| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.14417531718569782, |
| "eval_steps": 500, |
| "global_step": 500, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0014417531718569781, |
| "grad_norm": 322.0, |
| "kd_loss": 1.997420072555542, |
| "learning_rate": 4.444444444444445e-07, |
| "loss": 2.1422, |
| "step": 5, |
| "student_loss": 0.0605761781334877, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0028835063437139563, |
| "grad_norm": 186.0, |
| "kd_loss": 1.8595829010009766, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 2.0829, |
| "step": 10, |
| "student_loss": 0.09249057620763779, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.004325259515570935, |
| "grad_norm": 117.5, |
| "kd_loss": 1.5637085437774658, |
| "learning_rate": 1.5555555555555558e-06, |
| "loss": 1.972, |
| "step": 15, |
| "student_loss": 0.026328546926379204, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0057670126874279125, |
| "grad_norm": 106.5, |
| "kd_loss": 1.4912058115005493, |
| "learning_rate": 2.1111111111111114e-06, |
| "loss": 1.8011, |
| "step": 20, |
| "student_loss": 0.13309946656227112, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.00720876585928489, |
| "grad_norm": 86.5, |
| "kd_loss": 1.1542646884918213, |
| "learning_rate": 2.666666666666667e-06, |
| "loss": 1.5427, |
| "step": 25, |
| "student_loss": 0.036511313170194626, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.00865051903114187, |
| "grad_norm": 58.5, |
| "kd_loss": 0.9403514266014099, |
| "learning_rate": 3.2222222222222227e-06, |
| "loss": 1.2794, |
| "step": 30, |
| "student_loss": 0.01801167242228985, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.010092272202998846, |
| "grad_norm": 31.5, |
| "kd_loss": 0.6764306426048279, |
| "learning_rate": 3.777777777777778e-06, |
| "loss": 1.0897, |
| "step": 35, |
| "student_loss": 0.0011638500727713108, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.011534025374855825, |
| "grad_norm": 27.625, |
| "kd_loss": 0.7702282667160034, |
| "learning_rate": 4.333333333333334e-06, |
| "loss": 0.9024, |
| "step": 40, |
| "student_loss": 0.08332191407680511, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.012975778546712802, |
| "grad_norm": 20.875, |
| "kd_loss": 1.050789475440979, |
| "learning_rate": 4.888888888888889e-06, |
| "loss": 0.7206, |
| "step": 45, |
| "student_loss": 0.07726961374282837, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.01441753171856978, |
| "grad_norm": 44.25, |
| "kd_loss": 0.4379478096961975, |
| "learning_rate": 5.444444444444445e-06, |
| "loss": 0.6754, |
| "step": 50, |
| "student_loss": 0.007436605170369148, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.015859284890426758, |
| "grad_norm": 17.5, |
| "kd_loss": 0.3016382157802582, |
| "learning_rate": 6e-06, |
| "loss": 0.5857, |
| "step": 55, |
| "student_loss": 0.002162981079891324, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.01730103806228374, |
| "grad_norm": 24.625, |
| "kd_loss": 0.2297545075416565, |
| "learning_rate": 6.555555555555556e-06, |
| "loss": 0.5265, |
| "step": 60, |
| "student_loss": 0.012505847960710526, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.018742791234140715, |
| "grad_norm": 14.75, |
| "kd_loss": 0.2053602635860443, |
| "learning_rate": 7.111111111111112e-06, |
| "loss": 0.5101, |
| "step": 65, |
| "student_loss": 0.04744679108262062, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.020184544405997693, |
| "grad_norm": 34.0, |
| "kd_loss": 0.7437167763710022, |
| "learning_rate": 7.666666666666667e-06, |
| "loss": 0.4523, |
| "step": 70, |
| "student_loss": 0.029973715543746948, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02162629757785467, |
