "learning_rate": 2.7793510220343856e-05, "loss": 0.0007, "step": 54500 }, { "epoch": 0.39, "learning_rate": 2.7755760763764557e-05, "loss": 0.0008, "step": 55000 }, { "epoch": 0.39, "learning_rate": 2.7717935656971673e-05, "loss": 0.0008, "step": 55500 }, { "epoch": 0.4, "learning_rate": 2.768011055017879e-05, "loss": 0.0007, "step": 56000 }, { "epoch": 0.4, "learning_rate": 2.7642285443385902e-05, "loss": 0.0007, "step": 56500 }, { "epoch": 0.41, "learning_rate": 2.7604535986806603e-05, "loss": 0.0007, "step": 57000 }, { "epoch": 0.41, "learning_rate": 2.756671088001372e-05, "loss": 0.0007, "step": 57500 }, { "epoch": 0.41, "learning_rate": 2.7528885773220835e-05, "loss": 0.0007, "step": 58000 }, { "epoch": 0.42, "learning_rate": 2.7491060666427948e-05, "loss": 0.0007, "step": 58500 }, { "epoch": 0.42, "learning_rate": 2.745331120984865e-05, "loss": 0.0007, "step": 59000 }, { "epoch": 0.42, "learning_rate": 2.7415486103055765e-05, "loss": 0.0007, "step": 59500 }, { "epoch": 0.43, "learning_rate": 2.737766099626288e-05, "loss": 0.0007, "step": 60000 }, { "epoch": 0.43, "learning_rate": 2.7339835889469994e-05, "loss": 0.0007, "step": 60500 }, { "epoch": 0.43, "learning_rate": 2.7302086432890695e-05, "loss": 0.0007, "step": 61000 }, { "epoch": 0.44, "learning_rate": 2.726426132609781e-05, "loss": 0.0007, "step": 61500 }, { "epoch": 0.44, "learning_rate": 2.7226436219304927e-05, "loss": 0.0007, "step": 62000 }, { "epoch": 0.44, "learning_rate": 2.7188611112512043e-05, "loss": 0.0007, "step": 62500 }, { "epoch": 0.45, "learning_rate": 2.7150861655932744e-05, "loss": 0.0007, "step": 63000 }, { "epoch": 0.45, "learning_rate": 2.7113036549139857e-05, "loss": 0.0007, "step": 63500 }, { "epoch": 0.46, "learning_rate": 2.7075211442346973e-05, "loss": 0.0007, "step": 64000 }, { "epoch": 0.46, "learning_rate": 2.703738633555409e-05, "loss": 0.0007, "step": 64500 }, { "epoch": 0.46, "learning_rate": 2.699963687897479e-05, "loss": 0.0007, "step": 65000 }, { "epoch": 0.47, "learning_rate": 2.6961811772181903e-05, "loss": 0.0007, "step": 65500 }, { "epoch": 0.47, "learning_rate": 2.692398666538902e-05, "loss": 0.0007, "step": 66000 }, { "epoch": 0.47, "learning_rate": 2.6886161558596135e-05, "loss": 0.0007, "step": 66500 }, { "epoch": 0.48, "learning_rate": 2.6848412102016836e-05, "loss": 0.0007, "step": 67000 }, { "epoch": 0.48, "learning_rate": 2.681058699522395e-05, "loss": 0.0007, "step": 67500 }, { "epoch": 0.48, "learning_rate": 2.6772761888431065e-05, "loss": 0.0007, "step": 68000 }, { "epoch": 0.49, "learning_rate": 2.673493678163818e-05, "loss": 0.0007, "step": 68500 }, { "epoch": 0.49, "learning_rate": 2.6697187325058882e-05, "loss": 0.0007, "step": 69000 }, { "epoch": 0.49, "learning_rate": 2.6659362218265998e-05, "loss": 0.0007, "step": 69500 }, { "epoch": 0.5, "learning_rate": 2.662153711147311e-05, "loss": 0.0007, "step": 70000 }, { "epoch": 0.5, "learning_rate": 2.6583712004680227e-05, "loss": 0.0007, "step": 70500 }, { "epoch": 0.5, "learning_rate": 2.6545962548100928e-05, "loss": 0.0007, "step": 71000 }, { "epoch": 0.51, "learning_rate": 2.6508137441308044e-05, "loss": 0.0007, "step": 71500 }, { "epoch": 0.51, "learning_rate": 2.6470312334515157e-05, "loss": 0.0007, "step": 72000 }, { "epoch": 0.52, "learning_rate": 2.6432487227722276e-05, "loss": 0.0007, "step": 72500 }, { "epoch": 0.52, "learning_rate": 2.6394737771142977e-05, "loss": 0.0007, "step": 73000 }, { "epoch": 0.52, "learning_rate": 2.635691266435009e-05, "loss": 0.0007, "step": 73500 }, { "epoch": 0.53, "learning_rate": 2.6319087557557203e-05, "loss": 0.0007, "step": 74000 }, { "epoch": 0.53, "learning_rate": 2.628126245076432e-05, "loss": 0.0007, "step": 74500 }, { "epoch": 0.53, "learning_rate": 2.624351299418502e-05, "loss": 0.0007, "step": 75000 }, { "epoch": 0.54, "learning_rate": 2.6205687887392136e-05, "loss": 0.0007, "step": 75500 }, { "epoch": 0.54, "learning_rate": 2.6167862780599252e-05, "loss": 0.0007, "step": 76000 }, { "epoch": 0.54, "learning_rate": 2.6130037673806365e-05, "loss": 0.0007, "step": 76500 }, { "epoch": 0.55, "learning_rate": 2.6092288217227066e-05, "loss": 0.0007, "step": 77000 }, { "epoch": 0.55, "learning_rate": 2.6054463110434182e-05, "loss": 0.0007, "step": 77500 }, { "epoch": 0.55, "learning_rate": 2.6016638003641298e-05, "loss": 0.0006, "step": 78000 }, { "epoch": 0.56, "learning_rate": 2.597881289684841e-05, "loss": 0.0007, "step": 78500 }, { "epoch": 0.56, "learning_rate": 2.5941063440269112e-05, "loss": 0.0006, "step": 79000 }, { "epoch": 0.57, "learning_rate": 2.590323833347623e-05, "loss": 0.0007, "step": 79500 }, { "epoch": 0.57, "learning_rate": 2.5865413226683344e-05, "loss": 0.0006, "step": 80000 }, { "epoch": 0.57, "learning_rate": 2.582758811989046e-05, "loss": 0.0006, "step": 80500 }, { "epoch": 0.58, "learning_rate": 2.5789838663311158e-05, "loss": 0.0007, "step": 81000 }, { "epoch": 0.58, "learning_rate": 2.5752013556518277e-05, "loss": 0.0007, "step": 81500 }, { "epoch": 0.58, "learning_rate": 2.571418844972539e-05, "loss": 0.0006, "step": 82000 }, { "epoch": 0.59, "learning_rate": 2.5676363342932506e-05, "loss": 0.0006, "step": 82500 }, { "epoch": 0.59, "learning_rate": 2.5638613886353207e-05, "loss": 0.0006, "step": 83000 }, { "epoch": 0.59, "learning_rate": 2.560078877956032e-05, "loss": 0.0006, "step": 83500 }, { "epoch": 0.6, "learning_rate": 2.556296367276744e-05, "loss": 0.0006, "step": 84000 }, { "epoch": 0.6, "learning_rate": 2.5525138565974552e-05, "loss": 0.0006, "step": 84500 }, { "epoch": 0.6, "learning_rate": 2.5487313459181665e-05, "loss": 0.0006, "step": 85000 }, { "epoch": 0.61, "learning_rate": 2.5449564002602366e-05, "loss": 0.0006, "step": 85500 }, { "epoch": 0.61, "learning_rate": 2.5411738895809485e-05, "loss": 0.0006, "step": 86000 }, { "epoch": 0.62, "learning_rate": 2.5373913789016598e-05, "loss": 0.0006, "step": 86500 }, { "epoch": 0.62, "learning_rate": 2.5336088682223714e-05, "loss": 0.0007, "step": 87000 }, { "epoch": 0.62, "learning_rate": 2.5298339225644415e-05, "loss": 0.0007, "step": 87500 }, { "epoch": 0.63, "learning_rate": 2.526051411885153e-05, "loss": 0.0007, "step": 88000 }, { "epoch": 0.63, "learning_rate": 2.5222689012058644e-05, "loss": 0.0006, "step": 88500 }, { "epoch": 0.63, "learning_rate": 2.518486390526576e-05, "loss": 0.0006, "step": 89000 }, { "epoch": 0.64, "learning_rate": 2.514711444868646e-05, "loss": 0.0006, "step": 89500 }, { "epoch": 0.64, "learning_rate": 2.5109289341893574e-05, "loss": 0.0006, "step": 90000 }, { "epoch": 0.64, "learning_rate": 2.5071464235100693e-05, "loss": 0.0006, "step": 90500 }, { "epoch": 0.65, "learning_rate": 2.5033639128307806e-05, "loss": 0.0006, "step": 91000 }, { "epoch": 0.65, "learning_rate": 2.4995889671728507e-05, "loss": 0.0006, "step": 91500 }, { "epoch": 0.65, "learning_rate": 2.495806456493562e-05, "loss": 0.0006, "step": 92000 }, { "epoch": 0.66, "learning_rate": 2.492023945814274e-05, "loss": 0.0006, "step": 92500 }, { "epoch": 0.66, "learning_rate": 2.4882414351349852e-05, "loss": 0.0006, "step": 93000 }, { "epoch": 0.66, "learning_rate": 2.4844664894770553e-05, "loss": 0.0006, "step": 93500 }, { "epoch": 0.67, "learning_rate": 2.480683978797767e-05, "loss": 0.0006, "step": 94000 }, { "epoch": 0.67, "learning_rate": 2.4769014681184785e-05, "loss": 0.0006, "step": 94500 }, { "epoch": 0.68, "learning_rate": 2.4731189574391898e-05, "loss": 0.0006, "step": 95000 }, { "epoch": 0.68, "learning_rate": 2.46934401178126e-05, "loss": 0.0006, "step": 95500 }, { "epoch": 0.68, "learning_rate": 2.4655615011019715e-05, "loss": 0.0006, "step": 96000 }, { "epoch": 0.69, "learning_rate": 2.461778990422683e-05, "loss": 0.0006, "step": 96500 }, { "epoch": 0.69, "learning_rate": 2.4579964797433947e-05, "loss": 0.0006, "step": 97000 }, { "epoch": 0.69, "learning_rate": 2.4542215340854648e-05, "loss": 0.0006, "step": 97500 }, { "epoch": 0.7, "learning_rate": 2.450439023406176e-05, "loss": 0.0006, "step": 98000 }, { "epoch": 0.7, "learning_rate": 2.4466565127268874e-05, "loss": 0.0006, "step": 98500 }, { "epoch": 0.7, "learning_rate": 2.4428740020475993e-05, "loss": 0.0006, "step": 99000 }, { "epoch": 0.71, "learning_rate": 2.4390990563896694e-05, "loss": 0.0006, "step": 99500 }, { "epoch": 0.71, "learning_rate": 2.4353165457103807e-05, "loss": 0.0006, "step": 100000 }, { "epoch": 0.71, "learning_rate": 2.4315340350310923e-05, "loss": 0.0006, "step": 100500 }, { "epoch": 0.72, "learning_rate": 2.427751524351804e-05, "loss": 0.0006, "step": 101000 }, { "epoch": 0.72, "learning_rate": 2.423976578693874e-05, "loss": 0.0006, "step": 101500 }, { "epoch": 0.73, "learning_rate": 2.4201940680145853e-05, "loss": 0.0006, "step": 102000 }, { "epoch": 0.73, "learning_rate": 2.416411557335297e-05, "loss": 0.0006, "step": 102500 }, { "epoch": 0.73, "learning_rate": 2.4126290466560085e-05, "loss": 0.0006, "step": 103000 }, { "epoch": 0.74, "learning_rate": 2.4088541009980786e-05, "loss": 0.0006, "step": 103500 }, { "epoch": 0.74, "learning_rate": 2.4050715903187902e-05, "loss": 0.0006, "step": 104000 }, { "epoch": 0.74, "learning_rate": 2.4012890796395015e-05, "loss": 0.0006, "step": 104500 }, { "epoch": 0.75, "learning_rate": 2.397506568960213e-05, "loss": 0.0006, "step": 105000 }, { "epoch": 0.75, "learning_rate": 2.3937316233022832e-05, "loss": 0.0006, "step": 105500 }, { "epoch": 0.75, "learning_rate": 2.3899491126229948e-05, "loss": 0.0006, "step": 106000 }, { "epoch": 0.76, "learning_rate": 2.386166601943706e-05, "loss": 0.0006, "step": 106500 }, { "epoch": 0.76, "learning_rate": 2.3823840912644177e-05, "loss": 0.0006, "step": 107000 }, { "epoch": 0.76, "learning_rate": 2.3786015805851293e-05, "loss": 0.0006, "step": 107500 }, { "epoch": 0.77, "learning_rate": 2.3748266349271994e-05, "loss": 0.0006, "step": 108000 }, { "epoch": 0.77, "learning_rate": 2.3710441242479107e-05, "loss": 0.0006, "step": 108500 }, { "epoch": 0.78, "learning_rate": 2.3672616135686223e-05, "loss": 0.0006, "step": 109000 }, { "epoch": 0.78, "learning_rate": 2.363479102889334e-05, "loss": 0.0006, "step": 109500 }, { "epoch": 0.78, "learning_rate": 2.359704157231404e-05, "loss": 0.0006, "step": 110000 }, { "epoch": 0.79, "learning_rate": 2.3559216465521156e-05, "loss": 0.0006, "step": 110500 }, { "epoch": 0.79, "learning_rate": 2.352139135872827e-05, "loss": 0.0006, "step": 111000 }, { "epoch": 0.79, "learning_rate": 2.3483566251935388e-05, "loss": 0.0006, "step": 111500 }, { "epoch": 0.8, "learning_rate": 2.3445816795356086e-05, "loss": 0.0006, "step": 112000 }, { "epoch": 0.8, "learning_rate": 2.3407991688563202e-05, "loss": 0.0006, "step": 112500 }, { "epoch": 0.8, "learning_rate": 2.3370166581770315e-05, "loss": 0.0006, "step": 113000 }, { "epoch": 0.81, "learning_rate": 2.333234147497743e-05, "loss": 0.0006, "step": 113500 }, { "epoch": 0.81, "learning_rate": 2.3294592018398132e-05, "loss": 0.0006, "step": 114000 }, { "epoch": 0.81, "learning_rate": 2.3256766911605248e-05, "loss": 0.0006, "step": 114500 }, { "epoch": 0.82, "learning_rate": 2.3218941804812364e-05, "loss": 0.0006, "step": 115000 }, { "epoch": 0.82, "learning_rate": 2.3181116698019477e-05, "loss": 0.0006, "step": 115500 }, { "epoch": 0.82, "learning_rate": 2.3143367241440178e-05, "loss": 0.0006, "step": 116000 }, { "epoch": 0.83, "learning_rate": 2.3105542134647294e-05, "loss": 0.0006, "step": 116500 }, { "epoch": 0.83, "learning_rate": 2.306771702785441e-05, "loss": 0.0006, "step": 117000 }, { "epoch": 0.84, "learning_rate": 2.3029891921061523e-05, "loss": 0.0006, "step": 117500 }, { "epoch": 0.84, "learning_rate": 2.2992142464482224e-05, "loss": 0.0006, "step": 118000 }, { "epoch": 0.84, "learning_rate": 2.2954317357689343e-05, "loss": 0.0006, "step": 118500 }, { "epoch": 0.85, "learning_rate": 2.2916492250896456e-05, "loss": 0.0006, "step": 119000 }, { "epoch": 0.85, "learning_rate": 2.287866714410357e-05, "loss": 0.0006, "step": 119500 }, { "epoch": 0.85, "learning_rate": 2.2840842037310688e-05, "loss": 0.0006, "step": 120000 }, { "epoch": 0.86, "learning_rate": 2.280309258073139e-05, "loss": 0.0006, "step": 120500 }, { "epoch": 0.86, "learning_rate": 2.2765267473938502e-05, "loss": 0.0006, "step": 121000 }, { "epoch": 0.86, "learning_rate": 2.2727442367145618e-05, "loss": 0.0006, "step": 121500 }, { "epoch": 0.87, "learning_rate": 2.268961726035273e-05, "loss": 0.0005, "step": 122000 }, { "epoch": 0.87, "learning_rate": 2.2651867803773432e-05, "loss": 0.0006, "step": 122500 }, { "epoch": 0.87, "learning_rate": 2.2614042696980548e-05, "loss": 0.0006, "step": 123000 }, { "epoch": 0.88, "learning_rate": 2.2576217590187664e-05, "loss": 0.0006, "step": 123500 }, { "epoch": 0.88, "learning_rate": 2.2538392483394777e-05, "loss": 0.0006, "step": 124000 }, { "epoch": 0.89, "learning_rate": 2.2500643026815478e-05, "loss": 0.0006, "step": 124500 }, { "epoch": 0.89, "learning_rate": 2.2462817920022597e-05, "loss": 0.0006, "step": 125000 }, { "epoch": 0.89, "learning_rate": 2.242499281322971e-05, "loss": 0.0006, "step": 125500 }, { "epoch": 0.9, "learning_rate": 2.2387167706436823e-05, "loss": 0.0006, "step": 126000 }, { "epoch": 0.9, "learning_rate": 2.2349418249857524e-05, "loss": 0.0006, "step": 126500 }, { "epoch": 0.9, "learning_rate": 2.2311593143064643e-05, "loss": 0.0006, "step": 127000 }, { "epoch": 0.91, "learning_rate": 2.2273768036271756e-05, "loss": 0.0006, "step": 127500 }, { "epoch": 0.91, "learning_rate": 2.2235942929478872e-05, "loss": 0.0006, "step": 128000 }, { "epoch": 0.91, "learning_rate": 2.2198193472899573e-05, "loss": 0.0006, "step": 128500 }, { "epoch": 0.92, "learning_rate": 2.216036836610669e-05, "loss": 0.0006, "step": 129000 }, { "epoch": 0.92, "learning_rate": 2.2122543259313802e-05, "loss": 0.0006, "step": 129500 }, { "epoch": 0.92, "learning_rate": 2.2084718152520918e-05, "loss": 0.0005, "step": 130000 }, { "epoch": 0.93, "learning_rate": 2.204696869594162e-05, "loss": 0.0006, "step": 130500 }, { "epoch": 0.93, "learning_rate": 2.2009143589148732e-05, "loss": 0.0006, "step": 131000 }, { "epoch": 0.94, "learning_rate": 2.197131848235585e-05, "loss": 0.0006, "step": 131500 }, { "epoch": 0.94, "learning_rate": 2.1933493375562964e-05, "loss": 0.0006, "step": 132000 }, { "epoch": 0.94, "learning_rate": 2.1895668268770077e-05, "loss": 0.0006, "step": 132500 }, { "epoch": 0.95, "learning_rate": 2.1857918812190778e-05, "loss": 0.0006, "step": 133000 }, { "epoch": 0.95, "learning_rate": 2.1820093705397897e-05, "loss": 0.0006, "step": 133500 }, { "epoch": 0.95, "learning_rate": 2.178226859860501e-05, "loss": 0.0006, "step": 134000 }, { "epoch": 0.96, "learning_rate": 2.1744443491812126e-05, "loss": 0.0006, "step": 134500 }, { "epoch": 0.96, "learning_rate": 2.1706694035232827e-05, "loss": 0.0005, "step": 135000 }, { "epoch": 0.96, "learning_rate": 2.1668868928439943e-05, "loss": 0.0006, "step": 135500 }, { "epoch": 0.97, "learning_rate": 2.1631043821647056e-05, "loss": 0.0006, "step": 136000 }, { "epoch": 0.97, "learning_rate": 2.1593218714854172e-05, "loss": 0.0006, "step": 136500 }, { "epoch": 0.97, "learning_rate": 2.1555393608061285e-05, "loss": 0.0006, "step": 137000 }, { "epoch": 0.98, "learning_rate": 2.1517644151481986e-05, "loss": 0.0006, "step": 137500 }, { "epoch": 0.98, "learning_rate": 2.1479819044689105e-05, "loss": 0.0006, "step": 138000 }, { "epoch": 0.98, "learning_rate": 2.1441993937896218e-05, "loss": 0.0005, "step": 138500 }, { "epoch": 0.99, "learning_rate": 2.1404168831103334e-05, "loss": 0.0006, "step": 139000 }, { "epoch": 0.99, "learning_rate": 2.136641937452403e-05, "loss": 0.0006, "step": 139500 }, { "epoch": 1.0, "learning_rate": 2.132859426773115e-05, "loss": 0.0005, "step": 140000 }, { "epoch": 1.0, "learning_rate": 2.1290769160938264e-05, "loss": 0.0005, "step": 140500 }, { "epoch": 1.0, "learning_rate": 2.125294405414538e-05, "loss": 0.0005, "step": 141000 }, { "epoch": 1.01, "learning_rate": 2.121519459756608e-05, "loss": 0.0005, "step": 141500 }, { "epoch": 1.01, "learning_rate": 2.1177369490773197e-05, "loss": 0.0005, "step": 142000 }, { "epoch": 1.01, "learning_rate": 2.1139544383980313e-05, "loss": 0.0005, "step": 142500 }, { "epoch": 1.02, "learning_rate": 2.1101719277187426e-05, "loss": 0.0005, "step": 143000 }, { "epoch": 1.02, "learning_rate": 2.1063969820608127e-05, "loss": 0.0005, "step": 143500 }, { "epoch": 1.02, "learning_rate": 2.1026144713815243e-05, "loss": 0.0005, "step": 144000 }, { "epoch": 1.03, "learning_rate": 2.098831960702236e-05, "loss": 0.0005, "step": 144500 }, { "epoch": 1.03, "learning_rate": 2.0950494500229472e-05, "loss": 0.0005, "step": 145000 }, { "epoch": 1.03, "learning_rate": 2.0912669393436588e-05, "loss": 0.0005, "step": 145500 }, { "epoch": 1.04, "learning_rate": 2.0874844286643704e-05, "loss": 0.0005, "step": 146000 }, { "epoch": 1.04, "learning_rate": 2.0837094830064405e-05, "loss": 0.0005, "step": 