| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 375680, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00019998669080068145, | |
| "loss": 10.2481, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 0.00019997338160136289, | |
| "loss": 5.6824, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 0.0001999600724020443, | |
| "loss": 5.172, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019994676320272574, | |
| "loss": 5.0986, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 0.00019993345400340717, | |
| "loss": 4.9605, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 0.0001999201448040886, | |
| "loss": 4.7834, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019990683560477005, | |
| "loss": 4.7696, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 0.00019989352640545146, | |
| "loss": 4.6645, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 0.00019988021720613287, | |
| "loss": 4.5375, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.0001998669080068143, | |
| "loss": 4.4608, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00019985359880749575, | |
| "loss": 4.3304, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 0.00019984028960817719, | |
| "loss": 4.2935, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.0001998269804088586, | |
| "loss": 4.0207, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 0.00019981367120954004, | |
| "loss": 4.1862, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 0.00019980036201022147, | |
| "loss": 4.1547, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.0001997870528109029, | |
| "loss": 4.0028, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 0.00019977374361158435, | |
| "loss": 3.9894, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 0.00019976043441226576, | |
| "loss": 4.0722, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.0001997471252129472, | |
| "loss": 3.8993, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 0.00019973381601362864, | |
| "loss": 3.8414, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.00019972050681431008, | |
| "loss": 3.8328, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019970719761499149, | |
| "loss": 3.731, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00019969388841567292, | |
| "loss": 3.8229, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 0.00019968057921635434, | |
| "loss": 3.6892, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.00019966727001703577, | |
| "loss": 3.5272, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 0.0001996539608177172, | |
| "loss": 3.7993, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 0.00019964065161839865, | |
| "loss": 3.5962, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001996273424190801, | |
| "loss": 3.6104, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 0.0001996140332197615, | |
| "loss": 3.5948, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 0.00019960072402044294, | |
| "loss": 3.5836, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00019958741482112438, | |
| "loss": 3.5037, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 0.00019957410562180581, | |
| "loss": 3.5228, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00019956079642248725, | |
| "loss": 3.4802, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.00019954748722316866, | |
| "loss": 3.3982, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 0.0001995341780238501, | |
| "loss": 3.4461, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 0.0001995208688245315, | |
| "loss": 3.426, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.00019950755962521295, | |
| "loss": 3.5081, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 0.0001994942504258944, | |
| "loss": 3.3901, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 0.0001994809412265758, | |
| "loss": 3.394, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00019946763202725724, | |
| "loss": 3.2572, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 0.00019945432282793868, | |
| "loss": 3.1521, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 0.00019944101362862011, | |
| "loss": 3.3596, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00019942770442930155, | |
| "loss": 3.2875, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00019941439522998296, | |
| "loss": 3.2877, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0001994010860306644, | |
| "loss": 3.2511, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00019938777683134584, | |
| "loss": 3.2489, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 0.00019937446763202728, | |
| "loss": 3.189, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 0.00019936115843270872, | |
| "loss": 3.1982, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00019934784923339013, | |
| "loss": 3.2822, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 0.00019933454003407156, | |
| "loss": 3.0918, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 0.00019932123083475298, | |
| "loss": 3.1034, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00019930792163543441, | |
| "loss": 3.1271, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 0.00019929461243611585, | |
| "loss": 3.0492, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0001992813032367973, | |
| "loss": 3.1039, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0001992679940374787, | |
| "loss": 3.1402, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.00019925468483816014, | |
| "loss": 3.1043, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 0.00019924137563884158, | |
| "loss": 3.1407, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00019922806643952302, | |
| "loss": 3.0666, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 0.00019921475724020445, | |
| "loss": 3.0374, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 0.00019920144804088586, | |
| "loss": 3.0348, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.0001991881388415673, | |
| "loss": 3.1047, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 0.00019917482964224874, | |
| "loss": 3.0446, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 0.00019916152044293018, | |
| "loss": 3.0877, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.00019914821124361162, | |
| "loss": 2.926, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.000199134902044293, | |
| "loss": 3.0038, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 0.00019912159284497444, | |
| "loss": 3.0531, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00019910828364565588, | |
| "loss": 3.0204, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.00019909497444633732, | |
| "loss": 3.0724, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 0.00019908166524701875, | |
| "loss": 3.1021, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.00019906835604770017, | |
| "loss": 3.0148, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 0.0001990550468483816, | |
| "loss": 2.9517, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 0.00019904173764906304, | |
| "loss": 2.9411, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00019902842844974448, | |
| "loss": 2.9533, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 0.00019901511925042592, | |
| "loss": 2.9113, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00019900181005110733, | |
| "loss": 2.9085, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.00019898850085178877, | |
| "loss": 2.9903, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 0.0001989751916524702, | |
| "loss": 2.8475, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00019896188245315164, | |
| "loss": 2.9614, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.00019894857325383305, | |
| "loss": 2.9126, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 0.0001989352640545145, | |
| "loss": 2.919, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 0.0001989219548551959, | |
| "loss": 2.911, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00019890864565587734, | |
| "loss": 2.8617, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 0.00019889533645655878, | |
| "loss": 2.8817, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 0.00019888202725724022, | |
| "loss": 2.9325, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00019886871805792166, | |
| "loss": 2.8276, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.00019885540885860307, | |
| "loss": 2.9285, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 0.0001988420996592845, | |
| "loss": 2.7656, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00019882879045996594, | |
| "loss": 2.9298, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 0.00019881548126064738, | |
| "loss": 2.8953, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00019880217206132882, | |
| "loss": 2.8127, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00019878886286201023, | |
| "loss": 2.8404, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 0.00019877555366269167, | |
| "loss": 2.8233, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 0.0001987622444633731, | |
| "loss": 2.9052, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00019874893526405452, | |
| "loss": 2.8838, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 0.00019873562606473596, | |
| "loss": 2.8243, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 0.00019872231686541737, | |
| "loss": 2.7138, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0001987090076660988, | |
| "loss": 2.7165, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.00019869569846678024, | |
| "loss": 2.8223, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 0.00019868238926746168, | |
| "loss": 2.8202, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00019866908006814312, | |
| "loss": 2.8057, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.00019865577086882453, | |
| "loss": 2.774, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 0.00019864246166950597, | |
| "loss": 2.8091, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.0001986291524701874, | |
| "loss": 2.785, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 0.00019861584327086885, | |
