| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.0, | |
| "global_step": 198420, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9874004636629375e-05, | |
| "loss": 2.688, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.974800927325874e-05, | |
| "loss": 2.2881, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 4.962201390988812e-05, | |
| "loss": 2.1112, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.949601854651749e-05, | |
| "loss": 1.9839, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.9370023183146866e-05, | |
| "loss": 1.9138, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.924402781977623e-05, | |
| "loss": 1.8549, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.9118032456405605e-05, | |
| "loss": 1.8107, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 4.899203709303498e-05, | |
| "loss": 1.7686, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.886604172966435e-05, | |
| "loss": 1.7388, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.874004636629372e-05, | |
| "loss": 1.6948, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.8614051002923096e-05, | |
| "loss": 1.6784, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.848805563955247e-05, | |
| "loss": 1.6304, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "learning_rate": 4.8362060276181834e-05, | |
| "loss": 1.6583, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8236064912811214e-05, | |
| "loss": 1.5898, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.811006954944059e-05, | |
| "loss": 1.5856, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.798407418606995e-05, | |
| "loss": 1.5539, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 4.7858078822699325e-05, | |
| "loss": 1.5607, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.77320834593287e-05, | |
| "loss": 1.5663, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.760608809595807e-05, | |
| "loss": 1.5291, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.7480092732587444e-05, | |
| "loss": 1.5196, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.7354097369216816e-05, | |
| "loss": 1.5014, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.722810200584619e-05, | |
| "loss": 1.4796, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 4.7102106642475555e-05, | |
| "loss": 1.4735, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.697611127910493e-05, | |
| "loss": 1.4725, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.685011591573431e-05, | |
| "loss": 1.4906, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.672412055236367e-05, | |
| "loss": 1.4598, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.6598125188993046e-05, | |
| "loss": 1.4293, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 4.647212982562242e-05, | |
| "loss": 1.4304, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.634613446225179e-05, | |
| "loss": 1.4401, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.6220139098881164e-05, | |
| "loss": 1.4463, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.609414373551054e-05, | |
| "loss": 1.3965, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.596814837213991e-05, | |
| "loss": 1.4147, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "learning_rate": 4.5842153008769276e-05, | |
| "loss": 1.3864, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.571615764539865e-05, | |
| "loss": 1.3718, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.559016228202803e-05, | |
| "loss": 1.3891, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.5464166918657394e-05, | |
| "loss": 1.3762, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.5338171555286767e-05, | |
| "loss": 1.3486, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.521217619191614e-05, | |
| "loss": 1.3795, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 4.508618082854551e-05, | |
| "loss": 1.3564, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.496018546517488e-05, | |
| "loss": 1.3572, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.483419010180426e-05, | |
| "loss": 1.3434, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 4.470819473843363e-05, | |
| "loss": 1.3337, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.4582199375063e-05, | |
| "loss": 1.3233, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.445620401169237e-05, | |
| "loss": 1.3345, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.433020864832174e-05, | |
| "loss": 1.3145, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.420421328495112e-05, | |
| "loss": 1.3329, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.407821792158049e-05, | |
| "loss": 1.3152, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 4.395222255820986e-05, | |
| "loss": 1.3237, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.382622719483923e-05, | |
| "loss": 1.3108, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.3700231831468605e-05, | |
| "loss": 1.3097, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.357423646809798e-05, | |
| "loss": 1.3101, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.344824110472735e-05, | |
| "loss": 1.3018, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "learning_rate": 4.3322245741356724e-05, | |
| "loss": 1.3023, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.319625037798609e-05, | |
| "loss": 1.2938, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.307025501461546e-05, | |
| "loss": 1.295, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.2944259651244835e-05, | |
| "loss": 1.2941, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.281826428787421e-05, | |
| "loss": 1.2825, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "learning_rate": 4.269226892450358e-05, | |
