|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 3.0, |
|
"global_step": 105003, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.01, |
|
"learning_rate": 4.9761911564431496e-05, |
|
"loss": 4.0373, |
|
"step": 500 |
|
}, |
|
{ |
|
"epoch": 0.03, |
|
"learning_rate": 4.952382312886299e-05, |
|
"loss": 3.5361, |
|
"step": 1000 |
|
}, |
|
{ |
|
"epoch": 0.04, |
|
"learning_rate": 4.9285734693294476e-05, |
|
"loss": 3.2302, |
|
"step": 1500 |
|
}, |
|
{ |
|
"epoch": 0.06, |
|
"learning_rate": 4.904764625772597e-05, |
|
"loss": 3.0339, |
|
"step": 2000 |
|
}, |
|
{ |
|
"epoch": 0.07, |
|
"learning_rate": 4.880955782215746e-05, |
|
"loss": 2.9055, |
|
"step": 2500 |
|
}, |
|
{ |
|
"epoch": 0.09, |
|
"learning_rate": 4.857146938658896e-05, |
|
"loss": 2.8156, |
|
"step": 3000 |
|
}, |
|
{ |
|
"epoch": 0.1, |
|
"learning_rate": 4.833338095102045e-05, |
|
"loss": 2.6683, |
|
"step": 3500 |
|
}, |
|
{ |
|
"epoch": 0.11, |
|
"learning_rate": 4.809529251545194e-05, |
|
"loss": 2.5815, |
|
"step": 4000 |
|
}, |
|
{ |
|
"epoch": 0.13, |
|
"learning_rate": 4.785720407988343e-05, |
|
"loss": 2.5335, |
|
"step": 4500 |
|
}, |
|
{ |
|
"epoch": 0.14, |
|
"learning_rate": 4.7619115644314924e-05, |
|
"loss": 2.4421, |
|
"step": 5000 |
|
}, |
|
{ |
|
"epoch": 0.16, |
|
"learning_rate": 4.738102720874642e-05, |
|
"loss": 2.4007, |
|
"step": 5500 |
|
}, |
|
{ |
|
"epoch": 0.17, |
|
"learning_rate": 4.714293877317791e-05, |
|
"loss": 2.3885, |
|
"step": 6000 |
|
}, |
|
{ |
|
"epoch": 0.19, |
|
"learning_rate": 4.6904850337609405e-05, |
|
"loss": 2.2688, |
|
"step": 6500 |
|
}, |
|
{ |
|
"epoch": 0.2, |
|
"learning_rate": 4.66667619020409e-05, |
|
"loss": 2.2576, |
|
"step": 7000 |
|
}, |
|
{ |
|
"epoch": 0.21, |
|
"learning_rate": 4.642867346647239e-05, |
|
"loss": 2.194, |
|
"step": 7500 |
|
}, |
|
{ |
|
"epoch": 0.23, |
|
"learning_rate": 4.6190585030903885e-05, |
|
"loss": 2.1845, |
|
"step": 8000 |
|
}, |
|
{ |
|
"epoch": 0.24, |
|
"learning_rate": 4.595249659533538e-05, |
|
"loss": 2.1494, |
|
"step": 8500 |
|
}, |
|
{ |
|
"epoch": 0.26, |
|
"learning_rate": 4.5714408159766866e-05, |
|
"loss": 2.0918, |
|
"step": 9000 |
|
}, |
|
{ |
|
"epoch": 0.27, |
|
"learning_rate": 4.547631972419836e-05, |
|
"loss": 2.0708, |
|
"step": 9500 |
|
}, |
|
{ |
|
"epoch": 0.29, |
|
"learning_rate": 4.523823128862985e-05, |
|
"loss": 2.0594, |
|
"step": 10000 |
|
}, |
|
{ |
|
"epoch": 0.3, |
|
"learning_rate": 4.5000142853061346e-05, |
|
"loss": 2.0434, |
|
"step": 10500 |
|
}, |
|
{ |
|
"epoch": 0.31, |
|
"learning_rate": 4.476205441749284e-05, |
|
"loss": 2.0143, |
|
"step": 11000 |
|
}, |
|
{ |
|
"epoch": 0.33, |
|
"learning_rate": 4.452396598192433e-05, |
|
"loss": 1.9919, |
|
"step": 11500 |
|
}, |
|
{ |
|
"epoch": 0.34, |
|
"learning_rate": 4.428587754635582e-05, |
|
"loss": 1.9623, |
|
"step": 12000 |
|
}, |
|
{ |
|
"epoch": 0.36, |
|
"learning_rate": 4.4047789110787314e-05, |
|
"loss": 1.9654, |
|
"step": 12500 |
|
}, |
|
{ |
|
"epoch": 0.37, |
|
"learning_rate": 4.380970067521881e-05, |
|
"loss": 1.9521, |
|
"step": 13000 |