146500 }, { "epoch": 1.05, "learning_rate": 2.0799269723271518e-05, "loss": 0.0005, "step": 147000 }, { "epoch": 1.05, "learning_rate": 2.0761444616478634e-05, "loss": 0.0005, "step": 147500 }, { "epoch": 1.05, "learning_rate": 2.072361950968575e-05, "loss": 0.0005, "step": 148000 }, { "epoch": 1.06, "learning_rate": 2.068587005310645e-05, "loss": 0.0005, "step": 148500 }, { "epoch": 1.06, "learning_rate": 2.0648044946313567e-05, "loss": 0.0005, "step": 149000 }, { "epoch": 1.06, "learning_rate": 2.061021983952068e-05, "loss": 0.0005, "step": 149500 }, { "epoch": 1.07, "learning_rate": 2.0572394732727796e-05, "loss": 0.0005, "step": 150000 }, { "epoch": 1.07, "learning_rate": 2.0534645276148497e-05, "loss": 0.0005, "step": 150500 }, { "epoch": 1.07, "learning_rate": 2.0496820169355613e-05, "loss": 0.0005, "step": 151000 }, { "epoch": 1.08, "learning_rate": 2.0458995062562726e-05, "loss": 0.0005, "step": 151500 }, { "epoch": 1.08, "learning_rate": 2.0421169955769845e-05, "loss": 0.0005, "step": 152000 }, { "epoch": 1.08, "learning_rate": 2.0383344848976958e-05, "loss": 0.0005, "step": 152500 }, { "epoch": 1.09, "learning_rate": 2.034559539239766e-05, "loss": 0.0005, "step": 153000 }, { "epoch": 1.09, "learning_rate": 2.030777028560477e-05, "loss": 0.0005, "step": 153500 }, { "epoch": 1.1, "learning_rate": 2.0269945178811888e-05, "loss": 0.0005, "step": 154000 }, { "epoch": 1.1, "learning_rate": 2.0232120072019004e-05, "loss": 0.0005, "step": 154500 }, { "epoch": 1.1, "learning_rate": 2.0194370615439705e-05, "loss": 0.0005, "step": 155000 }, { "epoch": 1.11, "learning_rate": 2.015654550864682e-05, "loss": 0.0005, "step": 155500 }, { "epoch": 1.11, "learning_rate": 2.0118720401853934e-05, "loss": 0.0005, "step": 156000 }, { "epoch": 1.11, "learning_rate": 2.008089529506105e-05, "loss": 0.0005, "step": 156500 }, { "epoch": 1.12, "learning_rate": 2.004314583848175e-05, "loss": 0.0005, "step": 157000 }, { "epoch": 1.12, "learning_rate": 2.0005320731688867e-05, "loss": 0.0005, "step": 157500 }, { "epoch": 1.12, "learning_rate": 1.996749562489598e-05, "loss": 0.0005, "step": 158000 }, { "epoch": 1.13, "learning_rate": 1.99296705181031e-05, "loss": 0.0005, "step": 158500 }, { "epoch": 1.13, "learning_rate": 1.9891845411310212e-05, "loss": 0.0005, "step": 159000 }, { "epoch": 1.13, "learning_rate": 1.9854095954730913e-05, "loss": 0.0005, "step": 159500 }, { "epoch": 1.14, "learning_rate": 1.9816270847938026e-05, "loss": 0.0005, "step": 160000 }, { "epoch": 1.14, "learning_rate": 1.977844574114514e-05, "loss": 0.0005, "step": 160500 }, { "epoch": 1.14, "learning_rate": 1.9740620634352258e-05, "loss": 0.0005, "step": 161000 }, { "epoch": 1.15, "learning_rate": 1.970287117777296e-05, "loss": 0.0005, "step": 161500 }, { "epoch": 1.15, "learning_rate": 1.9665046070980075e-05, "loss": 0.0005, "step": 162000 }, { "epoch": 1.16, "learning_rate": 1.9627220964187188e-05, "loss": 0.0005, "step": 162500 }, { "epoch": 1.16, "learning_rate": 1.9589395857394307e-05, "loss": 0.0005, "step": 163000 }, { "epoch": 1.16, "learning_rate": 1.9551646400815005e-05, "loss": 0.0005, "step": 163500 }, { "epoch": 1.17, "learning_rate": 1.951382129402212e-05, "loss": 0.0005, "step": 164000 }, { "epoch": 1.17, "learning_rate": 1.9475996187229234e-05, "loss": 0.0005, "step": 164500 }, { "epoch": 1.17, "learning_rate": 1.9438171080436353e-05, "loss": 0.0005, "step": 165000 }, { "epoch": 1.18, "learning_rate": 1.9400345973643466e-05, "loss": 0.0005, "step": 165500 }, { "epoch": 1.18, "learning_rate": 1.9362596517064167e-05, "loss": 0.0005, "step": 166000 }, { "epoch": 1.18, "learning_rate": 1.9324771410271283e-05, "loss": 0.0005, "step": 166500 }, { "epoch": 1.19, "learning_rate": 1.92869463034784e-05, "loss": 0.0005, "step": 167000 }, { "epoch": 1.19, "learning_rate": 1.924912119668551e-05, "loss": 0.0005, "step": 167500 }, { "epoch": 1.19, "learning_rate": 1.9211296089892628e-05, "loss": 0.0005, "step": 168000 }, { "epoch": 1.2, "learning_rate": 1.917354663331333e-05, "loss": 0.0005, "step": 168500 }, { "epoch": 1.2, "learning_rate": 1.913572152652044e-05, "loss": 0.0005, "step": 169000 }, { "epoch": 1.21, "learning_rate": 1.909789641972756e-05, "loss": 0.0005, "step": 169500 }, { "epoch": 1.21, "learning_rate": 1.9060071312934674e-05, "loss": 0.0005, "step": 170000 }, { "epoch": 1.21, "learning_rate": 1.9022321856355375e-05, "loss": 0.0005, "step": 170500 }, { "epoch": 1.22, "learning_rate": 1.8984496749562488e-05, "loss": 0.0005, "step": 171000 }, { "epoch": 1.22, "learning_rate": 1.8946671642769607e-05, "loss": 0.0005, "step": 171500 }, { "epoch": 1.22, "learning_rate": 1.890884653597672e-05, "loss": 0.0005, "step": 172000 }, { "epoch": 1.23, "learning_rate": 1.8871021429183836e-05, "loss": 0.0005, "step": 172500 }, { "epoch": 1.23, "learning_rate": 1.8833271972604537e-05, "loss": 0.0005, "step": 173000 }, { "epoch": 1.23, "learning_rate": 1.8795446865811653e-05, "loss": 0.0005, "step": 173500 }, { "epoch": 1.24, "learning_rate": 1.8757621759018766e-05, "loss": 0.0005, "step": 174000 }, { "epoch": 1.24, "learning_rate": 1.8719796652225882e-05, "loss": 0.0005, "step": 174500 }, { "epoch": 1.24, "learning_rate": 1.8682047195646583e-05, "loss": 0.0005, "step": 175000 }, { "epoch": 1.25, "learning_rate": 1.86442220888537e-05, "loss": 0.0005, "step": 175500 }, { "epoch": 1.25, "learning_rate": 1.8606396982060815e-05, "loss": 0.0005, "step": 176000 }, { "epoch": 1.26, "learning_rate": 1.8568571875267928e-05, "loss": 0.0005, "step": 176500 }, { "epoch": 1.26, "learning_rate": 1.853074676847504e-05, "loss": 0.0005, "step": 177000 }, { "epoch": 1.26, "learning_rate": 1.849299731189574e-05, "loss": 0.0005, "step": 177500 }, { "epoch": 1.27, "learning_rate": 1.845517220510286e-05, "loss": 0.0005, "step": 178000 }, { "epoch": 1.27, "learning_rate": 1.8417347098309974e-05, "loss": 0.0005, "step": 178500 }, { "epoch": 1.27, "learning_rate": 1.837952199151709e-05, "loss": 0.0005, "step": 179000 }, { "epoch": 1.28, "learning_rate": 1.834177253493779e-05, "loss": 0.0005, "step": 179500 }, { "epoch": 1.28, "learning_rate": 1.8303947428144907e-05, "loss": 0.0005, "step": 180000 }, { "epoch": 1.28, "learning_rate": 1.826612232135202e-05, "loss": 0.0005, "step": 180500 }, { "epoch": 1.29, "learning_rate": 1.8228297214559136e-05, "loss": 0.0005, "step": 181000 }, { "epoch": 1.29, "learning_rate": 1.8190472107766252e-05, "loss": 0.0005, "step": 181500 }, { "epoch": 1.29, "learning_rate": 1.8152722651186953e-05, "loss": 0.0005, "step": 182000 }, { "epoch": 1.3, "learning_rate": 1.811489754439407e-05, "loss": 0.0005, "step": 182500 }, { "epoch": 1.3, "learning_rate": 1.807707243760118e-05, "loss": 0.0005, "step": 183000 }, { "epoch": 1.3, "learning_rate": 1.8039247330808298e-05, "loss": 0.0005, "step": 183500 }, { "epoch": 1.31, "learning_rate": 1.8001422224015414e-05, "loss": 0.0005, "step": 184000 }, { "epoch": 1.31, "learning_rate": 1.7963672767436115e-05, "loss": 0.0005, "step": 184500 }, { "epoch": 1.32, "learning_rate": 1.7925847660643228e-05, "loss": 0.0005, "step": 185000 }, { "epoch": 1.32, "learning_rate": 1.7888022553850344e-05, "loss": 0.0005, "step": 185500 }, { "epoch": 1.32, "learning_rate": 1.785019744705746e-05, "loss": 0.0005, "step": 186000 }, { "epoch": 1.33, "learning_rate": 1.781244799047816e-05, "loss": 0.0005, "step": 186500 }, { "epoch": 1.33, "learning_rate": 1.7774622883685277e-05, "loss": 0.0005, "step": 187000 }, { "epoch": 1.33, "learning_rate": 1.773679777689239e-05, "loss": 0.0005, "step": 187500 }, { "epoch": 1.34, "learning_rate": 1.7698972670099506e-05, "loss": 0.0005, "step": 188000 }, { "epoch": 1.34, "learning_rate": 1.7661147563306622e-05, "loss": 0.0005, "step": 188500 }, { "epoch": 1.34, "learning_rate": 1.7623398106727323e-05, "loss": 0.0005, "step": 189000 }, { "epoch": 1.35, "learning_rate": 1.7585572999934436e-05, "loss": 0.0005, "step": 189500 }, { "epoch": 1.35, "learning_rate": 1.7547747893141555e-05, "loss": 0.0005, "step": 190000 }, { "epoch": 1.35, "learning_rate": 1.7509922786348668e-05, "loss": 0.0005, "step": 190500 }, { "epoch": 1.36, "learning_rate": 1.747209767955578e-05, "loss": 0.0005, "step": 191000 }, { "epoch": 1.36, "learning_rate": 1.7434272572762896e-05, "loss": 0.0005, "step": 191500 }, { "epoch": 1.37, "learning_rate": 1.7396523116183598e-05, "loss": 0.0005, "step": 192000 }, { "epoch": 1.37, "learning_rate": 1.7358698009390714e-05, "loss": 