| "loss": 2.6157, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 0.00019860253407155028, | |
| "loss": 2.8632, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.0001985892248722317, | |
| "loss": 2.7628, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 0.00019857591567291313, | |
| "loss": 2.8141, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00019856260647359454, | |
| "loss": 2.7369, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00019854929727427598, | |
| "loss": 2.721, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 0.00019853598807495742, | |
| "loss": 2.7448, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 0.00019852267887563886, | |
| "loss": 2.6667, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.00019850936967632027, | |
| "loss": 2.7889, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0001984960604770017, | |
| "loss": 2.823, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 0.00019848275127768315, | |
| "loss": 2.7421, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00019846944207836458, | |
| "loss": 2.8042, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 0.00019845613287904602, | |
| "loss": 2.69, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 0.00019844282367972743, | |
| "loss": 2.7895, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.00019842951448040887, | |
| "loss": 2.7649, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0001984162052810903, | |
| "loss": 2.7463, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 0.00019840289608177175, | |
| "loss": 2.7047, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00019838958688245316, | |
| "loss": 2.73, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 0.00019837627768313457, | |
| "loss": 2.7422, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.000198362968483816, | |
| "loss": 2.7778, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00019834965928449745, | |
| "loss": 2.5151, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 0.00019833635008517888, | |
| "loss": 2.7631, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 0.00019832304088586032, | |
| "loss": 2.7046, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00019830973168654173, | |
| "loss": 2.6186, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 0.00019829642248722317, | |
| "loss": 2.6684, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0001982831132879046, | |
| "loss": 2.687, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00019826980408858605, | |
| "loss": 2.6079, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 0.00019825649488926749, | |
| "loss": 2.6386, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 0.0001982431856899489, | |
| "loss": 2.5986, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00019822987649063033, | |
| "loss": 2.639, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 0.00019821656729131177, | |
| "loss": 2.6357, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0001982032580919932, | |
| "loss": 2.6278, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00019818994889267465, | |
| "loss": 2.633, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 0.00019817663969335606, | |
| "loss": 2.7147, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.00019816333049403747, | |
| "loss": 2.6298, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 0.0001981500212947189, | |
| "loss": 2.6853, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00019813671209540035, | |
| "loss": 2.5161, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.00019812340289608179, | |
| "loss": 2.6841, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 0.0001981100936967632, | |
| "loss": 2.7514, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 0.00019809678449744464, | |
| "loss": 2.6619, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.00019808347529812607, | |
| "loss": 2.6089, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 0.0001980701660988075, | |
| "loss": 2.5672, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.00019805685689948895, | |
| "loss": 2.6084, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.00019804354770017036, | |
| "loss": 2.6425, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 0.0001980302385008518, | |
| "loss": 2.5539, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 0.00019801692930153324, | |
| "loss": 2.588, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00019800362010221468, | |
| "loss": 2.6845, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_loss": 2.289619207382202, | |
| "eval_runtime": 153.5049, | |
| "eval_samples_per_second": 61.19, | |
| "eval_steps_per_second": 61.19, | |
| "step": 75136 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.00019799031090289609, | |
| "loss": 2.4516, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 0.00019797700170357752, | |
| "loss": 2.3205, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00019796369250425894, | |
| "loss": 2.3807, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 0.00019795038330494037, | |
| "loss": 2.4826, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 0.0001979370741056218, | |
| "loss": 2.4244, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.00019792376490630325, | |
| "loss": 2.4297, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 0.0001979104557069847, | |
| "loss": 2.5217, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.0001978971465076661, | |
| "loss": 2.4128, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00019788383730834754, | |
| "loss": 2.4708, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 0.00019787052810902898, | |
| "loss": 2.3967, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 0.0001978572189097104, | |
| "loss": 2.5101, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00019784390971039185, | |
| "loss": 2.4615, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00019783060051107326, | |
| "loss": 2.4171, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 0.0001978172913117547, | |
| "loss": 2.4693, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.0001978039821124361, | |
| "loss": 2.4741, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 0.00019779067291311755, | |
| "loss": 2.3287, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 0.000197777363713799, | |
| "loss": 2.4375, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0001977640545144804, | |
| "loss": 2.4086, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.00019775074531516184, | |
| "loss": 2.499, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 0.00019773743611584328, | |
| "loss": 2.5159, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00019772412691652471, | |
| "loss": 2.4019, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 0.00019771081771720615, | |
| "loss": 2.399, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00019769750851788756, | |
| "loss": 2.5017, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.000197684199318569, | |
| "loss": 2.4847, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 0.00019767089011925044, | |
| "loss": 2.4126, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 0.00019765758091993188, | |
| "loss": 2.4905, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00019764427172061332, | |
| "loss": 2.3939, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 0.00019763096252129473, | |
| "loss": 2.483, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 0.00019761765332197614, | |
| "loss": 2.3784, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00019760434412265758, | |
| "loss": 2.3567, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00019759103492333901, | |
| "loss": 2.4193, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 0.00019757772572402045, | |
| "loss": 2.4165, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001975644165247019, | |
| "loss": 2.4415, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.0001975511073253833, | |
| "loss": 2.417, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 0.00019753779812606474, | |
| "loss": 2.3009, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00019752448892674618, | |
| "loss": 2.4853, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 0.00019751117972742762, | |
| "loss": 2.3344, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 0.00019749787052810905, | |
| "loss": 2.4312, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.00019748456132879046, | |
| "loss": 2.4414, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 0.0001974712521294719, | |
| "loss": 2.4249, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.00019745794293015334, | |
| "loss": 2.356, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00019744463373083478, | |
| "loss": 2.3665, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 0.00019743132453151622, | |
| "loss": 2.4506, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.0001974180153321976, | |
| "loss": 2.413, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00019740470613287904, | |
| "loss": 2.4709, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 0.00019739139693356048, | |
| "loss": 2.3756, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 0.00019737808773424192, | |
| "loss": 2.3728, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00019736477853492335, | |
| "loss": 2.4195, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 0.00019735146933560476, | |
| "loss": 2.454, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 0.0001973381601362862, | |
| "loss": 2.3611, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00019732485093696764, | |
| "loss": 2.4303, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 0.00019731154173764908, | |
| "loss": 2.4555, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00019729823253833052, | |
| "loss": 2.3825, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00019728492333901193, | |
| "loss": 2.3164, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.00019727161413969337, | |
| "loss": 2.4844, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 0.0001972583049403748, | |
| "loss": 2.4749, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00019724499574105624, | |