| "loss": 1.2581, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.256627356113295e-05, | |
| "loss": 1.2742, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.2440278197762326e-05, | |
| "loss": 1.2623, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.231428283439169e-05, | |
| "loss": 1.245, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.218828747102107e-05, | |
| "loss": 1.2664, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.2062292107650444e-05, | |
| "loss": 1.2687, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "learning_rate": 4.193629674427981e-05, | |
| "loss": 1.2493, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.181030138090918e-05, | |
| "loss": 1.2562, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.1684306017538556e-05, | |
| "loss": 1.2501, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.155831065416793e-05, | |
| "loss": 1.2298, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.14323152907973e-05, | |
| "loss": 1.2554, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 4.1306319927426674e-05, | |
| "loss": 1.2565, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.1180324564056046e-05, | |
| "loss": 1.2386, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.105432920068541e-05, | |
| "loss": 1.2129, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.0928333837314785e-05, | |
| "loss": 1.2246, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "learning_rate": 4.0802338473944165e-05, | |
| "loss": 1.2231, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.067634311057353e-05, | |
| "loss": 1.2392, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.05503477472029e-05, | |
| "loss": 1.2326, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.0424352383832276e-05, | |
| "loss": 1.2224, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.029835702046165e-05, | |
| "loss": 1.2076, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "learning_rate": 4.017236165709102e-05, | |
| "loss": 1.2093, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.0046366293720394e-05, | |
| "loss": 1.2156, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 3.992037093034977e-05, | |
| "loss": 1.2158, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 3.979437556697913e-05, | |
| "loss": 1.2221, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 3.9668380203608506e-05, | |
| "loss": 1.2, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "learning_rate": 3.9542384840237885e-05, | |
| "loss": 1.203, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.941638947686726e-05, | |
| "loss": 1.209, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 3.9290394113496624e-05, | |
| "loss": 1.184, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 3.9164398750126e-05, | |
| "loss": 1.2053, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 3.903840338675537e-05, | |
| "loss": 1.1807, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.891240802338474e-05, | |
| "loss": 1.1921, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 3.8786412660014115e-05, | |
| "loss": 1.1877, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 3.866041729664349e-05, | |
| "loss": 1.2047, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 3.853442193327286e-05, | |
| "loss": 1.1715, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.8408426569902226e-05, | |
| "loss": 1.1882, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 3.82824312065316e-05, | |
| "loss": 1.1839, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 3.815643584316098e-05, | |
| "loss": 1.1821, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 3.8030440479790345e-05, | |
| "loss": 1.1737, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.790444511641972e-05, | |
| "loss": 1.1848, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 3.777844975304909e-05, | |
| "loss": 1.1982, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "learning_rate": 3.765245438967846e-05, | |
| "loss": 1.1675, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 3.7526459026307835e-05, | |
| "loss": 1.1568, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.740046366293721e-05, | |
| "loss": 1.1629, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 3.727446829956658e-05, | |
| "loss": 1.1451, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 3.714847293619595e-05, | |
| "loss": 1.1649, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "learning_rate": 3.702247757282532e-05, | |
| "loss": 1.1627, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.689648220945469e-05, | |
| "loss": 1.1587, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 3.6770486846084065e-05, | |
| "loss": 1.1531, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 3.664449148271344e-05, | |
| "loss": 1.1359, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 3.651849611934281e-05, | |
| "loss": 1.1591, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.639250075597218e-05, | |
| "loss": 1.1475, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 3.626650539260155e-05, | |
| "loss": 1.1728, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 3.614051002923093e-05, | |
| "loss": 1.1666, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 3.60145146658603e-05, | |
| "loss": 1.1417, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.588851930248967e-05, | |
| "loss": 1.1421, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 3.576252393911904e-05, | |
| "loss": 1.1618, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 3.563652857574841e-05, | |
| "loss": 1.141, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 3.5510533212377786e-05, | |
| "loss": 1.1487, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.538453784900716e-05, | |
| "loss": 1.1327, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 3.525854248563653e-05, | |