0.0005, "step": 192500 }, { "epoch": 1.37, "learning_rate": 1.732087290259783e-05, "loss": 0.0005, "step": 193000 }, { "epoch": 1.38, "learning_rate": 1.7283047795804942e-05, "loss": 0.0005, "step": 193500 }, { "epoch": 1.38, "learning_rate": 1.724522268901206e-05, "loss": 0.0005, "step": 194000 }, { "epoch": 1.38, "learning_rate": 1.720747323243276e-05, "loss": 0.0005, "step": 194500 }, { "epoch": 1.39, "learning_rate": 1.7169648125639876e-05, "loss": 0.0005, "step": 195000 }, { "epoch": 1.39, "learning_rate": 1.713182301884699e-05, "loss": 0.0005, "step": 195500 }, { "epoch": 1.39, "learning_rate": 1.7093997912054108e-05, "loss": 0.0005, "step": 196000 }, { "epoch": 1.4, "learning_rate": 1.705617280526122e-05, "loss": 0.0005, "step": 196500 }, { "epoch": 1.4, "learning_rate": 1.701842334868192e-05, "loss": 0.0005, "step": 197000 }, { "epoch": 1.4, "learning_rate": 1.6980598241889034e-05, "loss": 0.0005, "step": 197500 }, { "epoch": 1.41, "learning_rate": 1.6942773135096154e-05, "loss": 0.0005, "step": 198000 }, { "epoch": 1.41, "learning_rate": 1.6904948028303266e-05, "loss": 0.0005, "step": 198500 }, { "epoch": 1.42, "learning_rate": 1.6867122921510383e-05, "loss": 0.0005, "step": 199000 }, { "epoch": 1.42, "learning_rate": 1.6829373464931084e-05, "loss": 0.0005, "step": 199500 }, { "epoch": 1.42, "learning_rate": 1.6791548358138196e-05, "loss": 0.0005, "step": 200000 }, { "epoch": 1.43, "learning_rate": 1.6753723251345316e-05, "loss": 0.0005, "step": 200500 }, { "epoch": 1.43, "learning_rate": 1.671589814455243e-05, "loss": 0.0005, "step": 201000 }, { "epoch": 1.43, "learning_rate": 1.667814868797313e-05, "loss": 0.0005, "step": 201500 }, { "epoch": 1.44, "learning_rate": 1.6640323581180242e-05, "loss": 0.0005, "step": 202000 }, { "epoch": 1.44, "learning_rate": 1.6602498474387362e-05, "loss": 0.0005, "step": 202500 }, { "epoch": 1.44, "learning_rate": 1.6564673367594474e-05, "loss": 0.0005, "step": 203000 }, { "epoch": 1.45, "learning_rate": 1.652684826080159e-05, "loss": 0.0005, "step": 203500 }, { "epoch": 1.45, "learning_rate": 1.648909880422229e-05, "loss": 0.0005, "step": 204000 }, { "epoch": 1.45, "learning_rate": 1.6451273697429408e-05, "loss": 0.0005, "step": 204500 }, { "epoch": 1.46, "learning_rate": 1.641344859063652e-05, "loss": 0.0005, "step": 205000 }, { "epoch": 1.46, "learning_rate": 1.6375623483843636e-05, "loss": 0.0004, "step": 205500 }, { "epoch": 1.46, "learning_rate": 1.6337874027264338e-05, "loss": 0.0005, "step": 206000 }, { "epoch": 1.47, "learning_rate": 1.630004892047145e-05, "loss": 0.0005, "step": 206500 }, { "epoch": 1.47, "learning_rate": 1.626222381367857e-05, "loss": 0.0005, "step": 207000 }, { "epoch": 1.48, "learning_rate": 1.6224398706885682e-05, "loss": 0.0004, "step": 207500 }, { "epoch": 1.48, "learning_rate": 1.6186573600092795e-05, "loss": 0.0005, "step": 208000 }, { "epoch": 1.48, "learning_rate": 1.6148824143513496e-05, "loss": 0.0005, "step": 208500 }, { "epoch": 1.49, "learning_rate": 1.6110999036720616e-05, "loss": 0.0005, "step": 209000 }, { "epoch": 1.49, "learning_rate": 1.607317392992773e-05, "loss": 0.0005, "step": 209500 }, { "epoch": 1.49, "learning_rate": 1.6035348823134844e-05, "loss": 0.0005, "step": 210000 }, { "epoch": 1.5, "learning_rate": 1.5997599366555546e-05, "loss": 0.0004, "step": 210500 }, { "epoch": 1.5, "learning_rate": 1.595977425976266e-05, "loss": 0.0005, "step": 211000 }, { "epoch": 1.5, "learning_rate": 1.5921949152969774e-05, "loss": 0.0005, "step": 211500 }, { "epoch": 1.51, "learning_rate": 1.588412404617689e-05, "loss": 0.0004, "step": 212000 }, { "epoch": 1.51, "learning_rate": 1.5846298939384006e-05, "loss": 0.0004, "step": 212500 }, { "epoch": 1.51, "learning_rate": 1.5808473832591123e-05, "loss": 0.0005, "step": 213000 }, { "epoch": 1.52, "learning_rate": 1.5770724376011824e-05, "loss": 0.0005, "step": 213500 }, { "epoch": 1.52, "learning_rate": 1.5732899269218936e-05, "loss": 0.0005, "step": 214000 }, { "epoch": 1.53, "learning_rate": 1.5695074162426052e-05, "loss": 0.0005, "step": 214500 }, { "epoch": 1.53, "learning_rate": 1.565724905563317e-05, "loss": 0.0005, "step": 215000 }, { "epoch": 1.53, "learning_rate": 1.561942394884028e-05, "loss": 0.0005, "step": 215500 }, { "epoch": 1.54, "learning_rate": 1.5581674492260982e-05, "loss": 0.0005, "step": 216000 }, { "epoch": 1.54, "learning_rate": 1.55438493854681e-05, "loss": 0.0005, "step": 216500 }, { "epoch": 1.54, "learning_rate": 1.5506024278675214e-05, "loss": 0.0005, "step": 217000 }, { "epoch": 1.55, "learning_rate": 1.546819917188233e-05, "loss": 0.0004, "step": 217500 }, { "epoch": 1.55, "learning_rate": 1.5430374065089443e-05, "loss": 0.0005, "step": 218000 }, { "epoch": 1.55, "learning_rate": 1.5392624608510144e-05, "loss": 0.0005, "step": 218500 }, { "epoch": 1.56, "learning_rate": 1.535479950171726e-05, "loss": 0.0005, "step": 219000 }, { "epoch": 1.56, "learning_rate": 1.5316974394924377e-05, "loss": 0.0004, "step": 219500 }, { "epoch": 1.56, "learning_rate": 1.527914928813149e-05, "loss": 0.0005, "step": 220000 }, { "epoch": 1.57, "learning_rate": 1.524139983155219e-05, "loss": 0.0005, "step": 220500 }, { "epoch": 1.57, "learning_rate": 1.5203574724759308e-05, "loss": 0.0005, "step": 221000 }, { "epoch": 1.58, "learning_rate": 1.5165749617966422e-05, "loss": 0.0004, "step": 221500 }, { "epoch": 1.58, "learning_rate": 1.5127924511173537e-05, "loss": 0.0005, "step": 222000 }, { "epoch": 1.58, "learning_rate": 1.5090099404380651e-05, "loss": 0.0005, "step": 222500 }, { "epoch": 1.59, "learning_rate": 1.5052349947801352e-05, "loss": 0.0004, "step": 223000 }, { "epoch": 1.59, "learning_rate": 1.5014524841008468e-05, "loss": 0.0005, "step": 223500 }, { "epoch": 1.59, "learning_rate": 1.4976699734215583e-05, "loss": 0.0004, "step": 224000 }, { "epoch": 1.6, "learning_rate": 1.4938874627422699e-05, "loss": 0.0005, "step": 224500 }, { "epoch": 1.6, "learning_rate": 1.49011251708434e-05, "loss": 0.0004, "step": 225000 }, { "epoch": 1.6, "learning_rate": 1.4863300064050514e-05, "loss": 0.0005, "step": 225500 }, { "epoch": 1.61, "learning_rate": 1.482547495725763e-05, "loss": 0.0005, "step": 226000 }, { "epoch": 1.61, "learning_rate": 1.4787649850464745e-05, "loss": 0.0005, "step": 226500 }, { "epoch": 1.61, "learning_rate": 1.474982474367186e-05, "loss": 0.0005, "step": 227000 }, { "epoch": 1.62, "learning_rate": 1.471207528709256e-05, "loss": 0.0004, "step": 227500 }, { "epoch": 1.62, "learning_rate": 1.4674250180299676e-05, "loss": 0.0004, "step": 228000 }, { "epoch": 1.62, "learning_rate": 1.463642507350679e-05, "loss": 0.0004, "step": 228500 }, { "epoch": 1.63, "learning_rate": 1.4598599966713907e-05, "loss": 0.0005, "step": 229000 }, { "epoch": 1.63, "learning_rate": 1.4560774859921023e-05, "loss": 0.0004, "step": 229500 }, { "epoch": 1.64, "learning_rate": 1.4522949753128136e-05, "loss": 0.0005, "step": 230000 }, { "epoch": 1.64, "learning_rate": 1.4485200296548837e-05, "loss": 0.0005, "step": 230500 }, { "epoch": 1.64, "learning_rate": 1.4447375189755953e-05, "loss": 0.0004, "step": 231000 }, { "epoch": 1.65, "learning_rate": 1.4409550082963067e-05, "loss": 0.0005, "step": 231500 }, { "epoch": 1.65, "learning_rate": 1.4371724976170183e-05, "loss": 0.0004, "step": 232000 }, { "epoch": 1.65, "learning_rate": 1.43338998693773e-05, "loss": 0.0004, "step": 232500 }, { "epoch": 1.66, "learning_rate": 1.4296150412798e-05, "loss": 0.0004, "step": 233000 }, { "epoch": 1.66, "learning_rate": 1.4258325306005113e-05, "loss": 0.0005, "step": 233500 }, { "epoch": 1.66, "learning_rate": 1.422050019921223e-05, "loss": 0.0004, "step": 234000 }, { "epoch": 1.67, "learning_rate": 1.4182675092419344e-05, "loss": 0.0004, "step": 234500 }, { "epoch": 1.67, "learning_rate": 1.414484998562646e-05, "loss": 0.0005, "step": 235000 }, { "epoch": 1.67, "learning_rate": 1.4107024878833576e-05, "loss": 0.0004, "step": 235500 }, { "epoch": 1.68, "learning_rate": 1.4069275422254277e-05, "loss": 0.0004, "step": 236000 }, { "epoch": 1.68, "learning_rate": 1.4031450315461391e-05, "loss": 0.0004, "step": 236500 }, { "epoch": 1.69, "learning_rate": 1.3993625208668506e-05, "loss": 0.0004, "step": 237000 }, { "epoch": 1.69, "learning_rate": 1.3955800101875622e-05, "loss": 0.0004, "step": 237500 }, { "epoch": 1.69, "learning_rate": 1.3917974995082736e-05, "loss": 0.0005, "step": 238000 }, { "epoch": 1.7, "learning_rate": 1.3880149888289852e-05, "loss": 