| "loss": 2.4141, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 0.00019723168654173765, | |
| "loss": 2.4732, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 0.0001972183773424191, | |
| "loss": 2.4103, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.0001972050681431005, | |
| "loss": 2.244, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 0.00019719175894378194, | |
| "loss": 2.4251, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 0.00019717844974446338, | |
| "loss": 2.4125, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00019716514054514482, | |
| "loss": 2.4054, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.00019715183134582626, | |
| "loss": 2.3574, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 0.00019713852214650767, | |
| "loss": 2.4949, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0001971252129471891, | |
| "loss": 2.3071, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.00019711190374787054, | |
| "loss": 2.3755, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 0.00019709859454855198, | |
| "loss": 2.3994, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00019708528534923342, | |
| "loss": 2.425, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 0.00019707197614991483, | |
| "loss": 2.3043, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 0.00019705866695059627, | |
| "loss": 2.4273, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00019704535775127768, | |
| "loss": 2.392, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 0.00019703204855195912, | |
| "loss": 2.2706, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 0.00019701873935264056, | |
| "loss": 2.3492, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00019700543015332197, | |
| "loss": 2.4536, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.0001969921209540034, | |
| "loss": 2.4908, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00019697881175468484, | |
| "loss": 2.2912, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00019696550255536628, | |
| "loss": 2.346, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 0.00019695219335604772, | |
| "loss": 2.3519, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 0.00019693888415672913, | |
| "loss": 2.3962, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.00019692557495741057, | |
| "loss": 2.2665, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 0.000196912265758092, | |
| "loss": 2.3963, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 0.00019689895655877345, | |
| "loss": 2.3922, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.00019688564735945488, | |
| "loss": 2.3997, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 0.0001968723381601363, | |
| "loss": 2.391, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.00019685902896081773, | |
| "loss": 2.4071, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00019684571976149914, | |
| "loss": 2.3302, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.00019683241056218058, | |
| "loss": 2.2663, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 0.00019681910136286202, | |
| "loss": 2.4085, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00019680579216354346, | |
| "loss": 2.3689, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 0.00019679248296422487, | |
| "loss": 2.466, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 0.0001967791737649063, | |
| "loss": 2.3651, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00019676586456558775, | |
| "loss": 2.3936, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 0.00019675255536626918, | |
| "loss": 2.3022, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 0.0001967392461669506, | |
| "loss": 2.3889, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00019672593696763203, | |
| "loss": 2.2829, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 0.00019671262776831347, | |
| "loss": 2.331, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0001966993185689949, | |
| "loss": 2.3535, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00019668600936967635, | |
| "loss": 2.3339, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 0.00019667270017035776, | |
| "loss": 2.3372, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 0.00019665939097103917, | |
| "loss": 2.4082, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.0001966460817717206, | |
| "loss": 2.4428, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 0.00019663277257240205, | |
| "loss": 2.3537, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 0.00019661946337308348, | |
| "loss": 2.423, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00019660615417376492, | |
| "loss": 2.2162, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 0.00019659284497444633, | |
| "loss": 2.3798, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 0.00019657953577512777, | |
| "loss": 2.3549, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0001965662265758092, | |
| "loss": 2.3361, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.00019655291737649065, | |
| "loss": 2.3528, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 0.00019653960817717209, | |
| "loss": 2.3553, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.0001965262989778535, | |
| "loss": 2.1969, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 0.00019651298977853493, | |
| "loss": 2.3564, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 0.00019649968057921637, | |
| "loss": 2.3577, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.0001964863713798978, | |
| "loss": 2.398, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 0.00019647306218057922, | |
| "loss": 2.4231, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 0.00019645975298126066, | |
| "loss": 2.3658, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.00019644644378194207, | |
| "loss": 2.3445, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 0.0001964331345826235, | |
| "loss": 2.2921, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00019641982538330495, | |
| "loss": 2.3487, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00019640651618398639, | |
| "loss": 2.2385, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.0001963932069846678, | |
| "loss": 2.4189, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 0.00019637989778534923, | |
| "loss": 2.3459, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.00019636658858603067, | |
| "loss": 2.3295, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 0.0001963532793867121, | |
| "loss": 2.2921, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 0.00019633997018739355, | |
| "loss": 2.3225, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.00019632666098807496, | |
| "loss": 2.4121, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 0.0001963133517887564, | |
| "loss": 2.4108, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 0.00019630004258943784, | |
| "loss": 2.4693, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00019628673339011927, | |
| "loss": 2.3493, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 0.00019627342419080069, | |
| "loss": 2.3214, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00019626011499148212, | |
| "loss": 2.3452, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00019624680579216354, | |
| "loss": 2.2276, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 0.00019623349659284497, | |
| "loss": 2.2652, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 0.0001962201873935264, | |
| "loss": 2.406, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.00019620687819420785, | |
| "loss": 2.2756, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 0.0001961935689948893, | |
| "loss": 2.2884, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 0.0001961802597955707, | |
| "loss": 2.27, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00019616695059625214, | |
| "loss": 2.3682, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 0.00019615364139693358, | |
| "loss": 2.2721, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 0.000196140332197615, | |
| "loss": 2.3077, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00019612702299829645, | |
| "loss": 2.2378, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00019611371379897786, | |
| "loss": 2.4656, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.0001961004045996593, | |
| "loss": 2.2858, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.0001960870954003407, | |
| "loss": 2.2983, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 0.00019607378620102215, | |
| "loss": 2.3561, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 0.0001960604770017036, | |
| "loss": 2.3414, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.000196047167802385, | |
| "loss": 2.279, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 0.00019603385860306644, | |
| "loss": 2.3095, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 0.00019602054940374788, | |
| "loss": 2.3459, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.0001960072402044293, | |
| "loss": 2.3968, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_loss": 2.083725929260254, | |
| "eval_runtime": 150.1089, | |
| "eval_samples_per_second": 62.575, | |
| "eval_steps_per_second": 62.575, | |
| "step": 150272 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 0.00019599393100511075, | |
| "loss": 2.245, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.00019598062180579216, | |
| "loss": 2.1479, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0001959673126064736, | |
| "loss": 2.1048, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.00019595400340715504, | |
| "loss": 2.028, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 0.00019594069420783648, | |
| "loss": 2.0738, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00019592738500851792, | |
| "loss": 2.1684, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 0.00019591407580919933, | |
| "loss": 2.0715, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 0.00019590076660988074, | |
| "loss": 2.1869, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.00019588745741056218, | |