| "loss": 1.1478, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 3.5132547122265904e-05, | |
| "loss": 1.1354, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 3.500655175889527e-05, | |
| "loss": 1.1326, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.488055639552464e-05, | |
| "loss": 1.1268, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 3.475456103215402e-05, | |
| "loss": 1.1475, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 3.4628565668783395e-05, | |
| "loss": 1.1112, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 3.450257030541276e-05, | |
| "loss": 1.1328, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4376574942042134e-05, | |
| "loss": 1.1271, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 3.4250579578671506e-05, | |
| "loss": 1.1272, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 3.412458421530088e-05, | |
| "loss": 1.1104, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 3.399858885193025e-05, | |
| "loss": 1.1017, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 3.3872593488559624e-05, | |
| "loss": 1.1137, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.3746598125189e-05, | |
| "loss": 1.1178, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 3.362060276181836e-05, | |
| "loss": 1.1247, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 3.3494607398447736e-05, | |
| "loss": 1.1114, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.3368612035077115e-05, | |
| "loss": 1.1013, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.324261667170648e-05, | |
| "loss": 1.032, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 3.3116621308335854e-05, | |
| "loss": 0.9762, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.299062594496523e-05, | |
| "loss": 0.9998, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.28646305815946e-05, | |
| "loss": 0.9951, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.273863521822397e-05, | |
| "loss": 0.984, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.2612639854853345e-05, | |
| "loss": 0.987, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 3.248664449148272e-05, | |
| "loss": 0.9793, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.2360649128112084e-05, | |
| "loss": 1.0024, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2234653764741457e-05, | |
| "loss": 1.0008, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.2108658401370836e-05, | |
| "loss": 0.996, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.19826630380002e-05, | |
| "loss": 0.9927, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.09, | |
| "learning_rate": 3.1856667674629575e-05, | |
| "loss": 0.9943, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.173067231125895e-05, | |
| "loss": 0.9905, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.160467694788832e-05, | |
| "loss": 0.997, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.1478681584517686e-05, | |
| "loss": 0.9831, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.1352686221147066e-05, | |
| "loss": 0.9863, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.122669085777644e-05, | |
| "loss": 0.9961, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 1.13, | |
| "learning_rate": 3.1100695494405804e-05, | |
| "loss": 0.985, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.097470013103518e-05, | |
| "loss": 0.9954, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.084870476766455e-05, | |
| "loss": 0.9813, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.072270940429392e-05, | |
| "loss": 0.9961, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.0596714040923295e-05, | |
| "loss": 0.9968, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 1.17, | |
| "learning_rate": 3.0470718677552668e-05, | |
| "loss": 1.0083, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.0344723314182037e-05, | |
| "loss": 0.9952, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.021872795081141e-05, | |
| "loss": 0.9919, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.0092732587440786e-05, | |
| "loss": 0.9836, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 2.9966737224070156e-05, | |
| "loss": 0.9868, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 2.984074186069953e-05, | |
| "loss": 1.012, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.9714746497328898e-05, | |
| "loss": 0.9844, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 2.958875113395827e-05, | |
| "loss": 0.9899, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 2.946275577058764e-05, | |
| "loss": 0.9693, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 1.24, | |
| "learning_rate": 2.933676040721702e-05, | |
| "loss": 0.9793, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.921076504384639e-05, | |
| "loss": 0.9868, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 2.908476968047576e-05, | |
| "loss": 0.9842, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 2.895877431710513e-05, | |
| "loss": 0.9685, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 2.88327789537345e-05, | |
| "loss": 0.9832, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 1.28, | |
| "learning_rate": 2.870678359036388e-05, | |
| "loss": 0.98, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.858078822699325e-05, | |
| "loss": 0.9895, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 2.845479286362262e-05, | |
| "loss": 0.9798, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 2.832879750025199e-05, | |
| "loss": 0.9704, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 2.8202802136881364e-05, | |
| "loss": 0.9718, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.807680677351074e-05, | |
| "loss": 0.9781, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 1.32, | |