0.0004, "step": 238500 }, { "epoch": 1.7, "learning_rate": 1.3842324781496967e-05, "loss": 0.0004, "step": 239000 }, { "epoch": 1.7, "learning_rate": 1.3804499674704083e-05, "loss": 0.0004, "step": 239500 }, { "epoch": 1.71, "learning_rate": 1.3766750218124782e-05, "loss": 0.0004, "step": 240000 }, { "epoch": 1.71, "learning_rate": 1.3728925111331898e-05, "loss": 0.0004, "step": 240500 }, { "epoch": 1.71, "learning_rate": 1.36911756547526e-05, "loss": 0.0004, "step": 241000 }, { "epoch": 1.72, "learning_rate": 1.3653350547959714e-05, "loss": 0.0004, "step": 241500 }, { "epoch": 1.72, "learning_rate": 1.361552544116683e-05, "loss": 0.0004, "step": 242000 }, { "epoch": 1.72, "learning_rate": 1.3577700334373944e-05, "loss": 0.0005, "step": 242500 }, { "epoch": 1.73, "learning_rate": 1.3539950877794645e-05, "loss": 0.0004, "step": 243000 }, { "epoch": 1.73, "learning_rate": 1.350212577100176e-05, "loss": 0.0004, "step": 243500 }, { "epoch": 1.74, "learning_rate": 1.3464300664208876e-05, "loss": 0.0004, "step": 244000 }, { "epoch": 1.74, "learning_rate": 1.342647555741599e-05, "loss": 0.0004, "step": 244500 }, { "epoch": 1.74, "learning_rate": 1.3388650450623106e-05, "loss": 0.0005, "step": 245000 }, { "epoch": 1.75, "learning_rate": 1.3350900994043807e-05, "loss": 0.0004, "step": 245500 }, { "epoch": 1.75, "learning_rate": 1.3313075887250923e-05, "loss": 0.0004, "step": 246000 }, { "epoch": 1.75, "learning_rate": 1.3275250780458038e-05, "loss": 0.0004, "step": 246500 }, { "epoch": 1.76, "learning_rate": 1.3237425673665152e-05, "loss": 0.0004, "step": 247000 }, { "epoch": 1.76, "learning_rate": 1.3199600566872266e-05, "loss": 0.0004, "step": 247500 }, { "epoch": 1.76, "learning_rate": 1.3161775460079382e-05, "loss": 0.0004, "step": 248000 }, { "epoch": 1.77, "learning_rate": 1.3123950353286499e-05, "loss": 0.0004, "step": 248500 }, { "epoch": 1.77, "learning_rate": 1.30862008967072e-05, "loss": 0.0004, "step": 249000 }, { "epoch": 1.77, "learning_rate": 1.3048375789914314e-05, "loss": 0.0004, "step": 249500 }, { "epoch": 1.78, "learning_rate": 1.301055068312143e-05, "loss": 0.0004, "step": 250000 }, { "epoch": 1.78, "learning_rate": 1.2972725576328543e-05, "loss": 0.0004, "step": 250500 }, { "epoch": 1.78, "learning_rate": 1.2934900469535659e-05, "loss": 0.0004, "step": 251000 }, { "epoch": 1.79, "learning_rate": 1.289715101295636e-05, "loss": 0.0004, "step": 251500 }, { "epoch": 1.79, "learning_rate": 1.2859325906163476e-05, "loss": 0.0004, "step": 252000 }, { "epoch": 1.8, "learning_rate": 1.282150079937059e-05, "loss": 0.0004, "step": 252500 }, { "epoch": 1.8, "learning_rate": 1.2783675692577707e-05, "loss": 0.0004, "step": 253000 }, { "epoch": 1.8, "learning_rate": 1.2745850585784821e-05, "loss": 0.0004, "step": 253500 }, { "epoch": 1.81, "learning_rate": 1.2708025478991935e-05, "loss": 0.0004, "step": 254000 }, { "epoch": 1.81, "learning_rate": 1.2670200372199051e-05, "loss": 0.0004, "step": 254500 }, { "epoch": 1.81, "learning_rate": 1.2632450915619753e-05, "loss": 0.0004, "step": 255000 }, { "epoch": 1.82, "learning_rate": 1.2594625808826867e-05, "loss": 0.0004, "step": 255500 }, { "epoch": 1.82, "learning_rate": 1.2556800702033983e-05, "loss": 0.0004, "step": 256000 }, { "epoch": 1.82, "learning_rate": 1.2518975595241099e-05, "loss": 0.0004, "step": 256500 }, { "epoch": 1.83, "learning_rate": 1.2481226138661798e-05, "loss": 0.0004, "step": 257000 }, { "epoch": 1.83, "learning_rate": 1.2443401031868913e-05, "loss": 0.0004, "step": 257500 }, { "epoch": 1.83, "learning_rate": 1.2405575925076029e-05, "loss": 0.0004, "step": 258000 }, { "epoch": 1.84, "learning_rate": 1.2367750818283143e-05, "loss": 0.0004, "step": 258500 }, { "epoch": 1.84, "learning_rate": 1.232992571149026e-05, "loss": 0.0004, "step": 259000 }, { "epoch": 1.85, "learning_rate": 1.2292100604697375e-05, "loss": 0.0004, "step": 259500 }, { "epoch": 1.85, "learning_rate": 1.2254351148118077e-05, "loss": 0.0004, "step": 260000 }, { "epoch": 1.85, "learning_rate": 1.221652604132519e-05, "loss": 0.0004, "step": 260500 }, { "epoch": 1.86, "learning_rate": 1.2178700934532305e-05, "loss": 0.0004, "step": 261000 }, { "epoch": 1.86, "learning_rate": 1.214087582773942e-05, "loss": 0.0004, "step": 261500 }, { "epoch": 1.86, "learning_rate": 1.2103050720946536e-05, "loss": 0.0004, "step": 262000 }, { "epoch": 1.87, "learning_rate": 1.2065301264367237e-05, "loss": 0.0004, "step": 262500 }, { "epoch": 1.87, "learning_rate": 1.2027476157574353e-05, "loss": 0.0004, "step": 263000 }, { "epoch": 1.87, "learning_rate": 1.1989651050781467e-05, "loss": 0.0004, "step": 263500 }, { "epoch": 1.88, "learning_rate": 1.1951825943988582e-05, "loss": 0.0004, "step": 264000 }, { "epoch": 1.88, "learning_rate": 1.1914000837195696e-05, "loss": 0.0004, "step": 264500 }, { "epoch": 1.88, "learning_rate": 1.1876175730402812e-05, "loss": 0.0004, "step": 265000 }, { "epoch": 1.89, "learning_rate": 1.1838350623609928e-05, "loss": 0.0004, "step": 265500 }, { "epoch": 1.89, "learning_rate": 1.1800525516817043e-05, "loss": 0.0004, "step": 266000 }, { "epoch": 1.9, "learning_rate": 1.176285171045133e-05, "loss": 0.0004, "step": 266500 }, { "epoch": 1.9, "learning_rate": 1.1725026603658445e-05, "loss": 0.0004, "step": 267000 }, { "epoch": 1.9, "learning_rate": 1.168720149686556e-05, "loss": 0.0004, "step": 267500 }, { "epoch": 1.91, "learning_rate": 1.1649376390072675e-05, "loss": 0.0004, "step": 268000 }, { "epoch": 1.91, "learning_rate": 1.161155128327979e-05, "loss": 0.0004, "step": 268500 }, { "epoch": 1.91, "learning_rate": 1.1573726176486906e-05, "loss": 0.0004, "step": 269000 }, { "epoch": 1.92, "learning_rate": 1.153590106969402e-05, "loss": 0.0004, "step": 269500 }, { "epoch": 1.92, "learning_rate": 1.1498075962901136e-05, "loss": 0.0004, "step": 270000 }, { "epoch": 1.92, "learning_rate": 1.1460326506321837e-05, "loss": 0.0004, "step": 270500 }, { "epoch": 1.93, "learning_rate": 1.1422501399528952e-05, "loss": 0.0004, "step": 271000 }, { "epoch": 1.93, "learning_rate": 1.1384676292736066e-05, "loss": 0.0004, "step": 271500 }, { "epoch": 1.93, "learning_rate": 1.1346926836156767e-05, "loss": 0.0004, "step": 272000 }, { "epoch": 1.94, "learning_rate": 1.1309101729363883e-05, "loss": 0.0004, "step": 272500 }, { "epoch": 1.94, "learning_rate": 1.1271276622570998e-05, "loss": 0.0004, "step": 273000 }, { "epoch": 1.94, "learning_rate": 1.1233451515778114e-05, "loss": 0.0004, "step": 273500 }, { "epoch": 1.95, "learning_rate": 1.1195626408985228e-05, "loss": 0.0004, "step": 274000 }, { "epoch": 1.95, "learning_rate": 1.1157801302192343e-05, "loss": 0.0004, "step": 274500 }, { "epoch": 1.96, "learning_rate": 1.1119976195399459e-05, "loss": 0.0004, "step": 275000 }, { "epoch": 1.96, "learning_rate": 1.108222673882016e-05, "loss": 0.0004, "step": 275500 }, { "epoch": 1.96, "learning_rate": 1.1044401632027274e-05, "loss": 0.0004, "step": 276000 }, { "epoch": 1.97, "learning_rate": 1.100657652523439e-05, "loss": 0.0004, "step": 276500 }, { "epoch": 1.97, "learning_rate": 1.0968751418441506e-05, "loss": 0.0004, "step": 277000 }, { "epoch": 1.97, "learning_rate": 1.0930926311648619e-05, "loss": 0.0004, "step": 277500 }, { "epoch": 1.98, "learning_rate": 1.089317685506932e-05, "loss": 0.0004, "step": 278000 }, { "epoch": 1.98, "learning_rate": 1.0855351748276436e-05, "loss": 0.0004, "step": 278500 }, { "epoch": 1.98, "learning_rate": 1.0817526641483552e-05, "loss": 0.0004, "step": 279000 }, { "epoch": 1.99, "learning_rate": 1.0779701534690667e-05, "loss": 0.0004, "step": 279500 }, { "epoch": 1.99, "learning_rate": 1.0741876427897783e-05, "loss": 0.0004, "step": 280000 }, { "epoch": 1.99, "learning_rate": 1.0704051321104897e-05, "loss": 0.0004, "step": 280500 }, { "epoch": 2.0, "learning_rate": 1.0666226214312011e-05, "loss": 0.0004, "step": 281000 }, { "epoch": 2.0, "learning_rate": 1.0628476757732713e-05, "loss": 0.0004, "step": 281500 }, { "epoch": 2.01, "learning_rate": 1.0590651650939829e-05, "loss": 0.0004, "step": 282000 }, { "epoch": 2.01, "learning_rate": 1.0552826544146943e-05, "loss": 0.0004, "step": 282500 }, { "epoch": 2.01, "learning_rate": 1.0515001437354059e-05, "loss": 0.0004, "step": 283000 }, { "epoch": 2.02, "learning_rate": 1.0477176330561173e-05, "loss": 0.0004, "step": 283500 }, { "epoch": 2.02, "learning_rate": 1.0439351223768288e-05, "loss": 0.0004, "step": 284000 }, { "epoch": 2.02, "learning_rate": 