|
}, |
|
{ |
|
"epoch": 0.39, |
|
"learning_rate": 4.35716122396503e-05, |
|
"loss": 1.9266, |
|
"step": 13500 |
|
}, |
|
{ |
|
"epoch": 0.4, |
|
"learning_rate": 4.333352380408179e-05, |
|
"loss": 1.8828, |
|
"step": 14000 |
|
}, |
|
{ |
|
"epoch": 0.41, |
|
"learning_rate": 4.309543536851328e-05, |
|
"loss": 1.8585, |
|
"step": 14500 |
|
}, |
|
{ |
|
"epoch": 0.43, |
|
"learning_rate": 4.2857346932944775e-05, |
|
"loss": 1.871, |
|
"step": 15000 |
|
}, |
|
{ |
|
"epoch": 0.44, |
|
"learning_rate": 4.261925849737627e-05, |
|
"loss": 1.8358, |
|
"step": 15500 |
|
}, |
|
{ |
|
"epoch": 0.46, |
|
"learning_rate": 4.238117006180776e-05, |
|
"loss": 1.8018, |
|
"step": 16000 |
|
}, |
|
{ |
|
"epoch": 0.47, |
|
"learning_rate": 4.214308162623925e-05, |
|
"loss": 1.8175, |
|
"step": 16500 |
|
}, |
|
{ |
|
"epoch": 0.49, |
|
"learning_rate": 4.190499319067074e-05, |
|
"loss": 1.7991, |
|
"step": 17000 |
|
}, |
|
{ |
|
"epoch": 0.5, |
|
"learning_rate": 4.1666904755102236e-05, |
|
"loss": 1.7898, |
|
"step": 17500 |
|
}, |
|
{ |
|
"epoch": 0.51, |
|
"learning_rate": 4.142881631953373e-05, |
|
"loss": 1.7767, |
|
"step": 18000 |
|
}, |
|
{ |
|
"epoch": 0.53, |
|
"learning_rate": 4.119072788396522e-05, |
|
"loss": 1.7547, |
|
"step": 18500 |
|
}, |
|
{ |
|
"epoch": 0.54, |
|
"learning_rate": 4.095263944839671e-05, |
|
"loss": 1.7509, |
|
"step": 19000 |
|
}, |
|
{ |
|
"epoch": 0.56, |
|
"learning_rate": 4.07145510128282e-05, |
|
"loss": 1.7453, |
|
"step": 19500 |
|
}, |
|
{ |
|
"epoch": 0.57, |
|
"learning_rate": 4.0476462577259696e-05, |
|
"loss": 1.7294, |
|
"step": 20000 |
|
}, |
|
{ |
|
"epoch": 0.59, |
|
"learning_rate": 4.023837414169119e-05, |
|
"loss": 1.7356, |
|
"step": 20500 |
|
}, |
|
{ |
|
"epoch": 0.6, |
|
"learning_rate": 4.0000285706122684e-05, |
|
"loss": 1.7289, |
|
"step": 21000 |
|
}, |
|
{ |
|
"epoch": 0.61, |
|
"learning_rate": 3.976219727055418e-05, |
|
"loss": 1.7115, |
|
"step": 21500 |
|
}, |
|
{ |
|
"epoch": 0.63, |
|
"learning_rate": 3.952410883498567e-05, |
|
"loss": 1.6996, |
|
"step": 22000 |
|
}, |
|
{ |
|
"epoch": 0.64, |
|
"learning_rate": 3.9286020399417164e-05, |
|
"loss": 1.6723, |
|
"step": 22500 |
|
}, |
|
{ |
|
"epoch": 0.66, |
|
"learning_rate": 3.904793196384866e-05, |
|
"loss": 1.7014, |
|
"step": 23000 |
|
}, |
|
{ |
|
"epoch": 0.67, |
|
"learning_rate": 3.880984352828015e-05, |
|
"loss": 1.6455, |
|
"step": 23500 |
|
}, |
|
{ |
|
"epoch": 0.69, |
|
"learning_rate": 3.857175509271164e-05, |
|
"loss": 1.6545, |
|
"step": 24000 |
|
}, |
|
{ |
|
"epoch": 0.7, |
|
"learning_rate": 3.833366665714313e-05, |
|
"loss": 1.6246, |
|
"step": 24500 |
|
}, |
|
{ |
|
"epoch": 0.71, |
|
"learning_rate": 3.8095578221574625e-05, |
|
"loss": 1.6465, |
|
"step": 25000 |
|
}, |
|
{ |
|
"epoch": 0.73, |
|
"learning_rate": 3.785748978600612e-05, |
|
"loss": 1.6781, |
|
"step": 25500 |
|
}, |
|
{ |
|
"epoch": 0.74, |
|
"learning_rate": 3.761940135043761e-05, |
|
"loss": 1.6267, |
|
"step": 26000 |
|
}, |
|
{ |
|
"epoch": 0.76, |
|
"learning_rate": 3.73813129148691e-05, |
|
"loss": 1.5977, |
|