| "grad_norm": 14.0625, |
| "kd_loss": 0.18727970123291016, |
| "learning_rate": 8.222222222222222e-06, |
| "loss": 0.4534, |
| "step": 75, |
| "student_loss": 0.001686512609012425, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02306805074971165, |
| "grad_norm": 31.0, |
| "kd_loss": 0.18602637946605682, |
| "learning_rate": 8.777777777777778e-06, |
| "loss": 0.4338, |
| "step": 80, |
| "student_loss": 0.0014192159287631512, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.024509803921568627, |
| "grad_norm": 14.75, |
| "kd_loss": 0.11737873405218124, |
| "learning_rate": 9.333333333333334e-06, |
| "loss": 0.4378, |
| "step": 85, |
| "student_loss": 0.000477277091704309, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.025951557093425604, |
| "grad_norm": 17.125, |
| "kd_loss": 0.7024360299110413, |
| "learning_rate": 9.88888888888889e-06, |
| "loss": 0.3816, |
| "step": 90, |
| "student_loss": 0.027783483266830444, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.027393310265282585, |
| "grad_norm": 19.125, |
| "kd_loss": 0.3916384279727936, |
| "learning_rate": 9.99995337991255e-06, |
| "loss": 0.3905, |
| "step": 95, |
| "student_loss": 0.013136160559952259, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02883506343713956, |
| "grad_norm": 13.9375, |
| "kd_loss": 0.09685378521680832, |
| "learning_rate": 9.999763987297266e-06, |
| "loss": 0.3485, |
| "step": 100, |
| "student_loss": 0.0006464557372964919, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03027681660899654, |
| "grad_norm": 12.5, |
| "kd_loss": 0.48202934861183167, |
| "learning_rate": 9.999428913912876e-06, |
| "loss": 0.3555, |
| "step": 105, |
| "student_loss": 0.005315156653523445, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.031718569780853516, |
| "grad_norm": 9.8125, |
| "kd_loss": 0.42474475502967834, |
| "learning_rate": 9.998948169522587e-06, |
| "loss": 0.3616, |
| "step": 110, |
| "student_loss": 0.0304458886384964, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03316032295271049, |
| "grad_norm": 11.9375, |
| "kd_loss": 0.30219101905822754, |
| "learning_rate": 9.998321768134101e-06, |
| "loss": 0.3669, |
| "step": 115, |
| "student_loss": 0.0007903591031208634, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03460207612456748, |
| "grad_norm": 47.0, |
| "kd_loss": 0.14825306832790375, |
| "learning_rate": 9.997549727999212e-06, |
| "loss": 0.3014, |
| "step": 120, |
| "student_loss": 0.0007933085435070097, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.036043829296424454, |
| "grad_norm": 10.6875, |
| "kd_loss": 0.6828317046165466, |
| "learning_rate": 9.996632071613257e-06, |
| "loss": 0.3416, |
| "step": 125, |
| "student_loss": 0.056072596460580826, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03748558246828143, |
| "grad_norm": 21.5, |
| "kd_loss": 0.16518603265285492, |
| "learning_rate": 9.995568825714479e-06, |
| "loss": 0.3623, |
| "step": 130, |
| "student_loss": 0.000613231910392642, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03892733564013841, |
| "grad_norm": 11.875, |
| "kd_loss": 0.4139902889728546, |
| "learning_rate": 9.994360021283235e-06, |
| "loss": 0.3269, |
| "step": 135, |
| "student_loss": 0.059860486537218094, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.040369088811995385, |
| "grad_norm": 17.0, |
| "kd_loss": 0.15953019261360168, |
| "learning_rate": 9.993005693541098e-06, |
| "loss": 0.2964, |
| "step": 140, |
| "student_loss": 0.0005189953953959048, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04181084198385236, |
| "grad_norm": 19.5, |
| "kd_loss": 0.7493298053741455, |
| "learning_rate": 9.991505881949837e-06, |
| "loss": 0.3058, |
| "step": 145, |
| "student_loss": 0.053292322903871536, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04325259515570934, |
| "grad_norm": 15.8125, |
| "kd_loss": 0.09782249480485916, |
| "learning_rate": 9.989860630210251e-06, |
| "loss": 0.3046, |
| "step": 150, |
| "student_loss": 0.0007303969468921423, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04469434832756632, |
| "grad_norm": 12.0625, |
| "kd_loss": 0.119688019156456, |
| "learning_rate": 9.98806998626091e-06, |
| "loss": 0.3088, |
| "step": 155, |
| "student_loss": 0.0005232616094872355, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0461361014994233, |
| "grad_norm": 17.625, |
| "kd_loss": 0.2501586973667145, |
| "learning_rate": 9.98613400227676e-06, |
| "loss": 0.3508, |
| "step": 160, |
| "student_loss": 0.0909157246351242, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04757785467128028, |
| "grad_norm": 25.75, |
| "kd_loss": 0.23997722566127777, |
| "learning_rate": 9.984052734667586e-06, |
| "loss": 0.309, |
| "step": 165, |
| "student_loss": 0.0003585258382372558, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.049019607843137254, |
| "grad_norm": 53.75, |
| "kd_loss": 0.08952173590660095, |
| "learning_rate": 9.981826244076387e-06, |
| "loss": 0.2906, |
| "step": 170, |
| "student_loss": 0.0005561846191994846, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05046136101499423, |
| "grad_norm": 40.0, |
| "kd_loss": 0.22894923388957977, |
| "learning_rate": 9.979454595377594e-06, |
| "loss": 0.2992, |
| "step": 175, |
| "student_loss": 0.0037626794073730707, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05190311418685121, |
| "grad_norm": 23.75, |
| "kd_loss": 0.23231007158756256, |
| "learning_rate": 9.97693785767519e-06, |
| "loss": 0.278, |
| "step": 180, |
| "student_loss": 0.0004788874357473105, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05334486735870819, |
| "grad_norm": 32.25, |
| "kd_loss": 0.32399287819862366, |
| "learning_rate": 9.9742761043007e-06, |
| "loss": 0.3037, |
| "step": 185, |
| "student_loss": 0.05988827347755432, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05478662053056517, |
| "grad_norm": 18.75, |
| "kd_loss": 0.46008631587028503, |
| "learning_rate": 9.971469412811032e-06, |
| "loss": 0.2833, |
| "step": 190, |
| "student_loss": 0.0008702761260792613, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.056228373702422146, |
| "grad_norm": 10.1875, |
| "kd_loss": 0.41200897097587585, |
| "learning_rate": 9.968517864986247e-06, |
| "loss": 0.2848, |
| "step": 195, |
| "student_loss": 0.01731002703309059, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05767012687427912, |
| "grad_norm": 10.875, |
| "kd_loss": 0.17150147259235382, |
| "learning_rate": 9.965421546827153e-06, |
| "loss": 0.2805, |
| "step": 200, |
| "student_loss": 0.0014504786813631654, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0591118800461361, |
| "grad_norm": 20.875, |
| "kd_loss": 0.1884479522705078, |
| "learning_rate": 9.962180548552812e-06, |
| "loss": 0.2697, |
| "step": 205, |
| "student_loss": 0.04227042570710182, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06055363321799308, |
| "grad_norm": 23.75, |
| "kd_loss": 0.24175123870372772, |
| "learning_rate": 9.958794964597906e-06, |
| "loss": 0.2654, |
| "step": 210, |
| "student_loss": 0.0005445634596981108, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.061995386389850055, |
| "grad_norm": 25.875, |
| "kd_loss": 0.2347751259803772, |
| "learning_rate": 9.955264893609986e-06, |
| "loss": 0.294, |
| "step": 215, |
| "student_loss": 0.0005844403640367091, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06343713956170703, |
| "grad_norm": 17.25, |
| "kd_loss": 0.34608691930770874, |
| "learning_rate": 9.951590438446597e-06, |
| "loss": 0.2777, |
| "step": 220, |
| "student_loss": 0.06770455092191696, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06487889273356401, |
| "grad_norm": 37.0, |
| "kd_loss": 0.13642175495624542, |
| "learning_rate": 9.947771706172285e-06, |
| "loss": 0.2711, |
| "step": 225, |
| "student_loss": 0.029741473495960236, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06632064590542099, |
| "grad_norm": 13.5, |
| "kd_loss": 0.24233467876911163, |
| "learning_rate": 9.943808808055472e-06, |
| "loss": 0.2623, |
| "step": 230, |
| "student_loss": 0.0009022229351103306, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06776239907727798, |
| "grad_norm": 11.875, |
| "kd_loss": 0.6708654165267944, |
| "learning_rate": 9.93970185956522e-06, |
| "loss": 0.2862, |
| "step": 235, |
| "student_loss": 0.019873904064297676, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06920415224913495, |
| "grad_norm": 28.0, |
| "kd_loss": 0.18894419074058533, |
| "learning_rate": 9.93545098036786e-06, |
| "loss": 0.2625, |
| "step": 240, |
| "student_loss": 0.0006731516914442182, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07064590542099193, |
| "grad_norm": 20.25, |
| "kd_loss": 0.13373719155788422, |
| "learning_rate": 9.931056294323507e-06, |
| "loss": 0.2638, |
| "step": 245, |
| "student_loss": 0.0005933287902735174, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07208765859284891, |
| "grad_norm": 10.0, |
| "kd_loss": 0.12812890112400055, |
| "learning_rate": 9.926517929482454e-06, |
| "loss": 0.2926, |
| "step": 250, |
| "student_loss": 0.00034213275648653507, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07352941176470588, |
| "grad_norm": 16.75, |
| "kd_loss": 0.08018118143081665, |
| "learning_rate": 9.921836018081438e-06, |
| "loss": 0.2428, |
| "step": 255, |
| "student_loss": 0.00025593984173610806, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07497116493656286, |
| "grad_norm": 10.75, |
| "kd_loss": 0.19167569279670715, |
| "learning_rate": 9.917010696539793e-06, |
| "loss": 0.305, |
| "step": 260, |
| "student_loss": 0.04356108978390694, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07641291810841984, |
| "grad_norm": 26.625, |
| "kd_loss": 0.0696549043059349, |
| "learning_rate": 9.912042105455462e-06, |
| "loss": 0.2747, |
| "step": 265, |
| "student_loss": 0.0003771705087274313, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07785467128027682, |
| "grad_norm": 12.0625, |
| "kd_loss": 0.06586078554391861, |
| "learning_rate": 9.906930389600917e-06, |
| "loss": 0.2801, |
| "step": 270, |
| "student_loss": 0.00041506686829961836, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07929642445213379, |
| "grad_norm": 20.75, |
| "kd_loss": 0.14978329837322235, |
| "learning_rate": 9.901675697918925e-06, |
| "loss": 0.2886, |
| "step": 275, |
| "student_loss": 0.04765838012099266, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08073817762399077, |
| "grad_norm": 21.75, |
| "kd_loss": 0.17263588309288025, |
| "learning_rate": 9.896278183518216e-06, |
| "loss": 0.2897, |
| "step": 280, |
| "student_loss": 0.0004967491258867085, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08217993079584775, |
| "grad_norm": 20.0, |
| "kd_loss": 0.10580945014953613, |
| "learning_rate": 9.890738003669029e-06, |
| "loss": 0.293, |
| "step": 285, |
| "student_loss": 0.0005849753506481647, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08362168396770472, |
| "grad_norm": 12.1875, |
| "kd_loss": 0.09176977723836899, |
| "learning_rate": 9.885055319798514e-06, |
| "loss": 0.3088, |
| "step": 290, |
| "student_loss": 0.0006238222122192383, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0850634371395617, |
| "grad_norm": 10.9375, |
| "kd_loss": 0.07891599833965302, |
| "learning_rate": 9.879230297486034e-06, |
| "loss": 0.2957, |
| "step": 295, |
| "student_loss": 0.0005840823869220912, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08650519031141868, |
| "grad_norm": 7.9375, |
| "kd_loss": 0.3616674840450287, |
| "learning_rate": 9.873263106458353e-06, |
| "loss": 0.2483, |
| "step": 300, |
| "student_loss": 0.019933484494686127, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08794694348327567, |
| "grad_norm": 15.9375, |
| "kd_loss": 0.15585829317569733, |
| "learning_rate": 9.86715392058467e-06, |
| "loss": 0.2522, |
| "step": 305, |
| "student_loss": 0.018561875447630882, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08938869665513265, |
| "grad_norm": 10.1875, |
| "kd_loss": 0.06951466202735901, |
| "learning_rate": 9.860902917871566e-06, |
| "loss": 0.2646, |
| "step": 310, |
| "student_loss": 0.00046623480739071965, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09083044982698962, |
| "grad_norm": 11.5, |
| "kd_loss": 0.2836111783981323, |
| "learning_rate": 9.854510280457819e-06, |
| "loss": 0.2393, |
| "step": 315, |
| "student_loss": 0.012322928756475449, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0922722029988466, |
| "grad_norm": 13.6875, |
| "kd_loss": 0.10447105020284653, |
| "learning_rate": 9.847976194609083e-06, |
| "loss": 0.2668, |
| "step": 320, |
| "student_loss": 0.00048153125680983067, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09371395617070358, |
| "grad_norm": 11.0625, |
| "kd_loss": 0.4263528883457184, |
| "learning_rate": 9.841300850712479e-06, |
| "loss": 0.2377, |
| "step": 325, |
| "student_loss": 0.00044430175330489874, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09515570934256055, |
| "grad_norm": 17.5, |
| "kd_loss": 0.6019083261489868, |
| "learning_rate": 9.834484443271035e-06, |
| "loss": 0.2697, |
| "step": 330, |
| "student_loss": 0.05738024041056633, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09659746251441753, |
| "grad_norm": 7.125, |
| "kd_loss": 0.0715835765004158, |
| "learning_rate": 9.827527170898019e-06, |
| "loss": 0.2054, |
| "step": 335, |
| "student_loss": 0.0005960066919215024, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09803921568627451, |
| "grad_norm": 15.1875, |
| "kd_loss": 0.10987883061170578, |
| "learning_rate": 9.820429236311158e-06, |
| "loss": 0.2695, |
| "step": 340, |
| "student_loss": 0.009573950432240963, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09948096885813149, |
| "grad_norm": 34.0, |
| "kd_loss": 0.17168864607810974, |
| "learning_rate": 9.81319084632673e-06, |
| "loss": 0.258, |
| "step": 345, |
| "student_loss": 0.016774671152234077, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10092272202998846, |
| "grad_norm": 21.375, |
| "kd_loss": 0.532006025314331, |
| "learning_rate": 9.805812211853534e-06, |
| "loss": 0.2368, |
| "step": 350, |
| "student_loss": 0.0042114779353141785, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10236447520184544, |
| "grad_norm": 16.625, |
| "kd_loss": 0.0480809211730957, |
| "learning_rate": 9.798293547886748e-06, |
| "loss": 0.2384, |
| "step": 355, |
| "student_loss": 0.000468328595161438, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10380622837370242, |
| "grad_norm": 19.625, |
| "kd_loss": 0.06214166805148125, |
| "learning_rate": 9.790635073501662e-06, |
| "loss": 0.2473, |
| "step": 360, |
| "student_loss": 0.0004774518311023712, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1052479815455594, |
| "grad_norm": 11.0, |
| "kd_loss": 0.23296360671520233, |
| "learning_rate": 9.782837011847294e-06, |
| "loss": 0.3018, |
| "step": 365, |
| "student_loss": 0.045026302337646484, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10668973471741638, |
| "grad_norm": 14.5625, |
| "kd_loss": 0.39091649651527405, |
| "learning_rate": 9.774899590139897e-06, |
| "loss": 0.2487, |
| "step": 370, |
| "student_loss": 0.019704369828104973, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10813148788927336, |
| "grad_norm": 9.1875, |
| "kd_loss": 0.1680285483598709, |
| "learning_rate": 9.766823039656328e-06, |
| "loss": 0.2398, |
| "step": 375, |
| "student_loss": 0.038442861288785934, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10957324106113034, |
| "grad_norm": 17.875, |
| "kd_loss": 0.06734355539083481, |
| "learning_rate": 9.758607595727309e-06, |
| "loss": 0.2636, |
| "step": 380, |
| "student_loss": 0.0004427676321938634, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11101499423298732, |
| "grad_norm": 16.25, |
| "kd_loss": 0.16070111095905304, |
| "learning_rate": 9.75025349773058e-06, |
| "loss": 0.2488, |
| "step": 385, |
| "student_loss": 0.0003361160634085536, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11245674740484429, |
| "grad_norm": 10.0, |
| "kd_loss": 0.06088823080062866, |
| "learning_rate": 9.741760989083914e-06, |
| "loss": 0.2351, |
| "step": 390, |
| "student_loss": 0.00025634359917603433, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11389850057670127, |
| "grad_norm": 17.375, |
| "kd_loss": 0.1013672724366188, |
| "learning_rate": 9.73313031723803e-06, |
| "loss": 0.2348, |
| "step": 395, |
| "student_loss": 0.00046920194290578365, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11534025374855825, |
| "grad_norm": 12.9375, |
| "kd_loss": 0.12313513457775116, |
| "learning_rate": 9.724361733669383e-06, |
| "loss": 0.1916, |
| "step": 400, |
| "student_loss": 0.0006792352651245892, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11678200692041522, |
| "grad_norm": 12.4375, |
| "kd_loss": 0.11874714493751526, |
| "learning_rate": 9.715455493872827e-06, |
| "loss": 0.2445, |
| "step": 405, |
| "student_loss": 0.0060541993007063866, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1182237600922722, |
| "grad_norm": 15.875, |
| "kd_loss": 0.3652341663837433, |
| "learning_rate": 9.70641185735419e-06, |
| "loss": 0.2546, |
| "step": 410, |
| "student_loss": 0.017243457958102226, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11966551326412918, |
| "grad_norm": 14.0, |
| "kd_loss": 0.1578780561685562, |
| "learning_rate": 9.697231087622691e-06, |
| "loss": 0.2759, |
| "step": 415, |
| "student_loss": 0.0010528146522119641, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12110726643598616, |
| "grad_norm": 13.1875, |
| "kd_loss": 0.3285977840423584, |
| "learning_rate": 9.687913452183277e-06, |
| "loss": 0.2341, |
| "step": 420, |
| "student_loss": 0.06141834706068039, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12254901960784313, |
| "grad_norm": 15.1875, |
| "kd_loss": 0.05784276872873306, |
| "learning_rate": 9.678459222528825e-06, |
| "loss": 0.2234, |
| "step": 425, |
| "student_loss": 0.0011342725483700633, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12399077277970011, |
| "grad_norm": 11.8125, |
| "kd_loss": 0.35227036476135254, |
| "learning_rate": 9.668868674132224e-06, |
| "loss": 0.3045, |
| "step": 430, |
| "student_loss": 0.05136450007557869, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1254325259515571, |
| "grad_norm": 13.0625, |
| "kd_loss": 0.09926026314496994, |
| "learning_rate": 9.65914208643836e-06, |
| "loss": 0.2054, |
| "step": 435, |
| "student_loss": 0.0007687933393754065, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12687427912341406, |
| "grad_norm": 10.8125, |
| "kd_loss": 0.3320644795894623, |
| "learning_rate": 9.649279742855964e-06, |
| "loss": 0.2635, |
| "step": 440, |
| "student_loss": 0.02549891546368599, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12831603229527105, |
| "grad_norm": 10.0, |
| "kd_loss": 0.19906648993492126, |
| "learning_rate": 9.639281930749363e-06, |
| "loss": 0.2341, |
| "step": 445, |
| "student_loss": 0.025979263707995415, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12975778546712802, |
| "grad_norm": 9.25, |
| "kd_loss": 0.164601668715477, |
| "learning_rate": 9.629148941430096e-06, |
| "loss": 0.2751, |
| "step": 450, |
| "student_loss": 0.0007244550506584346, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.131199538638985, |
| "grad_norm": 7.78125, |
| "kd_loss": 0.04286447539925575, |
| "learning_rate": 9.618881070148436e-06, |
| "loss": 0.2417, |
| "step": 455, |
| "student_loss": 0.000796566077042371, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13264129181084197, |
| "grad_norm": 13.1875, |
| "kd_loss": 0.17657798528671265, |
| "learning_rate": 9.608478616084784e-06, |
| "loss": 0.1985, |
| "step": 460, |
| "student_loss": 0.00047529107541777194, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13408304498269896, |
| "grad_norm": 6.90625, |
| "kd_loss": 0.2981259524822235, |
| "learning_rate": 9.597941882340947e-06, |
| "loss": 0.2786, |
| "step": 465, |
| "student_loss": 0.00040858631837181747, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13552479815455595, |
| "grad_norm": 25.0, |
| "kd_loss": 0.09742365032434464, |
| "learning_rate": 9.587271175931313e-06, |
| "loss": 0.2365, |
| "step": 470, |
| "student_loss": 0.0003401920839678496, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13696655132641292, |
| "grad_norm": 23.625, |
| "kd_loss": 0.24699804186820984, |
| "learning_rate": 9.5764668077739e-06, |
| "loss": 0.1986, |
| "step": 475, |
| "student_loss": 0.0005240993923507631, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1384083044982699, |
| "grad_norm": 21.75, |
| "kd_loss": 0.19472193717956543, |
| "learning_rate": 9.5655290926813e-06, |
| "loss": 0.2274, |
| "step": 480, |
| "student_loss": 0.008671707473695278, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13985005767012687, |
| "grad_norm": 17.125, |
| "kd_loss": 0.055427636951208115, |
| "learning_rate": 9.554458349351507e-06, |
| "loss": 0.2293, |
| "step": 485, |
| "student_loss": 0.0014853382017463446, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14129181084198386, |
| "grad_norm": 13.3125, |
| "kd_loss": 0.2377857267856598, |
| "learning_rate": 9.54325490035863e-06, |
| "loss": 0.2595, |
| "step": 490, |
| "student_loss": 0.000622899504378438, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14273356401384082, |
| "grad_norm": 9.3125, |
| "kd_loss": 0.3495025932788849, |
| "learning_rate": 9.531919072143488e-06, |
| "loss": 0.2468, |
| "step": 495, |
| "student_loss": 0.0005211852840147913, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14417531718569782, |
| "grad_norm": 10.3125, |
| "kd_loss": 0.18743589520454407, |
| "learning_rate": 9.520451195004107e-06, |
| "loss": 0.2461, |
| "step": 500, |
| "student_loss": 0.0002995518152602017, |
| "teacher_loss": 0.0 |
| } |
| ], |
| "logging_steps": 5, |
| "max_steps": 3000, |
| "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": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 0.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