1.0401601767188989e-05, "loss": 0.0004, "step": 284500 }, { "epoch": 2.03, "learning_rate": 1.0363776660396105e-05, "loss": 0.0004, "step": 285000 }, { "epoch": 2.03, "learning_rate": 1.032595155360322e-05, "loss": 0.0004, "step": 285500 }, { "epoch": 2.03, "learning_rate": 1.0288126446810335e-05, "loss": 0.0004, "step": 286000 }, { "epoch": 2.04, "learning_rate": 1.025030134001745e-05, "loss": 0.0004, "step": 286500 }, { "epoch": 2.04, "learning_rate": 1.0212476233224566e-05, "loss": 0.0004, "step": 287000 }, { "epoch": 2.04, "learning_rate": 1.017465112643168e-05, "loss": 0.0004, "step": 287500 }, { "epoch": 2.05, "learning_rate": 1.0136826019638795e-05, "loss": 0.0004, "step": 288000 }, { "epoch": 2.05, "learning_rate": 1.0099076563059496e-05, "loss": 0.0004, "step": 288500 }, { "epoch": 2.06, "learning_rate": 1.0061327106480197e-05, "loss": 0.0004, "step": 289000 }, { "epoch": 2.06, "learning_rate": 1.0023501999687313e-05, "loss": 0.0004, "step": 289500 }, { "epoch": 2.06, "learning_rate": 9.985676892894429e-06, "loss": 0.0004, "step": 290000 }, { "epoch": 2.07, "learning_rate": 9.947851786101543e-06, "loss": 0.0004, "step": 290500 }, { "epoch": 2.07, "learning_rate": 9.910026679308658e-06, "loss": 0.0004, "step": 291000 }, { "epoch": 2.07, "learning_rate": 9.872201572515772e-06, "loss": 0.0004, "step": 291500 }, { "epoch": 2.08, "learning_rate": 9.834376465722888e-06, "loss": 0.0004, "step": 292000 }, { "epoch": 2.08, "learning_rate": 9.796551358930004e-06, "loss": 0.0004, "step": 292500 }, { "epoch": 2.08, "learning_rate": 9.758801902350705e-06, "loss": 0.0004, "step": 293000 }, { "epoch": 2.09, "learning_rate": 9.72097679555782e-06, "loss": 0.0004, "step": 293500 }, { "epoch": 2.09, "learning_rate": 9.683151688764936e-06, "loss": 0.0004, "step": 294000 }, { "epoch": 2.09, "learning_rate": 9.645326581972049e-06, "loss": 0.0004, "step": 294500 }, { "epoch": 2.1, "learning_rate": 9.60757712539275e-06, "loss": 0.0004, "step": 295000 }, { "epoch": 2.1, "learning_rate": 9.569827668813451e-06, "loss": 0.0004, "step": 295500 }, { "epoch": 2.1, "learning_rate": 9.532002562020567e-06, "loss": 0.0004, "step": 296000 }, { "epoch": 2.11, "learning_rate": 9.494177455227683e-06, "loss": 0.0004, "step": 296500 }, { "epoch": 2.11, "learning_rate": 9.456352348434797e-06, "loss": 0.0004, "step": 297000 }, { "epoch": 2.12, "learning_rate": 9.418527241641913e-06, "loss": 0.0004, "step": 297500 }, { "epoch": 2.12, "learning_rate": 9.380702134849026e-06, "loss": 0.0004, "step": 298000 }, { "epoch": 2.12, "learning_rate": 9.342952678269727e-06, "loss": 0.0004, "step": 298500 }, { "epoch": 2.13, "learning_rate": 9.305127571476843e-06, "loss": 0.0004, "step": 299000 }, { "epoch": 2.13, "learning_rate": 9.26730246468396e-06, "loss": 0.0004, "step": 299500 }, { "epoch": 2.13, "learning_rate": 9.229477357891074e-06, "loss": 0.0004, "step": 300000 }, { "epoch": 2.14, "learning_rate": 9.19165225109819e-06, "loss": 0.0004, "step": 300500 }, { "epoch": 2.14, "learning_rate": 9.153827144305304e-06, "loss": 0.0004, "step": 301000 }, { "epoch": 2.14, "learning_rate": 9.116002037512419e-06, "loss": 0.0004, "step": 301500 }, { "epoch": 2.15, "learning_rate": 9.078176930719535e-06, "loss": 0.0004, "step": 302000 }, { "epoch": 2.15, "learning_rate": 9.040427474140236e-06, "loss": 0.0004, "step": 302500 }, { "epoch": 2.15, "learning_rate": 9.00260236734735e-06, "loss": 0.0004, "step": 303000 }, { "epoch": 2.16, "learning_rate": 8.964852910768051e-06, "loss": 0.0004, "step": 303500 }, { "epoch": 2.16, "learning_rate": 8.927027803975167e-06, "loss": 0.0004, "step": 304000 }, { "epoch": 2.17, "learning_rate": 8.889202697182282e-06, "loss": 0.0004, "step": 304500 }, { "epoch": 2.17, "learning_rate": 8.851377590389396e-06, "loss": 0.0004, "step": 305000 }, { "epoch": 2.17, "learning_rate": 8.813552483596512e-06, "loss": 0.0004, "step": 305500 }, { "epoch": 2.18, "learning_rate": 8.775727376803627e-06, "loss": 0.0004, "step": 306000 }, { "epoch": 2.18, "learning_rate": 8.737902270010743e-06, "loss": 0.0004, "step": 306500 }, { "epoch": 2.18, "learning_rate": 8.700152813431444e-06, "loss": 0.0004, "step": 307000 }, { "epoch": 2.19, "learning_rate": 8.66232770663856e-06, "loss": 0.0004, "step": 307500 }, { "epoch": 2.19, "learning_rate": 8.624502599845673e-06, "loss": 0.0004, "step": 308000 }, { "epoch": 2.19, "learning_rate": 8.586677493052789e-06, "loss": 0.0004, "step": 308500 }, { "epoch": 2.2, "learning_rate": 8.548852386259903e-06, "loss": 0.0004, "step": 309000 }, { "epoch": 2.2, "learning_rate": 8.511027279467019e-06, "loss": 0.0004, "step": 309500 }, { "epoch": 2.2, "learning_rate": 8.473202172674135e-06, "loss": 0.0004, "step": 310000 }, { "epoch": 2.21, "learning_rate": 8.43537706588125e-06, "loss": 0.0004, "step": 310500 }, { "epoch": 2.21, "learning_rate": 8.39762760930195e-06, "loss": 0.0004, "step": 311000 }, { "epoch": 2.22, "learning_rate": 8.359802502509065e-06, "loss": 0.0004, "step": 311500 }, { "epoch": 2.22, "learning_rate": 8.322053045929766e-06, "loss": 0.0004, "step": 312000 }, { "epoch": 2.22, "learning_rate": 8.284227939136882e-06, "loss": 0.0004, "step": 312500 }, { "epoch": 2.23, "learning_rate": 8.246402832343997e-06, "loss": 0.0004, "step": 313000 }, { "epoch": 2.23, "learning_rate": 8.208577725551113e-06, "loss": 0.0004, "step": 313500 }, { "epoch": 2.23, "learning_rate": 8.170752618758227e-06, "loss": 0.0004, "step": 314000 }, { "epoch": 2.24, "learning_rate": 8.132927511965343e-06, "loss": 0.0004, "step": 314500 }, { "epoch": 2.24, "learning_rate": 8.095102405172458e-06, "loss": 0.0004, "step": 315000 }, { "epoch": 2.24, "learning_rate": 8.057277298379572e-06, "loss": 0.0004, "step": 315500 }, { "epoch": 2.25, "learning_rate": 8.019527841800273e-06, "loss": 0.0004, "step": 316000 }, { "epoch": 2.25, "learning_rate": 7.981702735007389e-06, "loss": 0.0004, "step": 316500 }, { "epoch": 2.25, "learning_rate": 7.943877628214504e-06, "loss": 0.0004, "step": 317000 }, { "epoch": 2.26, "learning_rate": 7.90605252142162e-06, "loss": 0.0004, "step": 317500 }, { "epoch": 2.26, "learning_rate": 7.86830306484232e-06, "loss": 0.0004, "step": 318000 }, { "epoch": 2.26, "learning_rate": 7.830477958049435e-06, "loss": 0.0004, "step": 318500 }, { "epoch": 2.27, "learning_rate": 7.79265285125655e-06, "loss": 0.0004, "step": 319000 }, { "epoch": 2.27, "learning_rate": 7.754827744463666e-06, "loss": 0.0004, "step": 319500 }, { "epoch": 2.28, "learning_rate": 7.71700263767078e-06, "loss": 0.0004, "step": 320000 }, { "epoch": 2.28, "learning_rate": 7.679177530877896e-06, "loss": 0.0004, "step": 320500 }, { "epoch": 2.28, "learning_rate": 7.641352424085012e-06, "loss": 0.0004, "step": 321000 }, { "epoch": 2.29, "learning_rate": 7.603527317292126e-06, "loss": 0.0004, "step": 321500 }, { "epoch": 2.29, "learning_rate": 7.565702210499242e-06, "loss": 0.0004, "step": 322000 }, { "epoch": 2.29, "learning_rate": 7.527952753919942e-06, "loss": 0.0004, "step": 322500 }, { "epoch": 2.3, "learning_rate": 7.490127647127057e-06, "loss": 0.0004, "step": 323000 }, { "epoch": 2.3, "learning_rate": 7.452302540334172e-06, "loss": 0.0004, "step": 323500 }, { "epoch": 2.3, "learning_rate": 7.414477433541288e-06, "loss": 0.0004, "step": 324000 }, { "epoch": 2.31, "learning_rate": 7.376727976961989e-06, "loss": 0.0004, "step": 324500 }, { "epoch": 2.31, "learning_rate": 7.338902870169104e-06, "loss": 0.0004, "step": 325000 }, { "epoch": 2.31, "learning_rate": 7.301077763376219e-06, "loss": 0.0004, "step": 325500 }, { "epoch": 2.32, "learning_rate": 7.2632526565833336e-06, "loss": 0.0004, "step": 326000 }, { "epoch": 2.32, "learning_rate": 7.225503200004035e-06, "loss": 0.0004, "step": 326500 }, { "epoch": 2.33, "learning_rate": 7.18767809321115e-06, "loss": 0.0004, "step": 327000 }, { "epoch": 2.33, "learning_rate": 7.149852986418265e-06, "loss": 0.0004, "step": 327500 }, { "epoch": 2.33, "learning_rate": 7.11202787962538e-06, "loss": 0.0004, "step": 328000 }, { "epoch": 2.34, "learning_rate": 7.0742784230460815e-06, "loss": 0.0004, "step": 328500 }, { "epoch": 2.34, "learning_rate": 7.036453316253197e-06, "loss": 0.0004, "step": 329000 }, { "epoch": 2.34, "learning_rate": 6.998628209460311e-06, "loss": 0.0004, "step": 329500 }, { "epoch": 2.35, "learning_rate": 6.960803102667427e-06, "loss": 0.0004, "step": 330000 }, { "epoch": 2.35, "learning_rate": 6.923053646088128e-06, "loss": 0.0004, "step": 330500 }, { "epoch": 2.35, "learning_rate": 6.885228539295243e-06, "loss": 0.0004, "step": 331000 }, { "epoch": 2.36, "learning_rate": 6.847403432502358e-06, "loss": 0.0004, "step": 331500 }, { "epoch": 2.36, "learning_rate": 6.809653975923059e-06, "loss": 0.0004, "step": 332000 }, { "epoch": 2.36, "learning_rate": 6.771828869130174e-06, "loss": 0.0004, "step": 332500 }, { "epoch": 2.37, "learning_rate": 6.734003762337289e-06, "loss": 0.0004, "step": 333000 }, { "epoch": 2.37, "learning_rate": 6.696178655544405e-06, "loss": 0.0004, "step": 333500 }, { "epoch": 2.38, "learning_rate": 6.65835354875152e-06, "loss": 0.0004, "step": 334000 }, { "epoch": 2.38, "learning_rate": 6.620528441958634e-06, "loss": 0.0004, "step": 334500 }, { "epoch": 2.38, "learning_rate": 6.5827033351657495e-06, "loss": 0.0004, "step": 335000 }, { "epoch": 2.39, "learning_rate": 6.544878228372866e-06, "loss": 0.0004, "step": 335500 }, { "epoch": 2.39, "learning_rate": 6.507128771793566e-06, "loss": 0.0004, "step": 336000 }, { "epoch": 2.39, "learning_rate": 6.469303665000681e-06, "loss": 0.0004, "step": 336500 }, { "epoch": 2.4, "learning_rate": 6.431478558207796e-06, "loss": 0.0004, "step": 337000 }, { "epoch": 2.4, "learning_rate": 6.393653451414911e-06, "loss": 0.0004, "step": 337500 }, { "epoch": 2.4, "learning_rate": 6.355903994835612e-06, "loss": 0.0004, "step": 338000 }, { "epoch": 2.41, "learning_rate": 6.318078888042727e-06, "loss": 0.0004, "step": 338500 }, { "epoch": 2.41, "learning_rate": 6.280253781249843e-06, "loss": 0.0004, "step": 339000 }, { "epoch": 2.41, "learning_rate": 6.2424286744569575e-06, "loss": 0.0004, "step": 339500 }, { "epoch": 2.42, "learning_rate": 6.204679217877659e-06, "loss": 0.0004, "step": 340000 }, { "epoch": 2.42, "learning_rate": 6.166854111084774e-06, "loss": 0.0004, "step": 340500 }, { "epoch": 2.42, "learning_rate": 6.129029004291888e-06, "loss": 0.0004, "step": 341000 }, { "epoch": 2.43, "learning_rate": 6.091203897499004e-06, "loss": 0.0004, "step": 341500 }, { "epoch": 2.43, "learning_rate": 6.0534544409197055e-06, "loss": 0.0004, "step": 342000 }, { "epoch": 2.44, "learning_rate": 6.015629334126821e-06, "loss": 0.0004, "step": 342500 }, { "epoch": 2.44, "learning_rate": 5.977804227333935e-06, "loss": 0.0004, "step": 343000 }, { "epoch": 2.44, "learning_rate": 5.93997912054105e-06, "loss": 0.0004, "step": 343500 }, { "epoch": 2.45, "learning_rate": 5.902229663961751e-06, "loss": 0.0004, "step": 344000 }, { "epoch": 2.45, "learning_rate": 5.864404557168866e-06, "loss": 0.0004, "step": 344500 }, { "epoch": 2.45, "learning_rate": 5.826579450375982e-06, "loss": 0.0004, "step": 345000 }, { "epoch": 2.46, "learning_rate": 5.788754343583097e-06, "loss": 0.0004, "step": 345500 }, { "epoch": 2.46, "learning_rate": 5.751004887003798e-06, "loss": 0.0004, "step": 346000 }, { "epoch": 2.46, "learning_rate": 5.713179780210913e-06, "loss": 0.0004, "step": 346500 }, { "epoch": 2.47, "learning_rate": 5.675354673418028e-06, "loss": 0.0004, "step": 347000 }, { "epoch": 2.47, "learning_rate": 5.637529566625144e-06, "loss": 0.0004, "step": 347500 }, { "epoch": 2.47, "learning_rate": 5.599780110045844e-06, "loss": 0.0004, "step": 348000 }, { "epoch": 2.48, "learning_rate": 5.561955003252959e-06, "loss": 0.0004, "step": 348500 }, { "epoch": 2.48, "learning_rate": 5.5242055466736605e-06, "loss": 0.0004, "step": 349000 }, { "epoch": 2.49, "learning_rate": 5.486380439880776e-06, "loss": 0.0004, "step": 349500 }, { "epoch": 2.49, "learning_rate": 5.44855533308789e-06, "loss": 0.0004, "step": 350000 }, { "epoch": 2.49, "learning_rate": 5.410730226295005e-06, "loss": 0.0004, "step": 350500 }, { "epoch": 2.5, "learning_rate": 5.3729051195021214e-06, "loss": 0.0004, "step": 351000 }, { "epoch": 2.5, "learning_rate": 5.335080012709236e-06, "loss": 0.0004, "step": 351500 }, { "epoch": 2.5, "learning_rate": 5.297254905916351e-06, "loss": 0.0004, "step": 352000 }, { "epoch": 2.51, "learning_rate": 5.259429799123466e-06, "loss": 0.0004, "step": 352500 }, { "epoch": 2.51, "learning_rate": 5.221680342544167e-06, "loss": 0.0004, "step": 353000 }, { "epoch": 2.51, "learning_rate": 5.183855235751283e-06, "loss": 0.0004, "step": 353500 }, { "epoch": 2.52, "learning_rate": 5.146030128958398e-06, "loss": 0.0004, "step": 354000 }, { "epoch": 2.52, "learning_rate": 5.108205022165513e-06, "loss": 0.0004, "step": 354500 }, { "epoch": 2.52, "learning_rate": 5.070455565586213e-06, "loss": 0.0004, "step": 355000 }, { "epoch": 2.53, "learning_rate": 5.0326304587933286e-06, "loss": 0.0004, "step": 355500 }, { "epoch": 2.53, "learning_rate": 4.994805352000444e-06, "loss": 0.0004, "step": 356000 }, { "epoch": 2.54, "learning_rate": 4.956980245207559e-06, "loss": 0.0004, "step": 356500 }, { "epoch": 2.54, "learning_rate": 4.91923078862826e-06, "loss": 0.0004, "step": 357000 }, { "epoch": 2.54, "learning_rate": 4.881405681835375e-06, "loss": 0.0004, "step": 357500 }, { "epoch": 2.55, "learning_rate": 4.843580575042491e-06, "loss": 0.0004, "step": 358000 }, { "epoch": 2.55, "learning_rate": 4.805755468249605e-06, "loss": 0.0004, "step": 358500 }, { "epoch": 2.55, "learning_rate": 4.768006011670306e-06, "loss": 0.0004, "step": 359000 }, { "epoch": 2.56, "learning_rate": 4.730180904877422e-06, "loss": 0.0003, "step": 359500 }, { "epoch": 2.56, "learning_rate": 4.6923557980845366e-06, "loss": 0.0004, "step": 360000 }, { "epoch": 2.56, "learning_rate": 4.654530691291652e-06, "loss": 0.0004, "step": 360500 }, { "epoch": 2.57, "learning_rate": 4.616781234712353e-06, "loss": 0.0004, "step": 361000 }, { "epoch": 2.57, "learning_rate": 4.578956127919468e-06, "loss": 0.0004, "step": 361500 }, { "epoch": 2.57, "learning_rate": 4.5411310211265825e-06, "loss": 0.0004, "step": 362000 }, { "epoch": 2.58, "learning_rate": 4.503305914333699e-06, "loss": 0.0004, "step": 362500 }, { "epoch": 2.58, "learning_rate": 4.4655564577544e-06, "loss": 0.0004, "step": 363000 }, { "epoch": 2.58, "learning_rate": 4.427731350961514e-06, "loss": 0.0004, "step": 363500 }, { "epoch": 2.59, "learning_rate": 4.389906244168629e-06, "loss": 0.0004, "step": 364000 }, { "epoch": 2.59, "learning_rate": 4.3520811373757445e-06, "loss": 0.0003, "step": 364500 }, { "epoch": 2.6, "learning_rate": 4.31425603058286e-06, "loss": 0.0004, "step": 365000 }, { "epoch": 2.6, "learning_rate": 4.276430923789975e-06, "loss": 0.0004, "step": 365500 }, { "epoch": 2.6, "learning_rate": 4.23860581699709e-06, "loss": 0.0004, "step": 366000 }, { "epoch": 2.61, "learning_rate": 4.2007807102042054e-06, "loss": 0.0004, "step": 366500 }, { "epoch": 2.61, "learning_rate": 4.163031253624906e-06, "loss": 0.0004, "step": 367000 }, { "epoch": 2.61, "learning_rate": 4.125206146832021e-06, "loss": 0.0004, "step": 367500 }, { "epoch": 2.62, "learning_rate": 4.087381040039137e-06, "loss": 0.0004, "step": 368000 }, { "epoch": 2.62, "learning_rate": 4.049555933246251e-06, "loss": 0.0004, "step": 368500 }, { "epoch": 2.62, "learning_rate": 4.0118064766669525e-06, "loss": 0.0004, "step": 369000 }, { "epoch": 2.63, "learning_rate": 3.973981369874068e-06, "loss": 0.0004, "step": 369500 }, { "epoch": 2.63, "learning_rate": 3.936156263081183e-06, "loss": 0.0004, "step": 370000 }, { "epoch": 2.63, "learning_rate": 3.898331156288298e-06, "loss": 0.0004, "step": 370500 }, { "epoch": 2.64, "learning_rate": 3.8605060494954134e-06, "loss": 0.0004, "step": 371000 }, { "epoch": 2.64, "learning_rate": 3.8227565929161146e-06, "loss": 0.0003, "step": 371500 }, { "epoch": 2.65, "learning_rate": 3.7849314861232294e-06, "loss": 0.0004, "step": 372000 }, { "epoch": 2.65, "learning_rate": 3.747106379330344e-06, "loss": 0.0004, "step": 372500 }, { "epoch": 2.65, "learning_rate": 3.70928127253746e-06, "loss": 0.0003, "step": 373000 }, { "epoch": 2.66, "learning_rate": 3.6715318159581605e-06, "loss": 0.0004, "step": 373500 }, { "epoch": 2.66, "learning_rate": 3.6337067091652753e-06, "loss": 0.0004, "step": 374000 }, { "epoch": 2.66, "learning_rate": 3.595881602372391e-06, "loss": 0.0004, "step": 374500 }, { "epoch": 2.67, "learning_rate": 3.5580564955795058e-06, "loss": 0.0004, "step": 375000 }, { "epoch": 2.67, "learning_rate": 3.520307039000207e-06, "loss": 0.0004, "step": 375500 }, { "epoch": 2.67, "learning_rate": 3.482481932207322e-06, "loss": 0.0004, "step": 376000 }, { "epoch": 2.68, "learning_rate": 3.444656825414437e-06, "loss": 0.0004, "step": 376500 }, { "epoch": 2.68, "learning_rate": 