"step": 26500 |
|
}, |
|
{ |
|
"epoch": 0.77, |
|
"learning_rate": 3.714322447930059e-05, |
|
"loss": 1.6134, |
|
"step": 27000 |
|
}, |
|
{ |
|
"epoch": 0.79, |
|
"learning_rate": 3.6905136043732086e-05, |
|
"loss": 1.5762, |
|
"step": 27500 |
|
}, |
|
{ |
|
"epoch": 0.8, |
|
"learning_rate": 3.666704760816358e-05, |
|
"loss": 1.5836, |
|
"step": 28000 |
|
}, |
|
{ |
|
"epoch": 0.81, |
|
"learning_rate": 3.642895917259507e-05, |
|
"loss": 1.5743, |
|
"step": 28500 |
|
}, |
|
{ |
|
"epoch": 0.83, |
|
"learning_rate": 3.619087073702656e-05, |
|
"loss": 1.5743, |
|
"step": 29000 |
|
}, |
|
{ |
|
"epoch": 0.84, |
|
"learning_rate": 3.595278230145805e-05, |
|
"loss": 1.5895, |
|
"step": 29500 |
|
}, |
|
{ |
|
"epoch": 0.86, |
|
"learning_rate": 3.571469386588955e-05, |
|
"loss": 1.5724, |
|
"step": 30000 |
|
}, |
|
{ |
|
"epoch": 0.87, |
|
"learning_rate": 3.547660543032104e-05, |
|
"loss": 1.5348, |
|
"step": 30500 |
|
}, |
|
{ |
|
"epoch": 0.89, |
|
"learning_rate": 3.5238516994752534e-05, |
|
"loss": 1.5205, |
|
"step": 31000 |
|
}, |
|
{ |
|
"epoch": 0.9, |
|
"learning_rate": 3.500042855918402e-05, |
|
"loss": 1.5438, |
|
"step": 31500 |
|
}, |
|
{ |
|
"epoch": 0.91, |
|
"learning_rate": 3.4762340123615514e-05, |
|
"loss": 1.5228, |
|
"step": 32000 |
|
}, |
|
{ |
|
"epoch": 0.93, |
|
"learning_rate": 3.452425168804701e-05, |
|
"loss": 1.546, |
|
"step": 32500 |
|
}, |
|
{ |
|
"epoch": 0.94, |
|
"learning_rate": 3.42861632524785e-05, |
|
"loss": 1.5256, |
|
"step": 33000 |
|
}, |
|
{ |
|
"epoch": 0.96, |
|
"learning_rate": 3.4048074816909995e-05, |
|
"loss": 1.5089, |
|
"step": 33500 |
|
}, |
|
{ |
|
"epoch": 0.97, |
|
"learning_rate": 3.380998638134149e-05, |
|
"loss": 1.5449, |
|
"step": 34000 |
|
}, |
|
{ |
|
"epoch": 0.99, |
|
"learning_rate": 3.357189794577298e-05, |
|
"loss": 1.4929, |
|
"step": 34500 |
|
}, |
|
{ |
|
"epoch": 1.0, |
|
"learning_rate": 3.3333809510204476e-05, |
|
"loss": 1.4842, |
|
"step": 35000 |
|
}, |
|
{ |
|
"epoch": 1.01, |
|
"learning_rate": 3.309572107463597e-05, |
|
"loss": 1.3251, |
|
"step": 35500 |
|
}, |
|
{ |
|
"epoch": 1.03, |
|
"learning_rate": 3.285763263906746e-05, |
|
"loss": 1.3659, |
|
"step": 36000 |
|
}, |
|
{ |
|
"epoch": 1.04, |
|
"learning_rate": 3.261954420349895e-05, |
|
"loss": 1.3282, |
|
"step": 36500 |
|
}, |
|
{ |
|
"epoch": 1.06, |
|
"learning_rate": 3.238145576793044e-05, |
|
"loss": 1.3274, |
|
"step": 37000 |
|
}, |
|
{ |
|
"epoch": 1.07, |
|
"learning_rate": 3.2143367332361936e-05, |
|
"loss": 1.3324, |
|
"step": 37500 |
|
}, |
|
{ |
|
"epoch": 1.09, |
|
"learning_rate": 3.190527889679343e-05, |
|
"loss": 1.3201, |
|
"step": 38000 |
|
}, |
|
{ |
|
"epoch": 1.1, |
|
"learning_rate": 3.1667190461224923e-05, |
|
"loss": 1.3183, |
|
"step": 38500 |
|
}, |
|
{ |
|
"epoch": 1.11, |
|
"learning_rate": 3.142910202565641e-05, |
|
"loss": 1.2984, |
|
"step": 39000 |
|
}, |
|
{ |
|
"epoch": 1.13, |
|
"learning_rate": 3.1191013590087904e-05, |
|
"loss": 1.3019, |
|
"step": 39500 |
|
}, |
|
{ |
|
"epoch": 1.14, |
|
"learning_rate": 3.09529251545194e-05, |
|