| "learning_rate": 2.795081141014011e-05, | |
| "loss": 0.9745, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 2.7824816046769482e-05, | |
| "loss": 0.9777, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 2.769882068339885e-05, | |
| "loss": 0.991, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7572825320028224e-05, | |
| "loss": 0.9719, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 2.7446829956657593e-05, | |
| "loss": 0.9648, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 2.732083459328697e-05, | |
| "loss": 0.9722, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 2.7194839229916342e-05, | |
| "loss": 0.9762, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.706884386654571e-05, | |
| "loss": 0.9843, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 2.6942848503175084e-05, | |
| "loss": 0.9765, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 1.39, | |
| "learning_rate": 2.6816853139804454e-05, | |
| "loss": 0.9799, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 2.669085777643383e-05, | |
| "loss": 0.982, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6564862413063203e-05, | |
| "loss": 0.9774, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 2.6438867049692572e-05, | |
| "loss": 0.9785, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 2.6312871686321945e-05, | |
| "loss": 0.9743, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 1.43, | |
| "learning_rate": 2.6186876322951314e-05, | |
| "loss": 0.9636, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.606088095958069e-05, | |
| "loss": 0.9695, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 2.5934885596210063e-05, | |
| "loss": 0.9645, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 2.5808890232839432e-05, | |
| "loss": 0.9772, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 2.5682894869468805e-05, | |
| "loss": 0.9525, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5556899506098174e-05, | |
| "loss": 0.9523, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 1.47, | |
| "learning_rate": 2.5430904142727547e-05, | |
| "loss": 0.9544, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 2.5304908779356923e-05, | |
| "loss": 0.9656, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 2.5178913415986292e-05, | |
| "loss": 0.964, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.5052918052615665e-05, | |
| "loss": 0.9629, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 2.4926922689245035e-05, | |
| "loss": 0.9499, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 2.480092732587441e-05, | |
| "loss": 0.9429, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 2.467493196250378e-05, | |
| "loss": 0.9622, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4548936599133153e-05, | |
| "loss": 0.9621, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 2.4422941235762525e-05, | |
| "loss": 0.9588, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 2.4296945872391895e-05, | |
| "loss": 0.9616, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 2.417095050902127e-05, | |
| "loss": 0.9563, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.404495514565064e-05, | |
| "loss": 0.9578, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 2.3918959782280013e-05, | |
| "loss": 0.9511, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 2.3792964418909386e-05, | |
| "loss": 0.9685, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 2.366696905553876e-05, | |
| "loss": 0.9645, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 2.354097369216813e-05, | |
| "loss": 0.9464, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.34149783287975e-05, | |
| "loss": 0.946, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 2.3288982965426873e-05, | |
| "loss": 0.9529, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 2.3162987602056246e-05, | |
| "loss": 0.9481, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 1.62, | |
| "learning_rate": 2.303699223868562e-05, | |
| "loss": 0.9501, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.2910996875314988e-05, | |
| "loss": 0.9357, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 2.278500151194436e-05, | |
| "loss": 0.9532, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 2.2659006148573734e-05, | |
| "loss": 0.9488, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 2.2533010785203106e-05, | |
| "loss": 0.9322, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.240701542183248e-05, | |
| "loss": 0.9514, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 1.66, | |
| "learning_rate": 2.228102005846185e-05, | |
| "loss": 0.9613, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 2.215502469509122e-05, | |
| "loss": 0.9399, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 2.2029029331720594e-05, | |
| "loss": 0.9492, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.1903033968349963e-05, | |
| "loss": 0.9261, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 2.177703860497934e-05, | |
| "loss": 0.9527, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 2.165104324160871e-05, | |
| "loss": 0.9458, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 2.1525047878238085e-05, | |
| "loss": 0.931, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1399052514867454e-05, | |
| "loss": 0.9366, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 2.1273057151496824e-05, | |
| "loss": 0.9383, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 2.11470617881262e-05, | |
| "loss": 0.9284, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 2.102106642475557e-05, | |
| "loss": 0.9305, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0895071061384942e-05, | |
| "loss": 0.9157, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 2.0769075698014314e-05, | |