| "loss": 2.0887, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 0.00019587414821124361, | |
| "loss": 2.0761, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 0.00019586083901192505, | |
| "loss": 2.0249, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0001958475298126065, | |
| "loss": 1.998, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0001958342206132879, | |
| "loss": 2.186, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 0.00019582091141396934, | |
| "loss": 2.2012, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00019580760221465078, | |
| "loss": 2.1327, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00019579429301533222, | |
| "loss": 2.0996, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 0.00019578098381601365, | |
| "loss": 2.0994, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.00019576767461669506, | |
| "loss": 2.057, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 0.0001957543654173765, | |
| "loss": 2.0814, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 0.00019574105621805794, | |
| "loss": 2.2237, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.00019572774701873938, | |
| "loss": 2.113, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 0.00019571443781942082, | |
| "loss": 2.2014, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.0001957011286201022, | |
| "loss": 2.1535, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.00019568781942078364, | |
| "loss": 2.1548, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 0.00019567451022146508, | |
| "loss": 2.2726, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.00019566120102214652, | |
| "loss": 2.0613, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00019564789182282795, | |
| "loss": 2.1457, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 0.00019563458262350936, | |
| "loss": 2.0963, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 0.0001956212734241908, | |
| "loss": 2.1481, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00019560796422487224, | |
| "loss": 2.104, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 0.00019559465502555368, | |
| "loss": 2.0308, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 0.00019558134582623512, | |
| "loss": 2.0471, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.00019556803662691653, | |
| "loss": 2.1354, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.00019555472742759797, | |
| "loss": 2.1223, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 0.0001955414182282794, | |
| "loss": 2.1158, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.00019552810902896084, | |
| "loss": 2.2403, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 0.00019551479982964225, | |
| "loss": 2.2066, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.0001955014906303237, | |
| "loss": 2.0638, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.0001954881814310051, | |
| "loss": 2.25, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 0.00019547487223168654, | |
| "loss": 2.1617, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 0.00019546156303236798, | |
| "loss": 2.2301, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.00019544825383304942, | |
| "loss": 2.165, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 0.00019543494463373086, | |
| "loss": 2.1395, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 0.00019542163543441227, | |
| "loss": 2.1827, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0001954083262350937, | |
| "loss": 2.1452, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.00019539501703577514, | |
| "loss": 2.1515, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 0.00019538170783645658, | |
| "loss": 2.1997, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00019536839863713802, | |
| "loss": 2.1392, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.00019535508943781943, | |
| "loss": 2.0866, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 0.00019534178023850087, | |
| "loss": 2.0945, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.00019532847103918228, | |
| "loss": 2.2065, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 0.00019531516183986372, | |
| "loss": 2.1329, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 0.00019530185264054516, | |
| "loss": 2.165, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.00019528854344122657, | |
| "loss": 2.1542, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 0.000195275234241908, | |
| "loss": 2.1892, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00019526192504258944, | |
| "loss": 2.1984, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.00019524861584327088, | |
| "loss": 2.2268, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 0.00019523530664395232, | |
| "loss": 2.0896, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 0.00019522199744463373, | |
| "loss": 2.146, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.00019520868824531517, | |
| "loss": 2.2207, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0001951953790459966, | |
| "loss": 2.2925, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 0.00019518206984667805, | |
| "loss": 2.1177, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.00019516876064735948, | |
| "loss": 2.1932, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 0.0001951554514480409, | |
| "loss": 2.1481, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 0.00019514214224872233, | |
| "loss": 2.0231, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00019512883304940374, | |
| "loss": 2.2687, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.00019511552385008518, | |
| "loss": 2.0866, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 0.00019510221465076662, | |
| "loss": 2.1907, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.00019508890545144806, | |
| "loss": 2.1635, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 0.00019507559625212947, | |
| "loss": 2.0831, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0001950622870528109, | |
| "loss": 2.1906, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.00019504897785349235, | |
| "loss": 2.2034, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 0.00019503566865417378, | |
| "loss": 2.2103, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 0.0001950223594548552, | |
| "loss": 2.1678, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.00019500905025553663, | |
| "loss": 2.0417, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 0.00019499574105621807, | |
| "loss": 2.1806, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0001949824318568995, | |
| "loss": 2.21, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00019496912265758095, | |
| "loss": 2.1416, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 0.00019495581345826236, | |
| "loss": 2.1792, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 0.00019494250425894377, | |
| "loss": 2.2475, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.0001949291950596252, | |
| "loss": 2.1561, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 0.00019491588586030665, | |
| "loss": 2.1854, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.00019490257666098808, | |
| "loss": 2.2039, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.00019488926746166952, | |
| "loss": 2.2249, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 0.00019487595826235093, | |
| "loss": 2.0889, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 0.00019486264906303237, | |
| "loss": 2.2046, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.0001948493398637138, | |
| "loss": 2.0463, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.00019483603066439525, | |
| "loss": 2.0919, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 0.00019482272146507669, | |
| "loss": 2.1278, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.0001948094122657581, | |
| "loss": 2.1866, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 0.00019479610306643953, | |
| "loss": 2.25, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 0.00019478279386712097, | |
| "loss": 2.038, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.0001947694846678024, | |
| "loss": 2.0665, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.00019475617546848382, | |
| "loss": 2.0884, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 0.00019474286626916526, | |
| "loss": 2.1792, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.00019472955706984667, | |
| "loss": 2.1597, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 0.0001947162478705281, | |
| "loss": 2.1973, | |
| "step": 198500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 0.00019470293867120955, | |
| "loss": 2.1074, | |
| "step": 199000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.00019468962947189099, | |
| "loss": 2.1212, | |
| "step": 199500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0001946763202725724, | |
| "loss": 2.1406, | |
| "step": 200000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 0.00019466301107325383, | |
| "loss": 2.1905, | |
| "step": 200500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.00019464970187393527, | |
| "loss": 2.0908, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 0.0001946363926746167, | |
| "loss": 2.1969, | |
| "step": 201500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 0.00019462308347529815, | |
| "loss": 2.246, | |
| "step": 202000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.00019460977427597956, | |
| "loss": 2.0437, | |
| "step": 202500 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.000194596465076661, | |
| "loss": 2.2312, | |
| "step": 203000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 0.00019458315587734244, | |
| "loss": 2.182, | |
| "step": 203500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.00019456984667802387, | |
| "loss": 2.2265, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 0.00019455653747870529, | |
| "loss": 2.1802, | |
| "step": 204500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.00019454322827938672, | |
| "loss": 2.1688, | |
| "step": 205000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.00019452991908006813, | |
| "loss": 2.1946, | |