3.4068317186215526e-06, "loss": 0.0004, "step": 377000 }, { "epoch": 2.68, "learning_rate": 3.3690822620422533e-06, "loss": 0.0004, "step": 377500 }, { "epoch": 2.69, "learning_rate": 3.3312571552493685e-06, "loss": 0.0004, "step": 378000 }, { "epoch": 2.69, "learning_rate": 3.2934320484564833e-06, "loss": 0.0004, "step": 378500 }, { "epoch": 2.7, "learning_rate": 3.255606941663599e-06, "loss": 0.0004, "step": 379000 }, { "epoch": 2.7, "learning_rate": 3.2178574850842997e-06, "loss": 0.0004, "step": 379500 }, { "epoch": 2.7, "learning_rate": 3.1800323782914145e-06, "loss": 0.0003, "step": 380000 }, { "epoch": 2.71, "learning_rate": 3.14220727149853e-06, "loss": 0.0004, "step": 380500 }, { "epoch": 2.71, "learning_rate": 3.104382164705645e-06, "loss": 0.0003, "step": 381000 }, { "epoch": 2.71, "learning_rate": 3.066632708126346e-06, "loss": 0.0004, "step": 381500 }, { "epoch": 2.72, "learning_rate": 3.0288076013334613e-06, "loss": 0.0004, "step": 382000 }, { "epoch": 2.72, "learning_rate": 2.9909824945405765e-06, "loss": 0.0004, "step": 382500 }, { "epoch": 2.72, "learning_rate": 2.9531573877476917e-06, "loss": 0.0004, "step": 383000 }, { "epoch": 2.73, "learning_rate": 2.9154079311683924e-06, "loss": 0.0004, "step": 383500 }, { "epoch": 2.73, "learning_rate": 2.8775828243755076e-06, "loss": 0.0004, "step": 384000 }, { "epoch": 2.73, "learning_rate": 2.8397577175826224e-06, "loss": 0.0004, "step": 384500 }, { "epoch": 2.74, "learning_rate": 2.801932610789738e-06, "loss": 0.0004, "step": 385000 }, { "epoch": 2.74, "learning_rate": 2.764183154210439e-06, "loss": 0.0003, "step": 385500 }, { "epoch": 2.74, "learning_rate": 2.726358047417554e-06, "loss": 0.0003, "step": 386000 }, { "epoch": 2.75, "learning_rate": 2.6885329406246692e-06, "loss": 0.0004, "step": 386500 }, { "epoch": 2.75, "learning_rate": 2.650707833831784e-06, "loss": 0.0004, "step": 387000 }, { "epoch": 2.76, "learning_rate": 2.612958377252485e-06, "loss": 0.0004, "step": 387500 }, { "epoch": 2.76, "learning_rate": 2.5751332704596004e-06, "loss": 0.0004, "step": 388000 }, { "epoch": 2.76, "learning_rate": 2.5373081636667156e-06, "loss": 0.0004, "step": 388500 }, { "epoch": 2.77, "learning_rate": 2.499483056873831e-06, "loss": 0.0004, "step": 389000 }, { "epoch": 2.77, "learning_rate": 2.4617336002945316e-06, "loss": 0.0004, "step": 389500 }, { "epoch": 2.77, "learning_rate": 2.4239084935016468e-06, "loss": 0.0004, "step": 390000 }, { "epoch": 2.78, "learning_rate": 2.3860833867087616e-06, "loss": 0.0004, "step": 390500 }, { "epoch": 2.78, "learning_rate": 2.3482582799158772e-06, "loss": 0.0004, "step": 391000 }, { "epoch": 2.78, "learning_rate": 2.310508823336578e-06, "loss": 0.0004, "step": 391500 }, { "epoch": 2.79, "learning_rate": 2.272683716543693e-06, "loss": 0.0004, "step": 392000 }, { "epoch": 2.79, "learning_rate": 2.2348586097508084e-06, "loss": 0.0004, "step": 392500 }, { "epoch": 2.79, "learning_rate": 2.197033502957923e-06, "loss": 0.0004, "step": 393000 }, { "epoch": 2.8, "learning_rate": 2.1592840463786243e-06, "loss": 0.0004, "step": 393500 }, { "epoch": 2.8, "learning_rate": 2.1214589395857395e-06, "loss": 0.0003, "step": 394000 }, { "epoch": 2.81, "learning_rate": 2.0836338327928548e-06, "loss": 0.0004, "step": 394500 }, { "epoch": 2.81, "learning_rate": 2.04580872599997e-06, "loss": 0.0003, "step": 395000 }, { "epoch": 2.81, "learning_rate": 2.008059269420671e-06, "loss": 0.0004, "step": 395500 }, { "epoch": 2.82, "learning_rate": 1.970234162627786e-06, "loss": 0.0004, "step": 396000 }, { "epoch": 2.82, "learning_rate": 1.9324090558349007e-06, "loss": 0.0004, "step": 396500 }, { "epoch": 2.82, "learning_rate": 1.8945839490420162e-06, "loss": 0.0003, "step": 397000 }, { "epoch": 2.83, "learning_rate": 1.856834492462717e-06, "loss": 0.0004, "step": 397500 }, { "epoch": 2.83, "learning_rate": 1.8190093856698323e-06, "loss": 0.0004, "step": 398000 }, { "epoch": 2.83, "learning_rate": 1.7811842788769473e-06, "loss": 0.0003, "step": 398500 }, { "epoch": 2.84, "learning_rate": 1.7433591720840625e-06, "loss": 0.0004, "step": 399000 }, { "epoch": 2.84, "learning_rate": 1.7056097155047635e-06, "loss": 0.0003, "step": 399500 }, { "epoch": 2.84, "learning_rate": 1.6677846087118787e-06, "loss": 0.0003, "step": 400000 }, { "epoch": 2.85, "learning_rate": 1.6299595019189937e-06, "loss": 0.0003, "step": 400500 }, { "epoch": 2.85, "learning_rate": 1.592134395126109e-06, "loss": 0.0004, "step": 401000 }, { "epoch": 2.86, "learning_rate": 1.5543849385468098e-06, "loss": 0.0003, "step": 401500 }, { "epoch": 2.86, "learning_rate": 1.516559831753925e-06, "loss": 0.0003, "step": 402000 }, { "epoch": 2.86, "learning_rate": 1.4787347249610403e-06, "loss": 0.0003, "step": 402500 }, { "epoch": 2.87, "learning_rate": 1.4409096181681553e-06, "loss": 0.0003, "step": 403000 }, { "epoch": 2.87, "learning_rate": 1.4030845113752705e-06, "loss": 0.0003, "step": 403500 }, { "epoch": 2.87, "learning_rate": 1.3653350547959714e-06, "loss": 0.0004, "step": 404000 }, { "epoch": 2.88, "learning_rate": 1.3275099480030865e-06, "loss": 0.0003, "step": 404500 }, { "epoch": 2.88, "learning_rate": 1.2896848412102017e-06, "loss": 0.0004, "step": 405000 }, { "epoch": 2.88, "learning_rate": 1.251859734417317e-06, "loss": 0.0004, "step": 405500 }, { "epoch": 2.89, "learning_rate": 1.2141102778380178e-06, "loss": 0.0003, "step": 406000 }, { "epoch": 2.89, "learning_rate": 1.1762851710451328e-06, "loss": 0.0004, "step": 406500 }, { "epoch": 2.89, "learning_rate": 1.138460064252248e-06, "loss": 0.0004, "step": 407000 }, { "epoch": 2.9, "learning_rate": 1.1006349574593633e-06, "loss": 0.0004, "step": 407500 }, { "epoch": 2.9, "learning_rate": 1.0628855008800642e-06, "loss": 0.0003, "step": 408000 }, { "epoch": 2.9, "learning_rate": 1.0250603940871794e-06, "loss": 0.0003, "step": 408500 }, { "epoch": 2.91, "learning_rate": 9.872352872942944e-07, "loss": 0.0004, "step": 409000 }, { "epoch": 2.91, "learning_rate": 9.494101805014097e-07, "loss": 0.0004, "step": 409500 }, { "epoch": 2.92, "learning_rate": 9.116607239221106e-07, "loss": 0.0003, "step": 410000 }, { "epoch": 2.92, "learning_rate": 8.738356171292257e-07, "loss": 0.0003, "step": 410500 }, { "epoch": 2.92, "learning_rate": 8.360105103363409e-07, "loss": 0.0003, "step": 411000 }, { "epoch": 2.93, "learning_rate": 7.98185403543456e-07, "loss": 0.0003, "step": 411500 }, { "epoch": 2.93, "learning_rate": 7.604359469641569e-07, "loss": 0.0003, "step": 412000 }, { "epoch": 2.93, "learning_rate": 7.226108401712721e-07, "loss": 0.0004, "step": 412500 }, { "epoch": 2.94, "learning_rate": 6.847857333783872e-07, "loss": 0.0003, "step": 413000 }, { "epoch": 2.94, "learning_rate": 6.469606265855024e-07, "loss": 0.0003, "step": 413500 }, { "epoch": 2.94, "learning_rate": 6.092111700062033e-07, "loss": 0.0003, "step": 414000 }, { "epoch": 2.95, "learning_rate": 5.713860632133186e-07, "loss": 0.0004, "step": 414500 }, { "epoch": 2.95, "learning_rate": 5.335609564204336e-07, "loss": 0.0004, "step": 415000 }, { "epoch": 2.95, "learning_rate": 4.957358496275488e-07, "loss": 0.0004, "step": 415500 }, { "epoch": 2.96, "learning_rate": 4.5806204326183547e-07, "loss": 0.0003, "step": 416000 }, { "epoch": 2.96, "learning_rate": 4.2023693646895064e-07, "loss": 0.0003, "step": 416500 }, { "epoch": 2.97, "learning_rate": 3.824118296760658e-07, "loss": 0.0004, "step": 417000 }, { "epoch": 2.97, "learning_rate": 3.4458672288318094e-07, "loss": 0.0003, "step": 417500 }, { "epoch": 2.97, "learning_rate": 3.067616160902961e-07, "loss": 0.0003, "step": 418000 }, { "epoch": 2.98, "learning_rate": 2.6893650929741123e-07, "loss": 0.0003, "step": 418500 }, { "epoch": 2.98, "learning_rate": 2.311114025045264e-07, "loss": 0.0003, "step": 419000 }, { "epoch": 2.98, "learning_rate": 1.9328629571164155e-07, "loss": 0.0004, "step": 419500 }, { "epoch": 2.99, "learning_rate": 1.555368391323425e-07, "loss": 0.0003, "step": 420000 }, { "epoch": 2.99, "learning_rate": 1.1771173233945763e-07, "loss": 0.0003, "step": 420500 }, { "epoch": 2.99, "learning_rate": 7.98866255465728e-08, "loss": 0.0004, "step": 421000 }, { "epoch": 3.0, "learning_rate": 4.206151875368795e-08, "loss": 0.0004, "step": 421500 }, { "epoch": 3.0, "step": 421875, "total_flos": 0, "train_runtime": 86205.303, "train_samples_per_second": 4.894 } ], "max_steps": 421875, "num_train_epochs": 3, "total_flos": 0, "trial_name": null, "trial_params": null }