| "loss": 0.9431, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 2.0643080334643687e-05, | |
| "loss": 0.9255, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 2.051708497127306e-05, | |
| "loss": 0.9323, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.039108960790243e-05, | |
| "loss": 0.9396, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 2.0265094244531802e-05, | |
| "loss": 0.9209, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 2.0139098881161175e-05, | |
| "loss": 0.9257, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 2.0013103517790548e-05, | |
| "loss": 0.9306, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.9887108154419917e-05, | |
| "loss": 0.9373, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 1.81, | |
| "learning_rate": 1.976111279104929e-05, | |
| "loss": 0.9222, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 1.9635117427678662e-05, | |
| "loss": 0.9197, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 1.950912206430803e-05, | |
| "loss": 0.9244, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9383126700937408e-05, | |
| "loss": 0.943, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 1.9257131337566777e-05, | |
| "loss": 0.9351, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 1.913113597419615e-05, | |
| "loss": 0.928, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 1.9005140610825523e-05, | |
| "loss": 0.9206, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8879145247454892e-05, | |
| "loss": 0.9219, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 1.8753149884084268e-05, | |
| "loss": 0.9073, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 1.8627154520713637e-05, | |
| "loss": 0.9252, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 1.850115915734301e-05, | |
| "loss": 0.9161, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 1.8375163793972383e-05, | |
| "loss": 0.9181, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.8249168430601756e-05, | |
| "loss": 0.9225, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 1.812317306723113e-05, | |
| "loss": 0.9108, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 1.7997177703860498e-05, | |
| "loss": 0.9115, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 1.787118234048987e-05, | |
| "loss": 0.9239, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7745186977119243e-05, | |
| "loss": 0.9213, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 1.7619191613748616e-05, | |
| "loss": 0.9211, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 1.7493196250377985e-05, | |
| "loss": 0.9252, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 1.7367200887007358e-05, | |
| "loss": 0.9165, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.724120552363673e-05, | |
| "loss": 0.928, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 1.97, | |
| "learning_rate": 1.7115210160266104e-05, | |
| "loss": 0.9208, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 1.6989214796895476e-05, | |
| "loss": 0.9182, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 1.6863219433524846e-05, | |
| "loss": 0.9167, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.673722407015422e-05, | |
| "loss": 0.9354, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 1.661122870678359e-05, | |
| "loss": 0.8513, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 1.648523334341296e-05, | |
| "loss": 0.8098, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 1.6359237980042337e-05, | |
| "loss": 0.7988, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6233242616671706e-05, | |
| "loss": 0.8004, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 1.6107247253301082e-05, | |
| "loss": 0.8006, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 1.598125188993045e-05, | |
| "loss": 0.7905, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 1.5855256526559824e-05, | |
| "loss": 0.7937, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5729261163189197e-05, | |
| "loss": 0.7973, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 1.5603265799818566e-05, | |
| "loss": 0.8132, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 1.547727043644794e-05, | |
| "loss": 0.8012, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 1.535127507307731e-05, | |
| "loss": 0.8014, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5225279709706683e-05, | |
| "loss": 0.8053, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 1.5099284346336057e-05, | |
| "loss": 0.7884, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 1.4973288982965428e-05, | |
| "loss": 0.8122, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 1.48472936195948e-05, | |
| "loss": 0.8029, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4721298256224172e-05, | |
| "loss": 0.8, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 2.12, | |
| "learning_rate": 1.4595302892853543e-05, | |
| "loss": 0.7929, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 1.4469307529482914e-05, | |
| "loss": 0.8066, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 1.4343312166112288e-05, | |
| "loss": 0.7947, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.421731680274166e-05, | |
| "loss": 0.7977, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 1.4091321439371032e-05, | |
| "loss": 0.7844, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 2.16, | |
| "learning_rate": 1.3965326076000403e-05, | |
| "loss": 0.7857, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 1.3839330712629774e-05, | |
| "loss": 0.7947, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.3713335349259149e-05, | |
| "loss": 0.7909, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 1.358733998588852e-05, | |
| "loss": 0.7802, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 1.346134462251789e-05, | |
| "loss": 0.79, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 2.2, | |
| "learning_rate": 1.3335349259147265e-05, | |
| "loss": 0.7871, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3209353895776636e-05, | |
| "loss": 0.7805, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 1.3083358532406009e-05, | |
| "loss": 0.7923, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 1.295736316903538e-05, | |
| "loss": 0.7878, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 1.2831367805664751e-05, | |
| "loss": 0.7967, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 1.2705372442294126e-05, | |
| "loss": 0.7885, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2579377078923497e-05, | |
| "loss": 0.7872, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 1.2453381715552868e-05, | |
| "loss": 0.8009, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 1.232738635218224e-05, | |
| "loss": 0.7915, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 2.27, | |
| "learning_rate": 1.2201390988811611e-05, | |
| "loss": 0.7895, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.2075395625440984e-05, | |
| "loss": 0.7752, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 1.1949400262070357e-05, | |
| "loss": 0.7927, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 1.182340489869973e-05, | |
| "loss": 0.7833, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 1.16974095353291e-05, | |
| "loss": 0.8001, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1571414171958472e-05, | |
| "loss": 0.7981, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 2.31, | |
| "learning_rate": 1.1445418808587844e-05, | |
| "loss": 0.7775, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 1.1319423445217217e-05, | |
| "loss": 0.7852, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 1.1193428081846588e-05, | |
| "loss": 0.7913, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.1067432718475961e-05, | |
| "loss": 0.7735, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 1.0941437355105334e-05, | |
| "loss": 0.7837, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 2.35, | |
| "learning_rate": 1.0815441991734705e-05, | |
| "loss": 0.7872, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 1.0689446628364076e-05, | |
| "loss": 0.786, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0563451264993449e-05, | |
| "loss": 0.7824, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 1.0437455901622821e-05, | |
| "loss": 0.7877, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 1.0311460538252194e-05, | |
| "loss": 0.7916, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 2.39, | |
| "learning_rate": 1.0185465174881565e-05, | |
| "loss": 0.7886, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 1.0059469811510936e-05, | |
| "loss": 0.7913, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 9.933474448140309e-06, | |
| "loss": 0.7895, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 9.807479084769682e-06, | |
| "loss": 0.7897, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 9.681483721399053e-06, | |
| "loss": 0.7791, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.555488358028425e-06, | |
| "loss": 0.7843, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 9.429492994657798e-06, | |
| "loss": 0.7766, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 9.303497631287169e-06, | |
| "loss": 0.7818, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 9.17750226791654e-06, | |
| "loss": 0.7693, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 9.051506904545913e-06, | |
| "loss": 0.7673, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 2.46, | |
| "learning_rate": 8.925511541175286e-06, | |
| "loss": 0.7785, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 8.799516177804658e-06, | |
| "loss": 0.7903, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 8.67352081443403e-06, | |
| "loss": 0.7746, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.5475254510634e-06, | |
| "loss": 0.789, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 8.421530087692773e-06, | |
| "loss": 0.785, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 2.5, | |
| "learning_rate": 8.295534724322146e-06, | |
| "loss": 0.7753, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 8.169539360951517e-06, | |
| "loss": 0.7868, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 8.04354399758089e-06, | |
| "loss": 0.7737, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 7.917548634210262e-06, | |
| "loss": 0.7822, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 7.791553270839633e-06, | |
| "loss": 0.7627, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 2.54, | |
| "learning_rate": 7.665557907469004e-06, | |
| "loss": 0.7791, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 7.539562544098377e-06, | |
| "loss": 0.7829, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.41356718072775e-06, | |
| "loss": 0.7713, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 7.287571817357122e-06, | |
| "loss": 0.7729, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 7.161576453986493e-06, | |
| "loss": 0.7776, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 7.035581090615866e-06, | |
| "loss": 0.7708, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.909585727245238e-06, | |
| "loss": 0.7757, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 6.78359036387461e-06, | |
| "loss": 0.7628, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 6.657595000503981e-06, | |
| "loss": 0.7799, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 6.531599637133354e-06, | |
| "loss": 0.7735, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.405604273762726e-06, | |
| "loss": 0.7719, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 6.279608910392099e-06, | |
| "loss": 0.775, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 6.1536135470214705e-06, | |
| "loss": 0.7621, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 6.027618183650842e-06, | |
| "loss": 0.7791, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.901622820280214e-06, | |
| "loss": 0.771, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 2.65, | |
| "learning_rate": 5.775627456909586e-06, | |
| "loss": 0.7671, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 5.649632093538958e-06, | |
| "loss": 0.776, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 5.52363673016833e-06, | |
| "loss": 0.7749, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.397641366797703e-06, | |
| "loss": 0.7824, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 5.271646003427074e-06, | |
| "loss": 0.7729, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 2.69, | |
| "learning_rate": 5.1456506400564465e-06, | |
| "loss": 0.7823, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 5.019655276685818e-06, | |
| "loss": 0.7834, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.89365991331519e-06, | |
| "loss": 0.7813, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 4.767664549944562e-06, | |
| "loss": 0.7683, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 4.641669186573934e-06, | |
| "loss": 0.761, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 4.515673823203306e-06, | |
| "loss": 0.7549, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.389678459832678e-06, | |
| "loss": 0.7734, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 4.2636830964620506e-06, | |
| "loss": 0.7648, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 4.1376877330914225e-06, | |
| "loss": 0.7518, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 4.011692369720794e-06, | |
| "loss": 0.7578, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.885697006350166e-06, | |
| "loss": 0.7673, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 3.7597016429795386e-06, | |
| "loss": 0.7848, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 3.6337062796089104e-06, | |
| "loss": 0.7678, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 3.5077109162382828e-06, | |
| "loss": 0.7702, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.3817155528676542e-06, | |
| "loss": 0.7643, | |
| "step": 185000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 3.2557201894970265e-06, | |
| "loss": 0.7612, | |
| "step": 185500 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 3.1297248261263984e-06, | |
| "loss": 0.7685, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 3.0037294627557707e-06, | |
| "loss": 0.7564, | |
| "step": 186500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.8777340993851426e-06, | |
| "loss": 0.7607, | |
| "step": 187000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.751738736014515e-06, | |
| "loss": 0.7596, | |
| "step": 187500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.625743372643887e-06, | |
| "loss": 0.7622, | |
| "step": 188000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 2.4997480092732587e-06, | |
| "loss": 0.7537, | |
| "step": 188500 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.373752645902631e-06, | |
| "loss": 0.7591, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.247757282532003e-06, | |
| "loss": 0.7586, | |
| "step": 189500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1217619191613752e-06, | |
| "loss": 0.7731, | |
| "step": 190000 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 1.995766555790747e-06, | |
| "loss": 0.7641, | |
| "step": 190500 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 1.869771192420119e-06, | |
| "loss": 0.7527, | |
| "step": 191000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.743775829049491e-06, | |
| "loss": 0.7585, | |
| "step": 191500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 1.6177804656788632e-06, | |
| "loss": 0.7585, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 1.4917851023082353e-06, | |
| "loss": 0.7581, | |
| "step": 192500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 1.3657897389376072e-06, | |
| "loss": 0.7578, | |
| "step": 193000 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.2397943755669793e-06, | |
| "loss": 0.7728, | |
| "step": 193500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 1.1137990121963514e-06, | |
| "loss": 0.7493, | |
| "step": 194000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 9.87803648825723e-07, | |
| "loss": 0.7584, | |
| "step": 194500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 8.618082854550952e-07, | |
| "loss": 0.7619, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 7.358129220844674e-07, | |
| "loss": 0.7611, | |
| "step": 195500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 6.098175587138394e-07, | |
| "loss": 0.7565, | |
| "step": 196000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 4.838221953432114e-07, | |
| "loss": 0.7565, | |
| "step": 196500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 3.5782683197258347e-07, | |
| "loss": 0.7528, | |
| "step": 197000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.318314686019555e-07, | |
| "loss": 0.7524, | |
| "step": 197500 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 1.058361052313275e-07, | |
| "loss": 0.7653, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 198420, | |
| "total_flos": 2.7061438809725338e+17, | |
| "train_loss": 1.0211255452190962, | |
| "train_runtime": 50903.0981, | |
| "train_samples_per_second": 38.98, | |
| "train_steps_per_second": 3.898 | |
| } | |
| ], | |
| "max_steps": 198420, | |
| "num_train_epochs": 3, | |
| "total_flos": 2.7061438809725338e+17, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |