|
{ |
|
"best_metric": null, |
|
"best_model_checkpoint": null, |
|
"epoch": 0.07925500297206262, |
|
"eval_steps": 500, |
|
"global_step": 100, |
|
"is_hyper_param_search": false, |
|
"is_local_process_zero": true, |
|
"is_world_process_zero": true, |
|
"log_history": [ |
|
{ |
|
"epoch": 0.0007925500297206261, |
|
"grad_norm": 0.08596306294202805, |
|
"learning_rate": 1e-05, |
|
"loss": 1.6869, |
|
"step": 1 |
|
}, |
|
{ |
|
"epoch": 0.0015851000594412521, |
|
"grad_norm": 0.07810184359550476, |
|
"learning_rate": 2e-05, |
|
"loss": 1.5316, |
|
"step": 2 |
|
}, |
|
{ |
|
"epoch": 0.0023776500891618782, |
|
"grad_norm": 0.0829155296087265, |
|
"learning_rate": 3e-05, |
|
"loss": 1.7178, |
|
"step": 3 |
|
}, |
|
{ |
|
"epoch": 0.0031702001188825043, |
|
"grad_norm": 0.0854589119553566, |
|
"learning_rate": 4e-05, |
|
"loss": 1.6508, |
|
"step": 4 |
|
}, |
|
{ |
|
"epoch": 0.003962750148603131, |
|
"grad_norm": 0.0787849947810173, |
|
"learning_rate": 5e-05, |
|
"loss": 1.5097, |
|
"step": 5 |
|
}, |
|
{ |
|
"epoch": 0.0047553001783237564, |
|
"grad_norm": 0.08113064616918564, |
|
"learning_rate": 6e-05, |
|
"loss": 1.6618, |
|
"step": 6 |
|
}, |
|
{ |
|
"epoch": 0.005547850208044383, |
|
"grad_norm": 0.0847715213894844, |
|
"learning_rate": 7e-05, |
|
"loss": 1.6121, |
|
"step": 7 |
|
}, |
|
{ |
|
"epoch": 0.006340400237765009, |
|
"grad_norm": 0.0853838175535202, |
|
"learning_rate": 8e-05, |
|
"loss": 1.6069, |
|
"step": 8 |
|
}, |
|
{ |
|
"epoch": 0.007132950267485635, |
|
"grad_norm": 0.09020455926656723, |
|
"learning_rate": 9e-05, |
|
"loss": 1.6444, |
|
"step": 9 |
|
}, |
|
{ |
|
"epoch": 0.007925500297206262, |
|
"grad_norm": 0.08465927839279175, |
|
"learning_rate": 0.0001, |
|
"loss": 1.6896, |
|
"step": 10 |
|
}, |
|
{ |
|
"epoch": 0.008718050326926888, |
|
"grad_norm": 0.10944877564907074, |
|
"learning_rate": 9.99695413509548e-05, |
|
"loss": 1.6168, |
|
"step": 11 |
|
}, |
|
{ |
|
"epoch": 0.009510600356647513, |
|
"grad_norm": 0.10147999972105026, |
|
"learning_rate": 9.987820251299122e-05, |
|
"loss": 1.6489, |
|
"step": 12 |
|
}, |
|
{ |
|
"epoch": 0.01030315038636814, |
|
"grad_norm": 0.10642598569393158, |
|
"learning_rate": 9.972609476841367e-05, |
|
"loss": 1.5926, |
|
"step": 13 |
|
}, |
|
{ |
|
"epoch": 0.011095700416088766, |
|
"grad_norm": 0.12236135452985764, |
|
"learning_rate": 9.951340343707852e-05, |
|
"loss": 1.4658, |
|
"step": 14 |
|
}, |
|
{ |
|
"epoch": 0.011888250445809392, |
|
"grad_norm": 0.12448609620332718, |
|
"learning_rate": 9.924038765061042e-05, |
|
"loss": 1.621, |
|
"step": 15 |
|
}, |
|
{ |
|
"epoch": 0.012680800475530017, |
|
"grad_norm": 0.13601672649383545, |
|
"learning_rate": 9.890738003669029e-05, |
|
"loss": 1.4682, |
|
"step": 16 |
|
}, |
|
{ |
|
"epoch": 0.013473350505250644, |
|
"grad_norm": 0.10898814350366592, |
|
"learning_rate": 9.851478631379982e-05, |
|
"loss": 1.5103, |
|
"step": 17 |
|
}, |
|
{ |
|
"epoch": 0.01426590053497127, |
|
"grad_norm": 0.11529192328453064, |
|
"learning_rate": 9.806308479691595e-05, |
|
"loss": 1.5579, |
|
"step": 18 |
|
}, |
|
{ |
|
"epoch": 0.015058450564691897, |
|
"grad_norm": 0.11997439712285995, |
|
"learning_rate": 9.755282581475769e-05, |
|
"loss": 1.368, |
|
"step": 19 |
|
}, |
|
{ |
|
"epoch": 0.015851000594412523, |
|
"grad_norm": 0.11383643746376038, |
|
"learning_rate": 9.698463103929542e-05, |
|
"loss": 1.403, |
|
"step": 20 |
|
}, |
|
{ |
|
"epoch": 0.016643550624133148, |
|
"grad_norm": 0.10081478208303452, |
|
"learning_rate": 9.635919272833938e-05, |
|
"loss": 1.4791, |
|
"step": 21 |
|
}, |
|
{ |
|
"epoch": 0.017436100653853776, |
|
"grad_norm": 0.11111775040626526, |
|
"learning_rate": 9.567727288213005e-05, |
|
"loss": 1.4708, |
|
"step": 22 |
|
}, |
|
{ |
|
"epoch": 0.0182286506835744, |
|
"grad_norm": 0.11225850135087967, |
|
"learning_rate": 9.493970231495835e-05, |
|
"loss": 1.4379, |
|
"step": 23 |
|
}, |
|
{ |
|
"epoch": 0.019021200713295026, |
|
"grad_norm": 0.10765302181243896, |
|
"learning_rate": 9.414737964294636e-05, |
|
"loss": 1.3136, |
|
"step": 24 |
|
}, |
|
{ |
|
"epoch": 0.019813750743015654, |
|
"grad_norm": 0.10010719299316406, |
|
"learning_rate": 9.330127018922194e-05, |
|
"loss": 1.4687, |
|
"step": 25 |
|
}, |
|
{ |
|
"epoch": 0.02060630077273628, |
|
"grad_norm": 0.11463580280542374, |
|
"learning_rate": 9.24024048078213e-05, |
|
"loss": 1.3397, |
|
"step": 26 |
|
}, |
|
{ |
|
"epoch": 0.021398850802456904, |
|
"grad_norm": 0.11755277216434479, |
|
"learning_rate": 9.145187862775209e-05, |
|
"loss": 1.3989, |
|
"step": 27 |
|
}, |
|
{ |
|
"epoch": 0.022191400832177532, |
|
"grad_norm": 0.11401020735502243, |
|
"learning_rate": 9.045084971874738e-05, |
|
"loss": 1.3147, |
|
"step": 28 |
|
}, |
|
{ |
|
"epoch": 0.022983950861898157, |
|
"grad_norm": 0.1177113801240921, |
|
"learning_rate": 8.940053768033609e-05, |
|
"loss": 1.5029, |
|
"step": 29 |
|
}, |
|
{ |
|
"epoch": 0.023776500891618785, |
|
"grad_norm": 0.13379471004009247, |
|
"learning_rate": 8.83022221559489e-05, |
|
"loss": 1.3791, |
|
"step": 30 |
|
}, |
|
{ |
|
"epoch": 0.02456905092133941, |
|
"grad_norm": 0.1319996416568756, |
|
"learning_rate": 8.715724127386972e-05, |
|
"loss": 1.4348, |
|
"step": 31 |
|
}, |
|
{ |
|
"epoch": 0.025361600951060034, |
|
"grad_norm": 0.11248724907636642, |
|
"learning_rate": 8.596699001693255e-05, |
|
"loss": 1.3354, |
|
"step": 32 |
|
}, |
|
{ |
|
"epoch": 0.026154150980780663, |
|
"grad_norm": 0.1153276264667511, |
|
"learning_rate": 8.473291852294987e-05, |
|
"loss": 1.3007, |
|
"step": 33 |
|
}, |
|
{ |
|
"epoch": 0.026946701010501287, |
|
"grad_norm": 0.11299668252468109, |
|
"learning_rate": 8.345653031794292e-05, |
|
"loss": 1.1996, |
|
"step": 34 |
|
}, |
|
{ |
|
"epoch": 0.027739251040221916, |
|
"grad_norm": 0.1060878112912178, |
|
"learning_rate": 8.213938048432697e-05, |
|
"loss": 1.3233, |
|
"step": 35 |
|
}, |
|
{ |
|
"epoch": 0.02853180106994254, |
|
"grad_norm": 0.11341679096221924, |
|
"learning_rate": 8.07830737662829e-05, |
|
"loss": 1.2947, |
|
"step": 36 |
|
}, |
|
{ |
|
"epoch": 0.029324351099663165, |
|
"grad_norm": 0.10489525645971298, |
|
"learning_rate": 7.938926261462366e-05, |
|
"loss": 1.2775, |
|
"step": 37 |
|
}, |
|
{ |
|
"epoch": 0.030116901129383793, |
|
"grad_norm": 0.11247427016496658, |
|
"learning_rate": 7.795964517353735e-05, |
|
"loss": 1.2449, |
|
"step": 38 |
|
}, |
|
{ |
|
"epoch": 0.030909451159104418, |
|
"grad_norm": 0.10019715875387192, |
|
"learning_rate": 7.649596321166024e-05, |
|
"loss": 1.3006, |
|
"step": 39 |
|
}, |
|
{ |
|
"epoch": 0.031702001188825046, |
|
"grad_norm": 0.1318451166152954, |
|
"learning_rate": 7.500000000000001e-05, |
|
"loss": 1.2783, |
|
"step": 40 |
|
}, |
|
{ |
|
"epoch": 0.03249455121854567, |
|
"grad_norm": 0.11795549839735031, |
|
"learning_rate": 7.347357813929454e-05, |
|
"loss": 1.1728, |
|
"step": 41 |
|
}, |
|
{ |
|
"epoch": 0.033287101248266296, |
|
"grad_norm": 0.10496120899915695, |
|
"learning_rate": 7.191855733945387e-05, |
|
"loss": 1.2775, |
|
"step": 42 |
|
}, |
|
{ |
|
"epoch": 0.03407965127798692, |
|
"grad_norm": 0.11640025675296783, |
|
"learning_rate": 7.033683215379002e-05, |
|
"loss": 1.2791, |
|
"step": 43 |
|
}, |
|
{ |
|
"epoch": 0.03487220130770755, |
|
"grad_norm": 0.12168875336647034, |
|
"learning_rate": 6.873032967079561e-05, |
|
"loss": 1.1291, |
|
"step": 44 |
|
}, |
|
{ |
|
"epoch": 0.03566475133742818, |
|
"grad_norm": 0.12284629791975021, |
|
"learning_rate": 6.710100716628344e-05, |
|
"loss": 1.2398, |
|
"step": 45 |
|
}, |
|
{ |
|
"epoch": 0.0364573013671488, |
|
"grad_norm": 0.11582006514072418, |
|
"learning_rate": 6.545084971874738e-05, |
|
"loss": 1.2377, |
|
"step": 46 |
|
}, |
|
{ |
|
"epoch": 0.03724985139686943, |
|
"grad_norm": 0.1095830574631691, |
|
"learning_rate": 6.378186779084995e-05, |
|
"loss": 1.1439, |
|
"step": 47 |
|
}, |
|
{ |
|
"epoch": 0.03804240142659005, |
|
"grad_norm": 0.11340509355068207, |
|
"learning_rate": 6.209609477998338e-05, |
|
"loss": 1.2213, |
|
"step": 48 |
|
}, |
|
{ |
|
"epoch": 0.038834951456310676, |
|
"grad_norm": 0.11721540987491608, |
|
"learning_rate": 6.0395584540887963e-05, |
|
"loss": 1.246, |
|
"step": 49 |
|
}, |
|
{ |
|
"epoch": 0.03962750148603131, |
|
"grad_norm": 0.12324697524309158, |
|
"learning_rate": 5.868240888334653e-05, |
|
"loss": 1.1844, |
|
"step": 50 |
|
}, |
|
{ |
|
"epoch": 0.04042005151575193, |
|
"grad_norm": 0.12552472949028015, |
|
"learning_rate": 5.695865504800327e-05, |
|
"loss": 1.2225, |
|
"step": 51 |
|
}, |
|
{ |
|
"epoch": 0.04121260154547256, |
|
"grad_norm": 0.11364790052175522, |
|
"learning_rate": 5.522642316338268e-05, |