| "step": 205500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 0.00019451660988074957, | |
| "loss": 2.2045, | |
| "step": 206000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 0.000194503300681431, | |
| "loss": 2.116, | |
| "step": 206500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.00019448999148211245, | |
| "loss": 2.046, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 0.0001944766822827939, | |
| "loss": 2.1339, | |
| "step": 207500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0001944633730834753, | |
| "loss": 2.2015, | |
| "step": 208000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.00019445006388415674, | |
| "loss": 2.1834, | |
| "step": 208500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 0.00019443675468483817, | |
| "loss": 2.147, | |
| "step": 209000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.0001944234454855196, | |
| "loss": 2.1619, | |
| "step": 209500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 0.00019441013628620105, | |
| "loss": 2.184, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00019439682708688246, | |
| "loss": 2.1712, | |
| "step": 210500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0001943835178875639, | |
| "loss": 2.2269, | |
| "step": 211000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 0.0001943702086882453, | |
| "loss": 2.1537, | |
| "step": 211500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 0.00019435689948892675, | |
| "loss": 2.1268, | |
| "step": 212000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0001943435902896082, | |
| "loss": 2.1349, | |
| "step": 212500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 0.0001943302810902896, | |
| "loss": 2.0699, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 0.00019431697189097104, | |
| "loss": 2.1889, | |
| "step": 213500 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.00019430366269165248, | |
| "loss": 2.1516, | |
| "step": 214000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0001942903534923339, | |
| "loss": 2.1827, | |
| "step": 214500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 0.00019427704429301535, | |
| "loss": 2.2482, | |
| "step": 215000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.00019426373509369676, | |
| "loss": 2.2037, | |
| "step": 215500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.0001942504258943782, | |
| "loss": 2.1865, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 0.00019423711669505964, | |
| "loss": 2.1926, | |
| "step": 216500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.00019422380749574108, | |
| "loss": 2.2028, | |
| "step": 217000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 0.00019421049829642252, | |
| "loss": 2.2157, | |
| "step": 217500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 0.00019419718909710393, | |
| "loss": 2.1386, | |
| "step": 218000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.00019418387989778534, | |
| "loss": 2.1053, | |
| "step": 218500 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 0.00019417057069846678, | |
| "loss": 2.1686, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.0001941572614991482, | |
| "loss": 2.1914, | |
| "step": 219500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.00019414395229982965, | |
| "loss": 2.2137, | |
| "step": 220000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 0.0001941306431005111, | |
| "loss": 2.0708, | |
| "step": 220500 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.0001941173339011925, | |
| "loss": 2.1136, | |
| "step": 221000 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.00019410402470187394, | |
| "loss": 2.071, | |
| "step": 221500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 0.00019409071550255538, | |
| "loss": 2.1486, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 0.00019407740630323682, | |
| "loss": 2.1378, | |
| "step": 222500 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.00019406409710391825, | |
| "loss": 2.0198, | |
| "step": 223000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 0.00019405078790459966, | |
| "loss": 2.2202, | |
| "step": 223500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 0.0001940374787052811, | |
| "loss": 2.1, | |
| "step": 224000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.00019402416950596254, | |
| "loss": 2.1922, | |
| "step": 224500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 0.00019401086030664398, | |
| "loss": 2.1636, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_loss": 2.014349937438965, | |
| "eval_runtime": 150.1185, | |
| "eval_samples_per_second": 62.571, | |
| "eval_steps_per_second": 62.571, | |
| "step": 225408 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.0001939975511073254, | |
| "loss": 2.0504, | |
| "step": 225500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.0001939842419080068, | |
| "loss": 1.995, | |
| "step": 226000 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.00019397093270868824, | |
| "loss": 1.9558, | |
| "step": 226500 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 0.00019395762350936968, | |
| "loss": 1.9273, | |
| "step": 227000 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 0.00019394431431005112, | |
| "loss": 1.9675, | |
| "step": 227500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 0.00019393100511073255, | |
| "loss": 1.9863, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 3.04, | |
| "learning_rate": 0.00019391769591141396, | |
| "loss": 1.9841, | |
| "step": 228500 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.0001939043867120954, | |
| "loss": 1.9306, | |
| "step": 229000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 0.00019389107751277684, | |
| "loss": 1.9797, | |
| "step": 229500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 0.00019387776831345828, | |
| "loss": 1.8735, | |
| "step": 230000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.00019386445911413972, | |
| "loss": 1.9623, | |
| "step": 230500 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.00019385114991482113, | |
| "loss": 1.9795, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 3.08, | |
| "learning_rate": 0.00019383784071550257, | |
| "loss": 1.862, | |
| "step": 231500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.000193824531516184, | |
| "loss": 2.0137, | |
| "step": 232000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.00019381122231686544, | |
| "loss": 1.9126, | |
| "step": 232500 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 0.00019379791311754685, | |
| "loss": 1.9946, | |
| "step": 233000 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 0.0001937846039182283, | |
| "loss": 1.9526, | |
| "step": 233500 | |
| }, | |
| { | |
| "epoch": 3.11, | |
| "learning_rate": 0.0001937712947189097, | |
| "loss": 1.954, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 0.00019375798551959114, | |
| "loss": 1.9714, | |
| "step": 234500 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.00019374467632027258, | |
| "loss": 1.917, | |
| "step": 235000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 0.00019373136712095402, | |
| "loss": 2.0031, | |
| "step": 235500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 0.00019371805792163546, | |
| "loss": 1.9253, | |
| "step": 236000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.00019370474872231687, | |
| "loss": 1.8825, | |
| "step": 236500 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.0001936914395229983, | |
| "loss": 1.9689, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 0.00019367813032367974, | |
| "loss": 2.0113, | |
| "step": 237500 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 0.00019366482112436118, | |
| "loss": 2.0767, | |
| "step": 238000 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 0.00019365151192504262, | |
| "loss": 2.0196, | |
| "step": 238500 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 0.00019363820272572403, | |
| "loss": 1.9681, | |
| "step": 239000 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00019362489352640547, | |
| "loss": 2.0794, | |
| "step": 239500 | |
| }, | |
| { | |
| "epoch": 3.19, | |
| "learning_rate": 0.00019361158432708688, | |
| "loss": 1.9347, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 0.00019359827512776832, | |
| "loss": 1.8842, | |
| "step": 240500 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 0.00019358496592844976, | |
| "loss": 2.1145, | |
| "step": 241000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 0.00019357165672913117, | |
| "loss": 2.0187, | |
| "step": 241500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0001935583475298126, | |
| "loss": 1.9615, | |
| "step": 242000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 0.00019354503833049404, | |
| "loss": 2.0395, | |
| "step": 242500 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 0.00019353172913117548, | |
| "loss": 1.9914, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 0.00019351841993185692, | |
| "loss": 1.9202, | |
| "step": 243500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 0.00019350511073253833, | |
| "loss": 1.9856, | |
| "step": 244000 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 0.00019349180153321977, | |
| "loss": 1.9678, | |
| "step": 244500 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 0.0001934784923339012, | |
| "loss": 1.9491, | |
| "step": 245000 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 0.00019346518313458264, | |
| "loss": 2.048, | |
| "step": 245500 | |
| }, | |
| { | |
| "epoch": 3.27, | |
| "learning_rate": 0.00019345187393526408, | |
| "loss": 2.0294, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 0.0001934385647359455, | |
| "loss": 1.9502, | |
| "step": 246500 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 0.0001934252555366269, | |
| "loss": 1.9172, | |
| "step": 247000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 0.00019341194633730834, | |
| "loss": 2.0995, | |
| "step": 247500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.00019339863713798978, | |
| "loss": 1.8991, | |
| "step": 248000 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 0.00019338532793867122, | |
| "loss": 1.9673, | |
| "step": 248500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 0.00019337201873935266, | |
| "loss": 1.8967, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 0.00019335870954003407, | |
| "loss": 1.9235, | |
| "step": 249500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.0001933454003407155, | |
| "loss": 2.0222, | |
| "step": 250000 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 0.00019333209114139695, | |
| "loss": 1.9349, | |
| "step": 250500 | |
| }, | |
| { | |
| "epoch": 3.34, | |
| "learning_rate": 0.00019331878194207838, | |
| "loss": 1.9891, | |
| "step": 251000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 0.0001933054727427598, | |
| "loss": 2.0221, | |
| "step": 251500 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 0.00019329216354344123, | |
| "loss": 1.92, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 0.00019327885434412267, | |
| "loss": 1.9894, | |
| "step": 252500 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.0001932655451448041, | |
| "loss": 2.1004, | |
| "step": 253000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.00019325223594548555, | |
| "loss": 2.0359, | |
| "step": 253500 | |
| }, | |
| { | |
| "epoch": 3.38, | |
| "learning_rate": 0.00019323892674616696, | |
| "loss": 1.9138, | |
| "step": 254000 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 0.00019322561754684837, | |
| "loss": 2.0319, | |
| "step": 254500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 0.0001932123083475298, | |
| "loss": 1.9884, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 0.00019319899914821125, | |
| "loss": 1.9501, | |
| "step": 255500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 0.00019318568994889268, | |
| "loss": 1.9661, | |
| "step": 256000 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 0.00019317238074957412, | |
| "loss": 2.1242, | |
| "step": 256500 | |
| }, | |
| { | |
| "epoch": 3.42, | |
| "learning_rate": 0.00019315907155025553, | |
| "loss": 1.9356, | |
| "step": 257000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 0.00019314576235093697, | |
| "loss": 2.0207, | |
| "step": 257500 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 0.0001931324531516184, | |
| "loss": 2.0019, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 0.00019311914395229985, | |
| "loss": 1.9281, | |
| "step": 258500 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 0.00019310583475298129, | |
| "loss": 2.0126, | |
| "step": 259000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 0.0001930925255536627, | |
| "loss": 1.9973, | |
| "step": 259500 | |
| }, | |
| { | |
| "epoch": 3.46, | |
| "learning_rate": 0.00019307921635434413, | |
| "loss": 2.0036, | |
| "step": 260000 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 0.00019306590715502557, | |
| "loss": 1.9307, | |
| "step": 260500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 0.000193052597955707, | |
| "loss": 2.0548, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 0.00019303928875638842, | |
| "loss": 1.9384, | |
| "step": 261500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 0.00019302597955706986, | |
| "loss": 2.0036, | |
| "step": 262000 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 0.00019301267035775127, | |
| "loss": 1.9239, | |
| "step": 262500 | |
| }, | |
| { | |
| "epoch": 3.5, | |
| "learning_rate": 0.0001929993611584327, | |
| "loss": 1.9539, | |
| "step": 263000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 0.00019298605195911415, | |
| "loss": 2.0679, | |
| "step": 263500 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 0.00019297274275979559, | |
| "loss": 1.9274, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 0.000192959433560477, | |
| "loss": 1.935, | |
| "step": 264500 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 0.00019294612436115843, | |
| "loss": 2.1159, | |
| "step": 265000 | |
| }, | |
| { | |
| "epoch": 3.53, | |
| "learning_rate": 0.00019293281516183987, | |
| "loss": 2.0521, | |
| "step": 265500 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 0.0001929195059625213, | |
| "loss": 2.1259, | |
| "step": 266000 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 0.00019290619676320275, | |
| "loss": 2.0283, | |
| "step": 266500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 0.00019289288756388416, | |
| "loss": 2.0267, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 0.0001928795783645656, | |
| "loss": 1.9899, | |
| "step": 267500 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.00019286626916524704, | |
| "loss": 1.9344, | |
| "step": 268000 | |
| }, | |
| { | |
| "epoch": 3.57, | |
| "learning_rate": 0.00019285295996592845, | |
| "loss": 2.1067, | |
| "step": 268500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 0.00019283965076660989, | |
| "loss": 2.0983, | |
| "step": 269000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 0.00019282634156729132, | |
| "loss": 2.0504, | |
| "step": 269500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 0.00019281303236797273, | |
| "loss": 1.9778, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 0.00019279972316865417, | |
| "loss": 2.0056, | |
| "step": 270500 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 0.0001927864139693356, | |
| "loss": 1.9893, | |
| "step": 271000 | |
| }, | |
| { | |
| "epoch": 3.61, | |
| "learning_rate": 0.00019277310477001705, | |
| "loss": 1.9795, | |
| "step": 271500 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 0.0001927597955706985, | |
| "loss": 1.9398, | |
| "step": 272000 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 0.0001927464863713799, | |
| "loss": 2.0212, | |
| "step": 272500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 0.00019273317717206134, | |
| "loss": 1.9342, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 0.00019271986797274277, | |
| "loss": 1.9867, | |
| "step": 273500 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 0.0001927065587734242, | |
| "loss": 2.0927, | |
| "step": 274000 | |
| }, | |
| { | |
| "epoch": 3.65, | |
| "learning_rate": 0.00019269324957410565, | |
| "loss": 2.0769, | |
| "step": 274500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 0.00019267994037478706, | |
| "loss": 1.9088, | |
| "step": 275000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.0001926666311754685, | |
| "loss": 2.068, | |
| "step": 275500 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.0001926533219761499, | |
| "loss": 1.9976, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 0.00019264001277683135, | |
| "loss": 1.9657, | |
| "step": 276500 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 0.0001926267035775128, | |
| "loss": 2.0356, | |
| "step": 277000 | |
| }, | |
| { | |
| "epoch": 3.69, | |
| "learning_rate": 0.0001926133943781942, | |
| "loss": 1.9525, | |
| "step": 277500 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 0.00019260008517887564, | |
| "loss": 1.9918, | |
| "step": 278000 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 0.00019258677597955707, | |
| "loss": 1.9586, | |
| "step": 278500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 0.0001925734667802385, | |
| "loss": 1.9425, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 3.72, | |
| "learning_rate": 0.00019256015758091995, | |
| "loss": 1.9502, | |
| "step": 279500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 0.00019254684838160136, | |
| "loss": 1.8562, | |
| "step": 280000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 0.0001925335391822828, | |
| "loss": 2.0119, | |
| "step": 280500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 0.00019252022998296424, | |
| "loss": 2.0455, | |
| "step": 281000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.00019250692078364568, | |
| "loss": 2.039, | |
| "step": 281500 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.00019249361158432711, | |
| "loss": 2.0412, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 3.76, | |
| "learning_rate": 0.00019248030238500853, | |
| "loss": 2.1478, | |
| "step": 282500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00019246699318568994, | |
| "loss": 1.941, | |
| "step": 283000 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 0.00019245368398637138, | |
| "loss": 1.9749, | |
| "step": 283500 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 0.0001924403747870528, | |
| "loss": 2.0763, | |
| "step": 284000 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 0.00019242706558773425, | |
| "loss": 2.0318, | |
| "step": 284500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 0.0001924137563884157, | |
| "loss": 2.0111, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 0.0001924004471890971, | |
| "loss": 1.9364, | |
| "step": 285500 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.00019238713798977854, | |
| "loss": 2.0628, | |
| "step": 286000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 0.00019237382879045998, | |
| "loss": 2.1622, | |
| "step": 286500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 0.00019236051959114142, | |
| "loss": 2.049, | |
| "step": 287000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00019234721039182285, | |
| "loss": 2.0455, | |
| "step": 287500 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 0.00019233390119250426, | |
| "loss": 1.991, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 3.84, | |
| "learning_rate": 0.0001923205919931857, | |
| "loss": 1.9896, | |
| "step": 288500 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 0.00019230728279386714, | |
| "loss": 2.1045, | |
| "step": 289000 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 0.00019229397359454858, | |
| "loss": 2.0259, | |
| "step": 289500 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 0.00019228066439523, | |
| "loss": 2.0056, | |
| "step": 290000 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 0.0001922673551959114, | |
| "loss": 1.9663, | |
| "step": 290500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 0.00019225404599659284, | |
| "loss": 2.0817, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 0.00019224073679727428, | |
| "loss": 2.1459, | |
| "step": 291500 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 0.00019222742759795572, | |
| "loss": 2.0219, | |
| "step": 292000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 0.00019221411839863715, | |
| "loss": 2.0752, | |
| "step": 292500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 0.00019220080919931856, | |
| "loss": 2.0652, | |
| "step": 293000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.0001921875, | |
| "loss": 2.0201, | |
| "step": 293500 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 0.00019217419080068144, | |
| "loss": 2.018, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 3.92, | |
| "learning_rate": 0.00019216088160136288, | |
| "loss": 2.0404, | |
| "step": 294500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00019214757240204432, | |
| "loss": 2.0208, | |
| "step": 295000 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 0.00019213426320272573, | |
| "loss": 1.9643, | |
| "step": 295500 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 0.00019212095400340717, | |
| "loss": 2.0174, | |
| "step": 296000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.0001921076448040886, | |
| "loss": 1.991, | |
| "step": 296500 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.00019209433560477004, | |
| "loss": 2.0344, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 0.00019208102640545145, | |
| "loss": 2.1951, | |
| "step": 297500 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.0001920677172061329, | |
| "loss": 2.0148, | |
| "step": 298000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.0001920544080068143, | |
| "loss": 1.9864, | |
| "step": 298500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 0.00019204109880749574, | |
| "loss": 2.0757, | |
| "step": 299000 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 0.00019202778960817718, | |
| "loss": 2.0382, | |
| "step": 299500 | |
| }, | |
| { | |
| "epoch": 3.99, | |
| "learning_rate": 0.00019201448040885862, | |
| "loss": 2.0325, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 0.00019200117120954006, | |
| "loss": 2.0484, | |
| "step": 300500 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "eval_loss": 1.9609955549240112, | |
| "eval_runtime": 152.8744, | |
| "eval_samples_per_second": 61.443, | |
| "eval_steps_per_second": 61.443, | |
| "step": 300544 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.00019198786201022147, | |
| "loss": 1.835, | |
| "step": 301000 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 0.0001919745528109029, | |
| "loss": 1.8727, | |
| "step": 301500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 0.00019196124361158434, | |
| "loss": 1.7339, | |
| "step": 302000 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 0.00019194793441226578, | |
| "loss": 1.8193, | |
| "step": 302500 | |
| }, | |
| { | |
| "epoch": 4.03, | |
| "learning_rate": 0.00019193462521294722, | |
| "loss": 1.8356, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 0.00019192131601362863, | |
| "loss": 1.8941, | |
| "step": 303500 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.00019190800681431007, | |
| "loss": 1.8211, | |
| "step": 304000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.00019189469761499148, | |
| "loss": 1.9222, | |
| "step": 304500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 0.00019188138841567292, | |
| "loss": 1.7926, | |
| "step": 305000 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.00019186807921635436, | |
| "loss": 1.8561, | |
| "step": 305500 | |
| }, | |
| { | |
| "epoch": 4.07, | |
| "learning_rate": 0.00019185477001703577, | |
| "loss": 1.7667, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 0.0001918414608177172, | |
| "loss": 1.803, | |
| "step": 306500 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.00019182815161839864, | |
| "loss": 1.7807, | |
| "step": 307000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.00019181484241908008, | |
| "loss": 1.8113, | |
| "step": 307500 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 0.00019180153321976152, | |
| "loss": 1.8102, | |
| "step": 308000 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 0.00019178822402044293, | |
| "loss": 1.8823, | |
| "step": 308500 | |
| }, | |
| { | |
| "epoch": 4.11, | |
| "learning_rate": 0.00019177491482112437, | |
| "loss": 1.7312, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 0.0001917616056218058, | |
| "loss": 1.8165, | |
| "step": 309500 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 0.00019174829642248724, | |
| "loss": 1.8188, | |
| "step": 310000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 0.00019173498722316868, | |
| "loss": 1.9123, | |
| "step": 310500 | |
| }, | |
| { | |
| "epoch": 4.14, | |
| "learning_rate": 0.0001917216780238501, | |
| "loss": 1.8474, | |
| "step": 311000 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.0001917083688245315, | |
| "loss": 1.8434, | |
| "step": 311500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 0.00019169505962521294, | |
| "loss": 1.832, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 0.00019168175042589438, | |
| "loss": 1.7631, | |
| "step": 312500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 0.00019166844122657582, | |
| "loss": 1.8939, | |
| "step": 313000 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 0.00019165513202725726, | |
| "loss": 1.8447, | |
| "step": 313500 | |
| }, | |
| { | |
| "epoch": 4.18, | |
| "learning_rate": 0.00019164182282793867, | |
| "loss": 1.8582, | |
| "step": 314000 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 0.0001916285136286201, | |
| "loss": 1.8515, | |
| "step": 314500 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 0.00019161520442930154, | |
| "loss": 1.7869, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.00019160189522998298, | |
| "loss": 1.9908, | |
| "step": 315500 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 0.0001915885860306644, | |
| "loss": 1.8234, | |
| "step": 316000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 0.00019157527683134583, | |
| "loss": 1.8885, | |
| "step": 316500 | |
| }, | |
| { | |
| "epoch": 4.22, | |
| "learning_rate": 0.00019156196763202727, | |
| "loss": 1.8792, | |
| "step": 317000 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 0.0001915486584327087, | |
| "loss": 1.9137, | |
| "step": 317500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 0.00019153534923339015, | |
| "loss": 1.8785, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 0.00019152204003407156, | |
| "loss": 1.8575, | |
| "step": 318500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.00019150873083475297, | |
| "loss": 1.8303, | |
| "step": 319000 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 0.0001914954216354344, | |
| "loss": 1.7941, | |
| "step": 319500 | |
| }, | |
| { | |
| "epoch": 4.26, | |
| "learning_rate": 0.00019148211243611585, | |
| "loss": 1.9304, | |
| "step": 320000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.00019146880323679728, | |
| "loss": 1.8178, | |
| "step": 320500 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.00019145549403747872, | |
| "loss": 1.9175, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 0.00019144218483816013, | |
| "loss": 1.8657, | |
| "step": 321500 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 0.00019142887563884157, | |
| "loss": 1.821, | |
| "step": 322000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 0.000191415566439523, | |
| "loss": 1.7716, | |
| "step": 322500 | |
| }, | |
| { | |
| "epoch": 4.3, | |
| "learning_rate": 0.00019140225724020445, | |
| "loss": 1.8235, | |
| "step": 323000 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 0.00019138894804088589, | |
| "loss": 1.8991, | |
| "step": 323500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 0.0001913756388415673, | |
| "loss": 1.9076, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 0.00019136232964224873, | |
| "loss": 1.9073, | |
| "step": 324500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 0.00019134902044293017, | |
| "loss": 2.0163, | |
| "step": 325000 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 0.0001913357112436116, | |
| "loss": 1.8352, | |
| "step": 325500 | |
| }, | |
| { | |
| "epoch": 4.34, | |
| "learning_rate": 0.00019132240204429302, | |
| "loss": 1.8498, | |
| "step": 326000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.00019130909284497446, | |
| "loss": 1.9367, | |
| "step": 326500 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.00019129578364565587, | |
| "loss": 1.847, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 0.0001912824744463373, | |
| "loss": 1.8725, | |
| "step": 327500 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 0.00019126916524701875, | |
| "loss": 1.9263, | |
| "step": 328000 | |
| }, | |
| { | |
| "epoch": 4.37, | |
| "learning_rate": 0.00019125585604770019, | |
| "loss": 1.8871, | |
| "step": 328500 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 0.0001912425468483816, | |
| "loss": 1.8387, | |
| "step": 329000 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 0.00019122923764906303, | |
| "loss": 1.8359, | |
| "step": 329500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 0.00019121592844974447, | |
| "loss": 1.967, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 0.0001912026192504259, | |
| "loss": 1.9031, | |
| "step": 330500 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 0.00019118931005110735, | |
| "loss": 1.9016, | |
| "step": 331000 | |
| }, | |
| { | |
| "epoch": 4.41, | |
| "learning_rate": 0.00019117600085178876, | |
| "loss": 1.8559, | |
| "step": 331500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 0.0001911626916524702, | |
| "loss": 1.8603, | |
| "step": 332000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 0.00019114938245315164, | |
| "loss": 1.9245, | |
| "step": 332500 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 0.00019113607325383305, | |
| "loss": 1.8627, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 0.00019112276405451449, | |
| "loss": 1.8681, | |
| "step": 333500 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 0.00019110945485519592, | |