"loss": 1.31, |
|
"step": 40000 |
|
}, |
|
{ |
|
"epoch": 1.16, |
|
"learning_rate": 3.071483671895089e-05, |
|
"loss": 1.3066, |
|
"step": 40500 |
|
}, |
|
{ |
|
"epoch": 1.17, |
|
"learning_rate": 3.0476748283382384e-05, |
|
"loss": 1.3006, |
|
"step": 41000 |
|
}, |
|
{ |
|
"epoch": 1.19, |
|
"learning_rate": 3.023865984781387e-05, |
|
"loss": 1.2894, |
|
"step": 41500 |
|
}, |
|
{ |
|
"epoch": 1.2, |
|
"learning_rate": 3.0000571412245365e-05, |
|
"loss": 1.3018, |
|
"step": 42000 |
|
}, |
|
{ |
|
"epoch": 1.21, |
|
"learning_rate": 2.9762482976676858e-05, |
|
"loss": 1.3055, |
|
"step": 42500 |
|
}, |
|
{ |
|
"epoch": 1.23, |
|
"learning_rate": 2.9524394541108352e-05, |
|
"loss": 1.306, |
|
"step": 43000 |
|
}, |
|
{ |
|
"epoch": 1.24, |
|
"learning_rate": 2.9286306105539845e-05, |
|
"loss": 1.285, |
|
"step": 43500 |
|
}, |
|
{ |
|
"epoch": 1.26, |
|
"learning_rate": 2.9048217669971332e-05, |
|
"loss": 1.2823, |
|
"step": 44000 |
|
}, |
|
{ |
|
"epoch": 1.27, |
|
"learning_rate": 2.8810129234402826e-05, |
|
"loss": 1.2871, |
|
"step": 44500 |
|
}, |
|
{ |
|
"epoch": 1.29, |
|
"learning_rate": 2.857204079883432e-05, |
|
"loss": 1.2921, |
|
"step": 45000 |
|
}, |
|
{ |
|
"epoch": 1.3, |
|
"learning_rate": 2.8333952363265813e-05, |
|
"loss": 1.2532, |
|
"step": 45500 |
|
}, |
|
{ |
|
"epoch": 1.31, |
|
"learning_rate": 2.8095863927697306e-05, |
|
"loss": 1.2852, |
|
"step": 46000 |
|
}, |
|
{ |
|
"epoch": 1.33, |
|
"learning_rate": 2.7857775492128796e-05, |
|
"loss": 1.2802, |
|
"step": 46500 |
|
}, |
|
{ |
|
"epoch": 1.34, |
|
"learning_rate": 2.761968705656029e-05, |
|
"loss": 1.2753, |
|
"step": 47000 |
|
}, |
|
{ |
|
"epoch": 1.36, |
|
"learning_rate": 2.7381598620991783e-05, |
|
"loss": 1.2729, |
|
"step": 47500 |
|
}, |
|
{ |
|
"epoch": 1.37, |
|
"learning_rate": 2.7143510185423277e-05, |
|
"loss": 1.2462, |
|
"step": 48000 |
|
}, |
|
{ |
|
"epoch": 1.39, |
|
"learning_rate": 2.690542174985477e-05, |
|
"loss": 1.2493, |
|
"step": 48500 |
|
}, |
|
{ |
|
"epoch": 1.4, |
|
"learning_rate": 2.6667333314286257e-05, |
|
"loss": 1.2481, |
|
"step": 49000 |
|
}, |
|
{ |
|
"epoch": 1.41, |
|
"learning_rate": 2.642924487871775e-05, |
|
"loss": 1.2634, |
|
"step": 49500 |
|
}, |
|
{ |
|
"epoch": 1.43, |
|
"learning_rate": 2.6191156443149244e-05, |
|
"loss": 1.2348, |
|
"step": 50000 |
|
}, |
|
{ |
|
"epoch": 1.44, |
|
"learning_rate": 2.5953068007580738e-05, |
|
"loss": 1.243, |
|
"step": 50500 |
|
}, |
|
{ |
|
"epoch": 1.46, |
|
"learning_rate": 2.571497957201223e-05, |
|
"loss": 1.2604, |
|
"step": 51000 |
|
}, |
|
{ |
|
"epoch": 1.47, |
|
"learning_rate": 2.547689113644372e-05, |
|
"loss": 1.2488, |
|
"step": 51500 |
|
}, |
|
{ |
|
"epoch": 1.49, |
|
"learning_rate": 2.5238802700875215e-05, |
|
"loss": 1.189, |
|
"step": 52000 |
|
}, |
|
{ |
|
"epoch": 1.5, |
|
"learning_rate": 2.500071426530671e-05, |
|
"loss": 1.2214, |
|
"step": 52500 |
|
}, |
|
{ |
|
"epoch": 1.51, |
|
"learning_rate": 2.47626258297382e-05, |
|
"loss": 1.2454, |
|
"step": 53000 |
|
}, |
|
{ |
|
"epoch": 1.53, |
|
"learning_rate": 2.4524537394169692e-05, |
|
"loss": 1.217, |
|
"step": 53500 |
|
}, |
|
{ |
|
"epoch": 1.54, |
|
"learning_rate": 2.4286448958601186e-05, |
|
"loss": 1.2253, |
|
"step": 54000 |
|
}, |
|
{ |
|
"epoch": 1.56, |
|
"learning_rate": 2.4048360523032676e-05, |
|
"loss": 1.2075, |
|
"step": 54500 |
|
}, |
|
{ |
|
"epoch": 1.57, |
|
"learning_rate": 2.381027208746417e-05, |
|
"loss": 1.2443, |
|
"step": 55000 |
|
}, |
|
{ |
|
"epoch": 1.59, |
|
"learning_rate": 2.357218365189566e-05, |
|
"loss": 1.2235, |
|
"step": 55500 |
|
}, |
|
{ |
|
"epoch": 1.6, |
|
"learning_rate": 2.3334095216327153e-05, |
|
"loss": 1.2038, |
|
"step": 56000 |
|
}, |
|
{ |
|
"epoch": 1.61, |
|
"learning_rate": 2.3096006780758647e-05, |
|
"loss": 1.2256, |
|
"step": 56500 |
|
}, |
|
{ |
|
"epoch": 1.63, |
|
"learning_rate": 2.2857918345190137e-05, |
|
"loss": 1.1911, |
|
"step": 57000 |
|
}, |
|
{ |
|
"epoch": 1.64, |
|
"learning_rate": 2.261982990962163e-05, |
|
"loss": 1.2346, |
|
"step": 57500 |
|
}, |
|
{ |
|
"epoch": 1.66, |
|
"learning_rate": 2.238174147405312e-05, |
|
"loss": 1.1922, |
|
"step": 58000 |
|
}, |
|
{ |
|
"epoch": 1.67, |
|
"learning_rate": 2.2143653038484614e-05, |
|
"loss": 1.2169, |
|
"step": 58500 |
|
}, |
|
{ |
|
"epoch": 1.69, |
|
"learning_rate": 2.1905564602916108e-05, |
|
"loss": 1.2035, |
|
"step": 59000 |
|
}, |
|
{ |
|
"epoch": 1.7, |
|
"learning_rate": 2.16674761673476e-05, |
|
"loss": 1.2302, |
|
"step": 59500 |
|
}, |
|
{ |
|
"epoch": 1.71, |
|
"learning_rate": 2.1429387731779095e-05, |
|
"loss": 1.1648, |
|
"step": 60000 |
|
}, |
|
{ |
|
"epoch": 1.73, |
|
"learning_rate": 2.1191299296210585e-05, |
|
"loss": 1.2085, |
|
"step": 60500 |
|
}, |
|
{ |
|
"epoch": 1.74, |
|
"learning_rate": 2.095321086064208e-05, |
|
"loss": 1.1842, |
|
"step": 61000 |
|
}, |
|
{ |
|
"epoch": 1.76, |
|
"learning_rate": 2.0715122425073572e-05, |
|
"loss": 1.1812, |
|
"step": 61500 |
|
}, |
|
{ |
|
"epoch": 1.77, |
|
"learning_rate": 2.0477033989505062e-05, |
|
"loss": 1.1811, |
|
"step": 62000 |
|
}, |
|
{ |
|
"epoch": 1.79, |
|
"learning_rate": 2.0238945553936556e-05, |
|
"loss": 1.1892, |
|
"step": 62500 |
|
}, |
|
{ |
|
"epoch": 1.8, |
|
"learning_rate": 2.0000857118368046e-05, |
|
"loss": 1.171, |
|
"step": 63000 |
|
}, |
|
{ |
|
"epoch": 1.81, |
|
"learning_rate": 1.976276868279954e-05, |
|
"loss": 1.1611, |
|
"step": 63500 |
|
}, |
|
{ |
|
"epoch": 1.83, |
|
"learning_rate": 1.9524680247231033e-05, |
|
"loss": 1.1365, |
|
"step": 64000 |
|
}, |
|
{ |
|
"epoch": 1.84, |
|
"learning_rate": 1.9286591811662523e-05, |
|
"loss": 1.1739, |
|
"step": 64500 |
|
}, |
|
{ |
|
"epoch": 1.86, |
|
"learning_rate": 1.9048503376094017e-05, |
|
"loss": 1.1728, |
|
"step": 65000 |
|
}, |
|
{ |
|
"epoch": 1.87, |
|
"learning_rate": 1.881041494052551e-05, |
|
"loss": 1.1646, |
|
"step": 65500 |
|
}, |
|
{ |
|
"epoch": 1.89, |
|
"learning_rate": 1.8572326504957004e-05, |
|
"loss": 1.1384, |
|
"step": 66000 |
|
}, |
|
{ |
|
"epoch": 1.9, |
|
"learning_rate": 1.8334238069388497e-05, |
|
"loss": 1.1416, |
|
"step": 66500 |
|
}, |
|
{ |
|
"epoch": 1.91, |
|
"learning_rate": 1.8096149633819987e-05, |
|
"loss": 1.1474, |
|
"step": 67000 |
|
}, |
|
{ |
|
"epoch": 1.93, |
|
"learning_rate": 1.785806119825148e-05, |
|
"loss": 1.1641, |
|
"step": 67500 |
|
}, |
|
{ |
|
"epoch": 1.94, |
|
"learning_rate": 1.761997276268297e-05, |
|
"loss": 1.165, |
|
"step": 68000 |
|
}, |
|
{ |
|
"epoch": 1.96, |
|
"learning_rate": 1.7381884327114465e-05, |
|
"loss": 1.1195, |
|
"step": 68500 |
|
}, |
|
{ |
|
"epoch": 1.97, |
|
"learning_rate": 1.7143795891545958e-05, |
|
"loss": 1.1765, |
|
"step": 69000 |
|
}, |
|
{ |
|
"epoch": 1.99, |
|
"learning_rate": 1.6905707455977448e-05, |
|
"loss": 1.1355, |
|
"step": 69500 |
|
}, |
|
{ |
|
"epoch": 2.0, |
|
"learning_rate": 1.6667619020408942e-05, |
|
"loss": 1.1765, |
|
"step": 70000 |
|
}, |
|
{ |
|
"epoch": 2.01, |
|
"learning_rate": 1.6429530584840432e-05, |
|
"loss": 0.9527, |
|
"step": 70500 |
|
}, |
|
{ |
|
"epoch": 2.03, |
|
"learning_rate": 1.6191442149271926e-05, |
|
"loss": 0.9703, |
|
"step": 71000 |
|
}, |
|
{ |
|
"epoch": 2.04, |
|
"learning_rate": 1.595335371370342e-05, |
|
"loss": 0.9582, |
|
"step": 71500 |
|
}, |
|
{ |
|
"epoch": 2.06, |
|
"learning_rate": 1.571526527813491e-05, |
|
"loss": 0.9646, |
|
"step": 72000 |
|
}, |
|
{ |
|
"epoch": 2.07, |
|
"learning_rate": 1.5477176842566403e-05, |
|
"loss": 0.9317, |
|
"step": 72500 |
|
}, |
|
{ |
|
"epoch": 2.09, |
|
"learning_rate": 1.5239088406997895e-05, |
|
"loss": 0.938, |
|
"step": 73000 |
|
}, |
|
{ |
|
"epoch": 2.1, |
|
"learning_rate": 1.5000999971429388e-05, |
|
"loss": 0.9991, |
|
"step": 73500 |
|
}, |
|
{ |
|
"epoch": 2.11, |
|
"learning_rate": 1.4762911535860882e-05, |
|
"loss": 0.9531, |
|
"step": 74000 |
|
}, |
|
{ |
|
"epoch": 2.13, |
|
"learning_rate": 1.4524823100292374e-05, |
|
"loss": 0.9612, |
|
"step": 74500 |
|
}, |
|
{ |
|
"epoch": 2.14, |
|
"learning_rate": 1.4286734664723867e-05, |
|
"loss": 0.958, |
|
"step": 75000 |
|
}, |
|
{ |
|
"epoch": 2.16, |
|
"learning_rate": 1.4048646229155357e-05, |
|
"loss": 0.9394, |
|
"step": 75500 |
|
}, |
|
{ |
|
"epoch": 2.17, |
|
"learning_rate": 1.381055779358685e-05, |
|
"loss": 0.9433, |
|
"step": 76000 |
|
}, |
|
{ |
|
"epoch": 2.19, |
|
"learning_rate": 1.3572469358018344e-05, |
|
"loss": 0.9431, |
|
"step": 76500 |
|
}, |
|
{ |
|
"epoch": 2.2, |
|
"learning_rate": 1.3334380922449834e-05, |
|
"loss": 0.9707, |
|
"step": 77000 |
|
}, |
|
{ |
|
"epoch": 2.21, |
|
"learning_rate": 1.3096292486881328e-05, |
|
"loss": 0.9453, |
|
"step": 77500 |
|
}, |
|
{ |
|
"epoch": 2.23, |
|
"learning_rate": 1.285820405131282e-05, |
|
"loss": 0.963, |
|
"step": 78000 |
|
}, |
|
{ |
|
"epoch": 2.24, |
|
"learning_rate": 1.2620115615744313e-05, |
|
"loss": 0.9777, |
|
"step": 78500 |
|
}, |
|
{ |
|
"epoch": 2.26, |
|
"learning_rate": 1.2382027180175805e-05, |
|
"loss": 0.9582, |
|
"step": 79000 |
|
}, |
|
{ |
|
"epoch": 2.27, |
|
"learning_rate": 1.2143938744607297e-05, |
|
"loss": 0.9625, |
|
"step": 79500 |
|
}, |
|
{ |
|
"epoch": 2.29, |
|
"learning_rate": 1.1905850309038789e-05, |
|
"loss": 0.9356, |
|
"step": 80000 |
|
}, |
|
{ |
|
"epoch": 2.3, |
|
"learning_rate": 1.1667761873470282e-05, |
|
"loss": 0.9154, |
|
"step": 80500 |
|
}, |
|
{ |
|
"epoch": 2.31, |
|
"learning_rate": 1.1429673437901774e-05, |
|
"loss": 0.9318, |
|
"step": 81000 |
|
}, |
|
{ |
|
"epoch": 2.33, |
|
"learning_rate": 1.1191585002333268e-05, |
|
"loss": 0.9404, |
|
"step": 81500 |
|
}, |
|
{ |
|
"epoch": 2.34, |
|
"learning_rate": 1.095349656676476e-05, |
|
"loss": 0.9471, |
|
"step": 82000 |
|
}, |
|
{ |
|
"epoch": 2.36, |
|
"learning_rate": 1.0715408131196251e-05, |
|
"loss": 0.9413, |
|
"step": 82500 |
|
}, |
|
{ |
|
"epoch": 2.37, |
|
"learning_rate": 1.0477319695627745e-05, |
|
"loss": 0.9553, |
|
"step": 83000 |
|
}, |
|
{ |
|
"epoch": 2.39, |
|
"learning_rate": 1.0239231260059237e-05, |
|
"loss": 0.9483, |
|
"step": 83500 |
|
}, |
|
{ |
|
"epoch": 2.4, |
|
"learning_rate": 1.0001142824490729e-05, |
|
"loss": 0.9399, |
|
"step": 84000 |
|
}, |
|
{ |
|
"epoch": 2.41, |
|
"learning_rate": 9.763054388922222e-06, |
|
"loss": 0.9261, |
|
"step": 84500 |
|
}, |
|
{ |
|
"epoch": 2.43, |
|
"learning_rate": 9.524965953353714e-06, |
|
"loss": 0.9511, |
|
"step": 85000 |
|
}, |
|
{ |
|
"epoch": 2.44, |
|
"learning_rate": 9.286877517785208e-06, |
|
"loss": 0.9393, |
|
"step": 85500 |
|
}, |
|
{ |
|
"epoch": 2.46, |
|
"learning_rate": 9.0487890822167e-06, |
|
"loss": 0.9468, |
|
"step": 86000 |
|
}, |
|
{ |
|
"epoch": 2.47, |
|
"learning_rate": 8.810700646648191e-06, |
|
"loss": 0.8873, |
|
"step": 86500 |
|
}, |
|
{ |
|
"epoch": 2.49, |
|
"learning_rate": 8.572612211079683e-06, |
|
"loss": 0.92, |
|
"step": 87000 |
|
}, |
|
{ |
|
"epoch": 2.5, |
|
"learning_rate": 8.334523775511175e-06, |
|
"loss": 0.9074, |
|
"step": 87500 |
|
}, |
|
{ |
|
"epoch": 2.51, |
|
"learning_rate": 8.096435339942669e-06, |
|
"loss": 0.8877, |
|
"step": 88000 |
|
}, |
|
{ |
|
"epoch": 2.53, |
|
"learning_rate": 7.858346904374162e-06, |
|
"loss": 0.9198, |
|
"step": 88500 |
|
}, |
|
{ |
|
"epoch": 2.54, |
|
"learning_rate": 7.620258468805654e-06, |
|
"loss": 0.9164, |
|
"step": 89000 |
|
}, |
|
{ |
|
"epoch": 2.56, |
|
"learning_rate": 7.382170033237146e-06, |
|
"loss": 0.9222, |
|
"step": 89500 |
|
}, |
|
{ |
|
"epoch": 2.57, |
|
"learning_rate": 7.144081597668638e-06, |
|
"loss": 0.9134, |
|
"step": 90000 |
|
}, |
|
{ |
|
"epoch": 2.59, |
|
"learning_rate": 6.905993162100131e-06, |
|
"loss": 0.8986, |
|
"step": 90500 |
|
}, |
|
{ |
|
"epoch": 2.6, |
|
"learning_rate": 6.667904726531624e-06, |
|
"loss": 0.9308, |
|
"step": 91000 |
|
}, |
|
{ |
|
"epoch": 2.61, |
|
"learning_rate": 6.429816290963116e-06, |
|
"loss": 0.8942, |
|
"step": 91500 |
|
}, |
|
{ |
|
"epoch": 2.63, |
|
"learning_rate": 6.1917278553946075e-06, |
|
"loss": 0.9094, |
|
"step": 92000 |
|
}, |
|
{ |
|
"epoch": 2.64, |
|
"learning_rate": 5.953639419826101e-06, |
|
"loss": 0.8926, |
|
"step": 92500 |
|
}, |
|
{ |
|
"epoch": 2.66, |
|
"learning_rate": 5.715550984257593e-06, |
|
"loss": 0.9168, |
|
"step": 93000 |
|
}, |
|
{ |
|
"epoch": 2.67, |
|
"learning_rate": 5.477462548689085e-06, |
|
"loss": 0.8975, |
|
"step": 93500 |
|
}, |
|
{ |
|
"epoch": 2.69, |
|
"learning_rate": 5.2393741131205775e-06, |
|
"loss": 0.8934, |
|
"step": 94000 |
|
}, |
|
{ |
|
"epoch": 2.7, |
|
"learning_rate": 5.00128567755207e-06, |
|
"loss": 0.9044, |
|
"step": 94500 |
|
}, |
|
{ |
|
"epoch": 2.71, |
|
"learning_rate": 4.763197241983563e-06, |
|
"loss": 0.9007, |
|
"step": 95000 |
|
}, |
|
{ |
|
"epoch": 2.73, |
|
"learning_rate": 4.525108806415055e-06, |
|
"loss": 0.9012, |
|
"step": 95500 |
|
}, |
|
{ |
|
"epoch": 2.74, |
|
"learning_rate": 4.287020370846547e-06, |
|
"loss": 0.8978, |
|
"step": 96000 |
|
}, |
|
{ |
|
"epoch": 2.76, |
|
"learning_rate": 4.04893193527804e-06, |
|
"loss": 0.9201, |
|
"step": 96500 |
|
}, |
|
{ |
|
"epoch": 2.77, |
|
"learning_rate": 3.8108434997095323e-06, |
|
"loss": 0.9187, |
|
"step": 97000 |
|
}, |
|
{ |
|
"epoch": 2.79, |
|
"learning_rate": 3.572755064141025e-06, |
|
"loss": 0.8854, |
|
"step": 97500 |
|
}, |
|
{ |
|
"epoch": 2.8, |
|
"learning_rate": 3.3346666285725173e-06, |
|
"loss": 0.882, |
|
"step": 98000 |
|
}, |
|
{ |
|
"epoch": 2.81, |
|
"learning_rate": 3.0965781930040096e-06, |
|
"loss": 0.9036, |
|
"step": 98500 |
|
}, |
|
{ |
|
"epoch": 2.83, |
|
"learning_rate": 2.8584897574355022e-06, |
|
"loss": 0.892, |
|
"step": 99000 |
|
}, |
|
{ |
|
"epoch": 2.84, |
|
"learning_rate": 2.620401321866994e-06, |
|
"loss": 0.9093, |
|
"step": 99500 |
|
}, |
|
{ |
|
"epoch": 2.86, |
|
"learning_rate": 2.3823128862984868e-06, |
|
"loss": 0.8837, |
|
"step": 100000 |
|
}, |
|
{ |
|
"epoch": 2.87, |
|
"learning_rate": 2.1442244507299795e-06, |
|
"loss": 0.856, |
|
"step": 100500 |
|
}, |
|
{ |
|
"epoch": 2.89, |
|
"learning_rate": 1.9061360151614717e-06, |
|
"loss": 0.8796, |
|
"step": 101000 |
|
}, |
|
{ |
|
"epoch": 2.9, |
|
"learning_rate": 1.6680475795929642e-06, |
|
"loss": 0.8924, |
|
"step": 101500 |
|
}, |
|
{ |
|
"epoch": 2.91, |
|
"learning_rate": 1.4299591440244565e-06, |
|
"loss": 0.8936, |
|
"step": 102000 |
|
}, |
|
{ |
|
"epoch": 2.93, |
|
"learning_rate": 1.191870708455949e-06, |
|
"loss": 0.8947, |
|
"step": 102500 |
|
}, |
|
{ |
|
"epoch": 2.94, |
|
"learning_rate": 9.537822728874414e-07, |
|
"loss": 0.8697, |
|
"step": 103000 |
|
}, |
|
{ |
|
"epoch": 2.96, |
|
"learning_rate": 7.156938373189338e-07, |
|
"loss": 0.8903, |
|
"step": 103500 |
|
}, |
|
{ |
|
"epoch": 2.97, |
|
"learning_rate": 4.776054017504262e-07, |
|
"loss": 0.8813, |
|
"step": 104000 |
|
}, |
|
{ |
|
"epoch": 2.99, |
|
"learning_rate": 2.395169661819186e-07, |
|
"loss": 0.8902, |
|
"step": 104500 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"learning_rate": 1.4285306134110454e-09, |
|
"loss": 0.8976, |
|
"step": 105000 |
|
}, |
|
{ |
|
"epoch": 3.0, |
|
"step": 105003, |
|
"total_flos": 3.655850016360038e+16, |
|
"train_loss": 1.3722702726657479, |
|
"train_runtime": 13766.7833, |
|
"train_samples_per_second": 30.508, |
|
"train_steps_per_second": 7.627 |
|
} |
|
], |
|
"max_steps": 105003, |
|
"num_train_epochs": 3, |
|
"total_flos": 3.655850016360038e+16, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|