|
"loss": 1.2093, |
|
"step": 52 |
|
}, |
|
{ |
|
"epoch": 0.04200515157519318, |
|
"grad_norm": 0.12709645926952362, |
|
"learning_rate": 5.348782368720626e-05, |
|
"loss": 1.1447, |
|
"step": 53 |
|
}, |
|
{ |
|
"epoch": 0.04279770160491381, |
|
"grad_norm": 0.11646372824907303, |
|
"learning_rate": 5.174497483512506e-05, |
|
"loss": 1.3036, |
|
"step": 54 |
|
}, |
|
{ |
|
"epoch": 0.04359025163463444, |
|
"grad_norm": 0.1108243390917778, |
|
"learning_rate": 5e-05, |
|
"loss": 1.1812, |
|
"step": 55 |
|
}, |
|
{ |
|
"epoch": 0.044382801664355064, |
|
"grad_norm": 0.11163365840911865, |
|
"learning_rate": 4.825502516487497e-05, |
|
"loss": 1.2556, |
|
"step": 56 |
|
}, |
|
{ |
|
"epoch": 0.04517535169407569, |
|
"grad_norm": 0.11043145507574081, |
|
"learning_rate": 4.6512176312793736e-05, |
|
"loss": 1.1275, |
|
"step": 57 |
|
}, |
|
{ |
|
"epoch": 0.04596790172379631, |
|
"grad_norm": 0.11449386924505234, |
|
"learning_rate": 4.477357683661734e-05, |
|
"loss": 1.2372, |
|
"step": 58 |
|
}, |
|
{ |
|
"epoch": 0.04676045175351694, |
|
"grad_norm": 0.11405756324529648, |
|
"learning_rate": 4.3041344951996746e-05, |
|
"loss": 1.2642, |
|
"step": 59 |
|
}, |
|
{ |
|
"epoch": 0.04755300178323757, |
|
"grad_norm": 0.12575648725032806, |
|
"learning_rate": 4.131759111665349e-05, |
|
"loss": 1.2295, |
|
"step": 60 |
|
}, |
|
{ |
|
"epoch": 0.048345551812958194, |
|
"grad_norm": 0.11009961366653442, |
|
"learning_rate": 3.960441545911204e-05, |
|
"loss": 1.2357, |
|
"step": 61 |
|
}, |
|
{ |
|
"epoch": 0.04913810184267882, |
|
"grad_norm": 0.10563351958990097, |
|
"learning_rate": 3.790390522001662e-05, |
|
"loss": 1.1675, |
|
"step": 62 |
|
}, |
|
{ |
|
"epoch": 0.049930651872399444, |
|
"grad_norm": 0.10982260853052139, |
|
"learning_rate": 3.6218132209150045e-05, |
|
"loss": 1.1918, |
|
"step": 63 |
|
}, |
|
{ |
|
"epoch": 0.05072320190212007, |
|
"grad_norm": 0.10726086795330048, |
|
"learning_rate": 3.4549150281252636e-05, |
|
"loss": 1.1175, |
|
"step": 64 |
|
}, |
|
{ |
|
"epoch": 0.0515157519318407, |
|
"grad_norm": 0.10884430259466171, |
|
"learning_rate": 3.289899283371657e-05, |
|
"loss": 1.262, |
|
"step": 65 |
|
}, |
|
{ |
|
"epoch": 0.052308301961561325, |
|
"grad_norm": 0.11094468086957932, |
|
"learning_rate": 3.12696703292044e-05, |
|
"loss": 1.2229, |
|
"step": 66 |
|
}, |
|
{ |
|
"epoch": 0.05310085199128195, |
|
"grad_norm": 0.11852382123470306, |
|
"learning_rate": 2.9663167846209998e-05, |
|
"loss": 1.1713, |
|
"step": 67 |
|
}, |
|
{ |
|
"epoch": 0.053893402021002575, |
|
"grad_norm": 0.11185895651578903, |
|
"learning_rate": 2.8081442660546125e-05, |
|
"loss": 1.0654, |
|
"step": 68 |
|
}, |
|
{ |
|
"epoch": 0.0546859520507232, |
|
"grad_norm": 0.11876579374074936, |
|
"learning_rate": 2.6526421860705473e-05, |
|
"loss": 1.1556, |
|
"step": 69 |
|
}, |
|
{ |
|
"epoch": 0.05547850208044383, |
|
"grad_norm": 0.11526551842689514, |
|
"learning_rate": 2.500000000000001e-05, |
|
"loss": 1.2392, |
|
"step": 70 |
|
}, |
|
{ |
|
"epoch": 0.056271052110164456, |
|
"grad_norm": 0.11417282372713089, |
|
"learning_rate": 2.350403678833976e-05, |
|
"loss": 1.093, |
|
"step": 71 |
|
}, |
|
{ |
|
"epoch": 0.05706360213988508, |
|
"grad_norm": 0.1079481914639473, |
|
"learning_rate": 2.2040354826462668e-05, |
|
"loss": 1.1071, |
|
"step": 72 |
|
}, |
|
{ |
|
"epoch": 0.057856152169605705, |
|
"grad_norm": 0.127486452460289, |
|
"learning_rate": 2.061073738537635e-05, |
|
"loss": 1.1641, |
|
"step": 73 |
|
}, |
|
{ |
|
"epoch": 0.05864870219932633, |
|
"grad_norm": 0.11122499406337738, |
|
"learning_rate": 1.9216926233717085e-05, |
|
"loss": 1.1811, |
|
"step": 74 |
|
}, |
|
{ |
|
"epoch": 0.05944125222904696, |
|
"grad_norm": 0.11476968228816986, |
|
"learning_rate": 1.7860619515673033e-05, |
|
"loss": 1.1977, |
|
"step": 75 |
|
}, |
|
{ |
|
"epoch": 0.06023380225876759, |
|
"grad_norm": 0.10940071940422058, |
|
"learning_rate": 1.6543469682057106e-05, |
|
"loss": 1.1444, |
|
"step": 76 |
|
}, |
|
{ |
|
"epoch": 0.06102635228848821, |
|
"grad_norm": 0.11059749126434326, |
|
"learning_rate": 1.526708147705013e-05, |
|
"loss": 1.2987, |
|
"step": 77 |
|
}, |
|
{ |
|
"epoch": 0.061818902318208836, |
|
"grad_norm": 0.12557165324687958, |
|
"learning_rate": 1.4033009983067452e-05, |
|
"loss": 1.1911, |
|
"step": 78 |
|
}, |
|
{ |
|
"epoch": 0.06261145234792946, |
|
"grad_norm": 0.11463788896799088, |
|
"learning_rate": 1.2842758726130283e-05, |
|
"loss": 1.1579, |
|
"step": 79 |
|
}, |
|
{ |
|
"epoch": 0.06340400237765009, |
|
"grad_norm": 0.11591491103172302, |
|
"learning_rate": 1.1697777844051105e-05, |
|
"loss": 1.0952, |
|
"step": 80 |
|
}, |
|
{ |
|
"epoch": 0.06419655240737071, |
|
"grad_norm": 0.11377400159835815, |
|
"learning_rate": 1.0599462319663905e-05, |
|
"loss": 1.1789, |
|
"step": 81 |
|
}, |
|
{ |
|
"epoch": 0.06498910243709134, |
|
"grad_norm": 0.1166919395327568, |
|
"learning_rate": 9.549150281252633e-06, |
|
"loss": 1.2323, |
|
"step": 82 |
|
}, |
|
{ |
|
"epoch": 0.06578165246681197, |
|
"grad_norm": 0.12292219698429108, |
|
"learning_rate": 8.548121372247918e-06, |
|
"loss": 1.0583, |
|
"step": 83 |
|
}, |
|
{ |
|
"epoch": 0.06657420249653259, |
|
"grad_norm": 0.10546883940696716, |
|
"learning_rate": 7.597595192178702e-06, |
|
"loss": 1.1301, |
|
"step": 84 |
|
}, |
|
{ |
|
"epoch": 0.06736675252625322, |
|
"grad_norm": 0.1269674152135849, |
|
"learning_rate": 6.698729810778065e-06, |
|
"loss": 1.1646, |
|
"step": 85 |
|
}, |
|
{ |
|
"epoch": 0.06815930255597384, |
|
"grad_norm": 0.11046171933412552, |
|
"learning_rate": 5.852620357053651e-06, |
|
"loss": 1.0562, |
|
"step": 86 |
|
}, |
|
{ |
|
"epoch": 0.06895185258569447, |
|
"grad_norm": 0.11050822585821152, |
|
"learning_rate": 5.060297685041659e-06, |
|
"loss": 1.1041, |
|
"step": 87 |
|
}, |
|
{ |
|
"epoch": 0.0697444026154151, |
|
"grad_norm": 0.10655710846185684, |
|
"learning_rate": 4.322727117869951e-06, |
|
"loss": 1.0813, |
|
"step": 88 |
|
}, |
|
{ |
|
"epoch": 0.07053695264513572, |
|
"grad_norm": 0.11349563300609589, |
|
"learning_rate": 3.6408072716606346e-06, |
|
"loss": 1.1111, |
|
"step": 89 |
|
}, |
|
{ |
|
"epoch": 0.07132950267485635, |
|
"grad_norm": 0.11467929184436798, |
|
"learning_rate": 3.0153689607045845e-06, |
|
"loss": 1.1875, |
|
"step": 90 |
|
}, |
|
{ |
|
"epoch": 0.07212205270457697, |
|
"grad_norm": 0.10557202994823456, |
|
"learning_rate": 2.4471741852423237e-06, |
|
"loss": 1.2174, |
|
"step": 91 |
|
}, |
|
{ |
|
"epoch": 0.0729146027342976, |
|
"grad_norm": 0.12264101952314377, |
|
"learning_rate": 1.9369152030840556e-06, |
|
"loss": 1.1609, |
|
"step": 92 |
|
}, |
|
{ |
|
"epoch": 0.07370715276401824, |
|
"grad_norm": 0.12894296646118164, |
|
"learning_rate": 1.4852136862001764e-06, |
|
"loss": 1.173, |
|
"step": 93 |
|
}, |
|
{ |
|
"epoch": 0.07449970279373885, |
|
"grad_norm": 0.10504671931266785, |
|
"learning_rate": 1.0926199633097157e-06, |
|
"loss": 1.193, |
|
"step": 94 |
|
}, |
|
{ |
|
"epoch": 0.07529225282345949, |
|
"grad_norm": 0.13509008288383484, |
|
"learning_rate": 7.596123493895991e-07, |
|
"loss": 1.2547, |
|
"step": 95 |
|
}, |
|
{ |
|
"epoch": 0.0760848028531801, |
|
"grad_norm": 0.11495807766914368, |
|
"learning_rate": 4.865965629214819e-07, |
|
"loss": 1.0605, |
|
"step": 96 |
|
}, |
|
{ |
|
"epoch": 0.07687735288290073, |
|
"grad_norm": 0.11578365415334702, |
|
"learning_rate": 2.7390523158633554e-07, |
|
"loss": 1.0972, |
|
"step": 97 |
|
}, |
|
{ |
|
"epoch": 0.07766990291262135, |
|
"grad_norm": 0.1086273044347763, |
|
"learning_rate": 1.2179748700879012e-07, |
|
"loss": 1.2008, |
|
"step": 98 |
|
}, |
|
{ |
|
"epoch": 0.07846245294234198, |
|
"grad_norm": 0.1122601330280304, |
|
"learning_rate": 3.04586490452119e-08, |
|
"loss": 1.1364, |
|
"step": 99 |
|
}, |
|
{ |
|
"epoch": 0.07925500297206262, |
|
"grad_norm": 0.10719715058803558, |
|
"learning_rate": 0.0, |
|
"loss": 1.2351, |
|
"step": 100 |
|
}, |
|
{ |
|
"epoch": 0.07925500297206262, |
|
"eval_loss": 1.1689194440841675, |
|
"eval_runtime": 16.0517, |
|
"eval_samples_per_second": 12.833, |
|
"eval_steps_per_second": 12.833, |
|
"step": 100 |
|
} |
|
], |
|
"logging_steps": 1, |
|
"max_steps": 100, |
|
"num_input_tokens_seen": 0, |
|
"num_train_epochs": 1, |
|
"save_steps": 500, |
|
"stateful_callbacks": { |
|
"TrainerControl": { |
|
"args": { |
|
"should_epoch_stop": false, |
|
"should_evaluate": false, |
|
"should_log": false, |
|
"should_save": true, |
|
"should_training_stop": true |
|
}, |
|
"attributes": {} |
|
} |
|
}, |
|
"total_flos": 5173041589911552.0, |
|
"train_batch_size": 1, |
|
"trial_name": null, |
|
"trial_params": null |
|
} |
|
|