| "loss": 1.9043, | |
| "step": 334000 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 0.00019109614565587733, | |
| "loss": 1.8662, | |
| "step": 334500 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 0.00019108283645655877, | |
| "loss": 1.9518, | |
| "step": 335000 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 0.0001910695272572402, | |
| "loss": 1.8168, | |
| "step": 335500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 0.00019105621805792165, | |
| "loss": 1.9287, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 0.0001910429088586031, | |
| "loss": 1.8791, | |
| "step": 336500 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.0001910295996592845, | |
| "loss": 1.8938, | |
| "step": 337000 | |
| }, | |
| { | |
| "epoch": 4.49, | |
| "learning_rate": 0.00019101629045996594, | |
| "loss": 1.9088, | |
| "step": 337500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 0.00019100298126064737, | |
| "loss": 1.9197, | |
| "step": 338000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 0.0001909896720613288, | |
| "loss": 2.082, | |
| "step": 338500 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 0.00019097636286201025, | |
| "loss": 1.8672, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.00019096305366269166, | |
| "loss": 1.9084, | |
| "step": 339500 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 0.0001909497444633731, | |
| "loss": 1.912, | |
| "step": 340000 | |
| }, | |
| { | |
| "epoch": 4.53, | |
| "learning_rate": 0.0001909364352640545, | |
| "loss": 1.9065, | |
| "step": 340500 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 0.00019092312606473595, | |
| "loss": 1.819, | |
| "step": 341000 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 0.0001909098168654174, | |
| "loss": 1.9697, | |
| "step": 341500 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 0.0001908965076660988, | |
| "loss": 1.9895, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 4.56, | |
| "learning_rate": 0.00019088319846678024, | |
| "loss": 1.933, | |
| "step": 342500 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 0.00019086988926746167, | |
| "loss": 1.8534, | |
| "step": 343000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 0.0001908565800681431, | |
| "loss": 1.948, | |
| "step": 343500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 0.00019084327086882455, | |
| "loss": 1.9291, | |
| "step": 344000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.00019082996166950596, | |
| "loss": 1.813, | |
| "step": 344500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.0001908166524701874, | |
| "loss": 1.9467, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 0.00019080334327086884, | |
| "loss": 1.9648, | |
| "step": 345500 | |
| }, | |
| { | |
| "epoch": 4.6, | |
| "learning_rate": 0.00019079003407155028, | |
| "loss": 1.9408, | |
| "step": 346000 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 0.00019077672487223171, | |
| "loss": 2.0716, | |
| "step": 346500 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 0.00019076341567291313, | |
| "loss": 1.8179, | |
| "step": 347000 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 0.00019075010647359454, | |
| "loss": 1.9267, | |
| "step": 347500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 0.00019073679727427598, | |
| "loss": 1.8477, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.0001907234880749574, | |
| "loss": 1.8807, | |
| "step": 348500 | |
| }, | |
| { | |
| "epoch": 4.64, | |
| "learning_rate": 0.00019071017887563885, | |
| "loss": 1.9378, | |
| "step": 349000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 0.0001906968696763203, | |
| "loss": 1.9433, | |
| "step": 349500 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 0.0001906835604770017, | |
| "loss": 1.9764, | |
| "step": 350000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 0.00019067025127768314, | |
| "loss": 1.9097, | |
| "step": 350500 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 0.00019065694207836458, | |
| "loss": 1.8964, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 0.00019064363287904601, | |
| "loss": 1.8199, | |
| "step": 351500 | |
| }, | |
| { | |
| "epoch": 4.68, | |
| "learning_rate": 0.00019063032367972745, | |
| "loss": 1.8813, | |
| "step": 352000 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 0.00019061701448040886, | |
| "loss": 1.9906, | |
| "step": 352500 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 0.0001906037052810903, | |
| "loss": 1.9196, | |
| "step": 353000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 0.00019059039608177174, | |
| "loss": 2.023, | |
| "step": 353500 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 0.00019057708688245318, | |
| "loss": 2.0203, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.0001905637776831346, | |
| "loss": 1.9758, | |
| "step": 354500 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.000190550468483816, | |
| "loss": 1.8171, | |
| "step": 355000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 0.00019053715928449744, | |
| "loss": 2.0035, | |
| "step": 355500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 0.00019052385008517888, | |
| "loss": 1.9566, | |
| "step": 356000 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 0.00019051054088586032, | |
| "loss": 1.8568, | |
| "step": 356500 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 0.00019049723168654175, | |
| "loss": 2.0018, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 0.00019048392248722316, | |
| "loss": 1.8047, | |
| "step": 357500 | |
| }, | |
| { | |
| "epoch": 4.76, | |
| "learning_rate": 0.0001904706132879046, | |
| "loss": 1.8328, | |
| "step": 358000 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 0.00019045730408858604, | |
| "loss": 1.9396, | |
| "step": 358500 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00019044399488926748, | |
| "loss": 1.8235, | |
| "step": 359000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 0.00019043068568994892, | |
| "loss": 1.8668, | |
| "step": 359500 | |
| }, | |
| { | |
| "epoch": 4.79, | |
| "learning_rate": 0.00019041737649063033, | |
| "loss": 2.0224, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 0.00019040406729131177, | |
| "loss": 1.9938, | |
| "step": 360500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 0.0001903907580919932, | |
| "loss": 1.9797, | |
| "step": 361000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 0.00019037744889267464, | |
| "loss": 1.9694, | |
| "step": 361500 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 0.00019036413969335605, | |
| "loss": 2.0293, | |
| "step": 362000 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 0.0001903508304940375, | |
| "loss": 1.9625, | |
| "step": 362500 | |
| }, | |
| { | |
| "epoch": 4.83, | |
| "learning_rate": 0.0001903375212947189, | |
| "loss": 1.9487, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 0.00019032421209540034, | |
| "loss": 1.9184, | |
| "step": 363500 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 0.00019031090289608178, | |
| "loss": 2.0013, | |
| "step": 364000 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 0.00019029759369676322, | |
| "loss": 1.9245, | |
| "step": 364500 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 0.00019028428449744466, | |
| "loss": 1.8413, | |
| "step": 365000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 0.00019027097529812607, | |
| "loss": 1.8837, | |
| "step": 365500 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 0.0001902576660988075, | |
| "loss": 2.0188, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 0.00019024435689948894, | |
| "loss": 1.9003, | |
| "step": 366500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 0.00019023104770017038, | |
| "loss": 1.9519, | |
| "step": 367000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 0.0001902177385008518, | |
| "loss": 1.9312, | |
| "step": 367500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 0.00019020442930153323, | |
| "loss": 1.8173, | |
| "step": 368000 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 0.00019019112010221467, | |
| "loss": 1.9114, | |
| "step": 368500 | |
| }, | |
| { | |
| "epoch": 4.91, | |
| "learning_rate": 0.00019017781090289608, | |
| "loss": 1.9343, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 0.00019016450170357752, | |
| "loss": 1.906, | |
| "step": 369500 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 0.00019015119250425896, | |
| "loss": 1.9389, | |
| "step": 370000 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 0.00019013788330494037, | |
| "loss": 1.917, | |
| "step": 370500 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 0.0001901245741056218, | |
| "loss": 1.9064, | |
| "step": 371000 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 0.00019011126490630324, | |
| "loss": 1.9388, | |
| "step": 371500 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.00019009795570698468, | |
| "loss": 1.8737, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.00019008464650766612, | |
| "loss": 1.9446, | |
| "step": 372500 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 0.00019007133730834753, | |
| "loss": 1.9625, | |
| "step": 373000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 0.00019005802810902897, | |
| "loss": 1.8929, | |
| "step": 373500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 0.0001900447189097104, | |
| "loss": 1.9421, | |
| "step": 374000 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 0.00019003140971039184, | |
| "loss": 2.0236, | |
| "step": 374500 | |
| }, | |
| { | |
| "epoch": 4.99, | |
| "learning_rate": 0.00019001810051107328, | |
| "loss": 2.0012, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 0.0001900047913117547, | |
| "loss": 2.0569, | |
| "step": 375500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "eval_loss": 1.9444704055786133, | |
| "eval_runtime": 150.8043, | |
| "eval_samples_per_second": 62.286, | |
| "eval_steps_per_second": 62.286, | |
| "step": 375680 | |
| } | |
| ], | |
| "max_steps": 7513600, | |
| "num_train_epochs": 100, | |
| "total_flos": 6244677543475200.0, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |