| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.28835063437139563, |
| "eval_steps": 500, |
| "global_step": 1000, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.0014417531718569781, |
| "grad_norm": 1952.0, |
| "kd_loss": 1.354009985923767, |
| "learning_rate": 4.444444444444445e-07, |
| "loss": 4.4194, |
| "step": 5, |
| "student_loss": 0.10154028236865997, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0028835063437139563, |
| "grad_norm": 768.0, |
| "kd_loss": 2.1859679222106934, |
| "learning_rate": 1.0000000000000002e-06, |
| "loss": 4.072, |
| "step": 10, |
| "student_loss": 0.23161625862121582, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.004325259515570935, |
| "grad_norm": 676.0, |
| "kd_loss": 0.8974658250808716, |
| "learning_rate": 1.5555555555555558e-06, |
| "loss": 3.0092, |
| "step": 15, |
| "student_loss": 0.07354968786239624, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0057670126874279125, |
| "grad_norm": 784.0, |
| "kd_loss": 5.459206581115723, |
| "learning_rate": 2.1111111111111114e-06, |
| "loss": 2.0182, |
| "step": 20, |
| "student_loss": 0.24417075514793396, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.00720876585928489, |
| "grad_norm": 132.0, |
| "kd_loss": 0.039198435842990875, |
| "learning_rate": 2.666666666666667e-06, |
| "loss": 0.9093, |
| "step": 25, |
| "student_loss": 0.3775336444377899, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.00865051903114187, |
| "grad_norm": 82.0, |
| "kd_loss": 0.01129975076764822, |
| "learning_rate": 3.2222222222222227e-06, |
| "loss": 0.6102, |
| "step": 30, |
| "student_loss": 0.48634102940559387, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.010092272202998846, |
| "grad_norm": 67.5, |
| "kd_loss": 0.01766686514019966, |
| "learning_rate": 3.777777777777778e-06, |
| "loss": 0.5753, |
| "step": 35, |
| "student_loss": 0.16448530554771423, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.011534025374855825, |
| "grad_norm": 66.0, |
| "kd_loss": 0.02908974140882492, |
| "learning_rate": 4.333333333333334e-06, |
| "loss": 0.441, |
| "step": 40, |
| "student_loss": 0.11244617402553558, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.012975778546712802, |
| "grad_norm": 68.0, |
| "kd_loss": 0.2916097640991211, |
| "learning_rate": 4.888888888888889e-06, |
| "loss": 0.2934, |
| "step": 45, |
| "student_loss": 0.07623665779829025, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.01441753171856978, |
| "grad_norm": 88.5, |
| "kd_loss": 0.19340172410011292, |
| "learning_rate": 5.444444444444445e-06, |
| "loss": 0.3841, |
| "step": 50, |
| "student_loss": 0.009455127641558647, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.015859284890426758, |
| "grad_norm": 115.0, |
| "kd_loss": 0.01024007610976696, |
| "learning_rate": 6e-06, |
| "loss": 0.2854, |
| "step": 55, |
| "student_loss": 0.0043172105215489864, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.01730103806228374, |
| "grad_norm": 57.75, |
| "kd_loss": 0.0018813807982951403, |
| "learning_rate": 6.555555555555556e-06, |
| "loss": 0.2133, |
| "step": 60, |
| "student_loss": 0.012564695440232754, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.018742791234140715, |
| "grad_norm": 56.0, |
| "kd_loss": 0.0018871173961088061, |
| "learning_rate": 7.111111111111112e-06, |
| "loss": 0.2158, |
| "step": 65, |
| "student_loss": 0.050387192517519, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.020184544405997693, |
| "grad_norm": 28.625, |
| "kd_loss": 0.4643922448158264, |
| "learning_rate": 7.666666666666667e-06, |
| "loss": 0.1818, |
| "step": 70, |
| "student_loss": 0.023501692339777946, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02162629757785467, |
| "grad_norm": 43.0, |
| "kd_loss": 0.010926949791610241, |
| "learning_rate": 8.222222222222222e-06, |
| "loss": 0.3426, |
| "step": 75, |
| "student_loss": 0.0022108303382992744, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02306805074971165, |
| "grad_norm": 13.4375, |
| "kd_loss": 0.004479927476495504, |
| "learning_rate": 8.777777777777778e-06, |
| "loss": 0.3733, |
| "step": 80, |
| "student_loss": 0.0041625467129051685, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.024509803921568627, |
| "grad_norm": 19.5, |
| "kd_loss": 0.003717327257618308, |
| "learning_rate": 9.333333333333334e-06, |
| "loss": 0.2466, |
| "step": 85, |
| "student_loss": 0.0008013551123440266, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.025951557093425604, |
| "grad_norm": 8.4375, |
| "kd_loss": 0.3134710490703583, |
| "learning_rate": 9.88888888888889e-06, |
| "loss": 0.2059, |
| "step": 90, |
| "student_loss": 0.016512297093868256, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.027393310265282585, |
| "grad_norm": 42.0, |
| "kd_loss": 0.0037511922419071198, |
| "learning_rate": 9.99995337991255e-06, |
| "loss": 0.1824, |
| "step": 95, |
| "student_loss": 0.014032241888344288, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.02883506343713956, |
| "grad_norm": 42.75, |
| "kd_loss": 0.0027243352960795164, |
| "learning_rate": 9.999763987297266e-06, |
| "loss": 0.1423, |
| "step": 100, |
| "student_loss": 0.0006586164236068726, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03027681660899654, |
| "grad_norm": 22.75, |
| "kd_loss": 0.007741209119558334, |
| "learning_rate": 9.999428913912876e-06, |
| "loss": 0.1432, |
| "step": 105, |
| "student_loss": 0.006049849092960358, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.031718569780853516, |
| "grad_norm": 8.0, |
| "kd_loss": 0.002043877262622118, |
| "learning_rate": 9.998948169522587e-06, |
| "loss": 0.2012, |
| "step": 110, |
| "student_loss": 0.03380074352025986, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03316032295271049, |
| "grad_norm": 12.9375, |
| "kd_loss": 0.13418245315551758, |
| "learning_rate": 9.998321768134101e-06, |
| "loss": 0.1955, |
| "step": 115, |
| "student_loss": 0.0011221985332667828, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03460207612456748, |
| "grad_norm": 5.5, |
| "kd_loss": 0.00385228986851871, |
| "learning_rate": 9.997549727999212e-06, |
| "loss": 0.1011, |
| "step": 120, |
| "student_loss": 0.0008886058931238949, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.036043829296424454, |
| "grad_norm": 21.75, |
| "kd_loss": 0.8833059072494507, |
| "learning_rate": 9.996632071613257e-06, |
| "loss": 0.1431, |
| "step": 125, |
| "student_loss": 0.037908557802438736, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03748558246828143, |
| "grad_norm": 22.75, |
| "kd_loss": 0.0020795725286006927, |
| "learning_rate": 9.995568825714479e-06, |
| "loss": 0.2122, |
| "step": 130, |
| "student_loss": 0.0007846044027246535, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.03892733564013841, |
| "grad_norm": 27.5, |
| "kd_loss": 0.6685954928398132, |
| "learning_rate": 9.994360021283235e-06, |
| "loss": 0.1744, |
| "step": 135, |
| "student_loss": 0.03987047076225281, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.040369088811995385, |
| "grad_norm": 6.34375, |
| "kd_loss": 0.0027170090470463037, |
| "learning_rate": 9.993005693541098e-06, |
| "loss": 0.1217, |
| "step": 140, |
| "student_loss": 0.0007884247461333871, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04181084198385236, |
| "grad_norm": 15.625, |
| "kd_loss": 0.7305672764778137, |
| "learning_rate": 9.991505881949837e-06, |
| "loss": 0.1177, |
| "step": 145, |
| "student_loss": 0.032465774565935135, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04325259515570934, |
| "grad_norm": 11.9375, |
| "kd_loss": 0.0009196309256367385, |
| "learning_rate": 9.989860630210251e-06, |
| "loss": 0.1145, |
| "step": 150, |
| "student_loss": 0.0006800334667786956, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04469434832756632, |
| "grad_norm": 3.75, |
| "kd_loss": 0.0008867154247127473, |
| "learning_rate": 9.98806998626091e-06, |
| "loss": 0.1703, |
| "step": 155, |
| "student_loss": 0.000513731618411839, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0461361014994233, |
| "grad_norm": 13.5625, |
| "kd_loss": 0.0031801851000636816, |
| "learning_rate": 9.98613400227676e-06, |
| "loss": 0.1881, |
| "step": 160, |
| "student_loss": 0.095796138048172, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.04757785467128028, |
| "grad_norm": 8.75, |
| "kd_loss": 0.009066148661077023, |
| "learning_rate": 9.984052734667586e-06, |
| "loss": 0.147, |
| "step": 165, |
| "student_loss": 0.00036973567330278456, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.049019607843137254, |
| "grad_norm": 19.875, |
| "kd_loss": 0.0007732495432719588, |
| "learning_rate": 9.981826244076387e-06, |
| "loss": 0.1159, |
| "step": 170, |
| "student_loss": 0.0006115734577178955, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05046136101499423, |
| "grad_norm": 17.75, |
| "kd_loss": 0.001440449501387775, |
| "learning_rate": 9.979454595377594e-06, |
| "loss": 0.1439, |
| "step": 175, |
| "student_loss": 0.0018107936484739184, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05190311418685121, |
| "grad_norm": 12.6875, |
| "kd_loss": 0.003575589507818222, |
| "learning_rate": 9.97693785767519e-06, |
| "loss": 0.0823, |
| "step": 180, |
| "student_loss": 0.0005937832174822688, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05334486735870819, |
| "grad_norm": 13.8125, |
| "kd_loss": 0.002183184027671814, |
| "learning_rate": 9.9742761043007e-06, |
| "loss": 0.2085, |
| "step": 185, |
| "student_loss": 0.058680690824985504, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05478662053056517, |
| "grad_norm": 7.9375, |
| "kd_loss": 0.13934114575386047, |
| "learning_rate": 9.971469412811032e-06, |
| "loss": 0.1039, |
| "step": 190, |
| "student_loss": 0.003180636093020439, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.056228373702422146, |
| "grad_norm": 14.5625, |
| "kd_loss": 0.2642575800418854, |
| "learning_rate": 9.968517864986247e-06, |
| "loss": 0.0963, |
| "step": 195, |
| "student_loss": 0.023328492417931557, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.05767012687427912, |
| "grad_norm": 6.1875, |
| "kd_loss": 0.0009822066640481353, |
| "learning_rate": 9.965421546827153e-06, |
| "loss": 0.1046, |
| "step": 200, |
| "student_loss": 0.000694891030434519, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0591118800461361, |
| "grad_norm": 13.5, |
| "kd_loss": 0.0020439866930246353, |
| "learning_rate": 9.962180548552812e-06, |
| "loss": 0.0897, |
| "step": 205, |
| "student_loss": 0.042425841093063354, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06055363321799308, |
| "grad_norm": 5.6875, |
| "kd_loss": 0.06946802884340286, |
| "learning_rate": 9.958794964597906e-06, |
| "loss": 0.0994, |
| "step": 210, |
| "student_loss": 0.0007058735354803503, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.061995386389850055, |
| "grad_norm": 8.75, |
| "kd_loss": 0.0022037022281438112, |
| "learning_rate": 9.955264893609986e-06, |
| "loss": 0.1323, |
| "step": 215, |
| "student_loss": 0.0006759354146197438, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06343713956170703, |
| "grad_norm": 23.5, |
| "kd_loss": 0.11546128988265991, |
| "learning_rate": 9.951590438446597e-06, |
| "loss": 0.108, |
| "step": 220, |
| "student_loss": 0.04613906517624855, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06487889273356401, |
| "grad_norm": 11.0, |
| "kd_loss": 0.006272319238632917, |
| "learning_rate": 9.947771706172285e-06, |
| "loss": 0.1205, |
| "step": 225, |
| "student_loss": 0.03350772336125374, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06632064590542099, |
| "grad_norm": 24.25, |
| "kd_loss": 0.26141810417175293, |
| "learning_rate": 9.943808808055472e-06, |
| "loss": 0.0788, |
| "step": 230, |
| "student_loss": 0.003604925936087966, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06776239907727798, |
| "grad_norm": 15.6875, |
| "kd_loss": 0.6798978447914124, |
| "learning_rate": 9.93970185956522e-06, |
| "loss": 0.0869, |
| "step": 235, |
| "student_loss": 0.015447350218892097, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.06920415224913495, |
| "grad_norm": 1.71875, |
| "kd_loss": 0.003386611584573984, |
| "learning_rate": 9.93545098036786e-06, |
| "loss": 0.1246, |
| "step": 240, |
| "student_loss": 0.0006252449820749462, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07064590542099193, |
| "grad_norm": 21.25, |
| "kd_loss": 0.2356187403202057, |
| "learning_rate": 9.931056294323507e-06, |
| "loss": 0.146, |
| "step": 245, |
| "student_loss": 0.0012863262090831995, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07208765859284891, |
| "grad_norm": 12.9375, |
| "kd_loss": 0.0037490571849048138, |
| "learning_rate": 9.926517929482454e-06, |
| "loss": 0.1485, |
| "step": 250, |
| "student_loss": 0.00040524674113839865, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07352941176470588, |
| "grad_norm": 21.125, |
| "kd_loss": 0.0018789542373269796, |
| "learning_rate": 9.921836018081438e-06, |
| "loss": 0.056, |
| "step": 255, |
| "student_loss": 0.0003012118686456233, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07497116493656286, |
| "grad_norm": 10.25, |
| "kd_loss": 0.008267119526863098, |
| "learning_rate": 9.917010696539793e-06, |
| "loss": 0.2125, |
| "step": 260, |
| "student_loss": 0.04526061564683914, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07641291810841984, |
| "grad_norm": 4.8125, |
| "kd_loss": 0.0011836340418085456, |
| "learning_rate": 9.912042105455462e-06, |
| "loss": 0.1144, |
| "step": 265, |
| "student_loss": 0.00038986533763818443, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07785467128027682, |
| "grad_norm": 12.875, |
| "kd_loss": 0.0008742965874262154, |
| "learning_rate": 9.906930389600917e-06, |
| "loss": 0.1514, |
| "step": 270, |
| "student_loss": 0.0005801369552500546, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.07929642445213379, |
| "grad_norm": 17.75, |
| "kd_loss": 0.0008876683423295617, |
| "learning_rate": 9.901675697918925e-06, |
| "loss": 0.1775, |
| "step": 275, |
| "student_loss": 0.053293921053409576, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08073817762399077, |
| "grad_norm": 26.75, |
| "kd_loss": 0.002074585296213627, |
| "learning_rate": 9.896278183518216e-06, |
| "loss": 0.1494, |
| "step": 280, |
| "student_loss": 0.004262140020728111, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08217993079584775, |
| "grad_norm": 7.65625, |
| "kd_loss": 0.0025187216233462095, |
| "learning_rate": 9.890738003669029e-06, |
| "loss": 0.1171, |
| "step": 285, |
| "student_loss": 0.0007169336313381791, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08362168396770472, |
| "grad_norm": 21.375, |
| "kd_loss": 0.007369345054030418, |
| "learning_rate": 9.885055319798514e-06, |
| "loss": 0.1821, |
| "step": 290, |
| "student_loss": 0.0005684102652594447, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0850634371395617, |
| "grad_norm": 16.0, |
| "kd_loss": 0.006104703992605209, |
| "learning_rate": 9.879230297486034e-06, |
| "loss": 0.1704, |
| "step": 295, |
| "student_loss": 0.0010396749712526798, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08650519031141868, |
| "grad_norm": 23.375, |
| "kd_loss": 0.08903778344392776, |
| "learning_rate": 9.873263106458353e-06, |
| "loss": 0.1398, |
| "step": 300, |
| "student_loss": 0.014311299659311771, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08794694348327567, |
| "grad_norm": 21.125, |
| "kd_loss": 0.003498164936900139, |
| "learning_rate": 9.86715392058467e-06, |
| "loss": 0.1304, |
| "step": 305, |
| "student_loss": 0.018731793388724327, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.08938869665513265, |
| "grad_norm": 22.25, |
| "kd_loss": 0.0002319709019502625, |
| "learning_rate": 9.860902917871566e-06, |
| "loss": 0.1595, |
| "step": 310, |
| "student_loss": 0.0005537475226446986, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09083044982698962, |
| "grad_norm": 28.125, |
| "kd_loss": 0.012743989005684853, |
| "learning_rate": 9.854510280457819e-06, |
| "loss": 0.1062, |
| "step": 315, |
| "student_loss": 0.012689951807260513, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.0922722029988466, |
| "grad_norm": 18.625, |
| "kd_loss": 0.0014882351970300078, |
| "learning_rate": 9.847976194609083e-06, |
| "loss": 0.1479, |
| "step": 320, |
| "student_loss": 0.0005163275054655969, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09371395617070358, |
| "grad_norm": 22.625, |
| "kd_loss": 0.3226740062236786, |
| "learning_rate": 9.841300850712479e-06, |
| "loss": 0.1206, |
| "step": 325, |
| "student_loss": 0.005041609983891249, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09515570934256055, |
| "grad_norm": 10.125, |
| "kd_loss": 0.4770796298980713, |
| "learning_rate": 9.834484443271035e-06, |
| "loss": 0.0984, |
| "step": 330, |
| "student_loss": 0.015074173919856548, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09659746251441753, |
| "grad_norm": 13.25, |
| "kd_loss": 0.000536711944732815, |
| "learning_rate": 9.827527170898019e-06, |
| "loss": 0.0722, |
| "step": 335, |
| "student_loss": 0.0006227240664884448, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09803921568627451, |
| "grad_norm": 26.0, |
| "kd_loss": 0.12674546241760254, |
| "learning_rate": 9.820429236311158e-06, |
| "loss": 0.1458, |
| "step": 340, |
| "student_loss": 0.012428171001374722, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.09948096885813149, |
| "grad_norm": 8.375, |
| "kd_loss": 0.0027160593308508396, |
| "learning_rate": 9.81319084632673e-06, |
| "loss": 0.1197, |
| "step": 345, |
| "student_loss": 0.01698848232626915, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10092272202998846, |
| "grad_norm": 25.5, |
| "kd_loss": 0.2688156068325043, |
| "learning_rate": 9.805812211853534e-06, |
| "loss": 0.0684, |
| "step": 350, |
| "student_loss": 0.006060065235942602, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10236447520184544, |
| "grad_norm": 26.0, |
| "kd_loss": 0.0023566237650811672, |
| "learning_rate": 9.798293547886748e-06, |
| "loss": 0.1159, |
| "step": 355, |
| "student_loss": 0.0004638656973838806, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10380622837370242, |
| "grad_norm": 9.125, |
| "kd_loss": 0.004494955763220787, |
| "learning_rate": 9.790635073501662e-06, |
| "loss": 0.082, |
| "step": 360, |
| "student_loss": 0.0004987778957001865, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1052479815455594, |
| "grad_norm": 10.375, |
| "kd_loss": 0.0029149462934583426, |
| "learning_rate": 9.782837011847294e-06, |
| "loss": 0.1926, |
| "step": 365, |
| "student_loss": 0.043784309178590775, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10668973471741638, |
| "grad_norm": 24.625, |
| "kd_loss": 0.07756392657756805, |
| "learning_rate": 9.774899590139897e-06, |
| "loss": 0.0849, |
| "step": 370, |
| "student_loss": 0.01941766031086445, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10813148788927336, |
| "grad_norm": 13.125, |
| "kd_loss": 0.00261987978592515, |
| "learning_rate": 9.766823039656328e-06, |
| "loss": 0.1517, |
| "step": 375, |
| "student_loss": 0.03552848473191261, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.10957324106113034, |
| "grad_norm": 6.28125, |
| "kd_loss": 0.0021297987550497055, |
| "learning_rate": 9.758607595727309e-06, |
| "loss": 0.1632, |
| "step": 380, |
| "student_loss": 0.0005093910149298608, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11101499423298732, |
| "grad_norm": 28.125, |
| "kd_loss": 0.012473474256694317, |
| "learning_rate": 9.75025349773058e-06, |
| "loss": 0.1271, |
| "step": 385, |
| "student_loss": 0.00041385076474398375, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11245674740484429, |
| "grad_norm": 24.375, |
| "kd_loss": 0.0008915087673813105, |
| "learning_rate": 9.741760989083914e-06, |
| "loss": 0.0936, |
| "step": 390, |
| "student_loss": 0.00028736997046507895, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11389850057670127, |
| "grad_norm": 12.875, |
| "kd_loss": 0.007614473812282085, |
| "learning_rate": 9.73313031723803e-06, |
| "loss": 0.0901, |
| "step": 395, |
| "student_loss": 0.0005106693133711815, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11534025374855825, |
| "grad_norm": 11.25, |
| "kd_loss": 0.00347925559617579, |
| "learning_rate": 9.724361733669383e-06, |
| "loss": 0.0546, |
| "step": 400, |
| "student_loss": 0.0008912020130082965, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11678200692041522, |
| "grad_norm": 8.0625, |
| "kd_loss": 0.0011954058427363634, |
| "learning_rate": 9.715455493872827e-06, |
| "loss": 0.1324, |
| "step": 405, |
| "student_loss": 0.004991908557713032, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1182237600922722, |
| "grad_norm": 9.1875, |
| "kd_loss": 0.0025417145807296038, |
| "learning_rate": 9.70641185735419e-06, |
| "loss": 0.1483, |
| "step": 410, |
| "student_loss": 0.01691470481455326, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.11966551326412918, |
| "grad_norm": 13.3125, |
| "kd_loss": 0.003752017393708229, |
| "learning_rate": 9.697231087622691e-06, |
| "loss": 0.1511, |
| "step": 415, |
| "student_loss": 0.0010631532641127706, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12110726643598616, |
| "grad_norm": 10.9375, |
| "kd_loss": 0.2554698586463928, |
| "learning_rate": 9.687913452183277e-06, |
| "loss": 0.1095, |
| "step": 420, |
| "student_loss": 0.06524818390607834, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12254901960784313, |
| "grad_norm": 9.125, |
| "kd_loss": 0.0005501045961864293, |
| "learning_rate": 9.678459222528825e-06, |
| "loss": 0.0577, |
| "step": 425, |
| "student_loss": 0.0013150365557521582, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12399077277970011, |
| "grad_norm": 10.0625, |
| "kd_loss": 0.00420792680233717, |
| "learning_rate": 9.668868674132224e-06, |
| "loss": 0.1381, |
| "step": 430, |
| "student_loss": 0.05358680710196495, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1254325259515571, |
| "grad_norm": 14.625, |
| "kd_loss": 0.0016143789980560541, |
| "learning_rate": 9.65914208643836e-06, |
| "loss": 0.0654, |
| "step": 435, |
| "student_loss": 0.000772066181525588, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12687427912341406, |
| "grad_norm": 7.15625, |
| "kd_loss": 0.0023973328061401844, |
| "learning_rate": 9.649279742855964e-06, |
| "loss": 0.1371, |
| "step": 440, |
| "student_loss": 0.025605041533708572, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12831603229527105, |
| "grad_norm": 23.0, |
| "kd_loss": 0.0013417273294180632, |
| "learning_rate": 9.639281930749363e-06, |
| "loss": 0.0814, |
| "step": 445, |
| "student_loss": 0.03157199174165726, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.12975778546712802, |
| "grad_norm": 13.25, |
| "kd_loss": 0.0016620807582512498, |
| "learning_rate": 9.629148941430096e-06, |
| "loss": 0.1442, |
| "step": 450, |
| "student_loss": 0.0006721625686623156, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.131199538638985, |
| "grad_norm": 28.0, |
| "kd_loss": 0.0016946454998105764, |
| "learning_rate": 9.618881070148436e-06, |
| "loss": 0.1405, |
| "step": 455, |
| "student_loss": 0.0009172292775474489, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13264129181084197, |
| "grad_norm": 10.125, |
| "kd_loss": 0.027760690078139305, |
| "learning_rate": 9.608478616084784e-06, |
| "loss": 0.0505, |
| "step": 460, |
| "student_loss": 0.0005534313968382776, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13408304498269896, |
| "grad_norm": 14.25, |
| "kd_loss": 0.008209657855331898, |
| "learning_rate": 9.597941882340947e-06, |
| "loss": 0.1803, |
| "step": 465, |
| "student_loss": 0.000469625141704455, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13552479815455595, |
| "grad_norm": 13.4375, |
| "kd_loss": 0.0006104095955379307, |
| "learning_rate": 9.587271175931313e-06, |
| "loss": 0.1148, |
| "step": 470, |
| "student_loss": 0.00037698415690101683, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13696655132641292, |
| "grad_norm": 6.5625, |
| "kd_loss": 0.012027841061353683, |
| "learning_rate": 9.5764668077739e-06, |
| "loss": 0.0626, |
| "step": 475, |
| "student_loss": 0.0005685262149199843, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1384083044982699, |
| "grad_norm": 9.0625, |
| "kd_loss": 0.0022413916885852814, |
| "learning_rate": 9.5655290926813e-06, |
| "loss": 0.099, |
| "step": 480, |
| "student_loss": 0.012239435687661171, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.13985005767012687, |
| "grad_norm": 25.75, |
| "kd_loss": 0.0016755703836679459, |
| "learning_rate": 9.554458349351507e-06, |
| "loss": 0.0891, |
| "step": 485, |
| "student_loss": 0.027401834726333618, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14129181084198386, |
| "grad_norm": 29.875, |
| "kd_loss": 0.46647658944129944, |
| "learning_rate": 9.54325490035863e-06, |
| "loss": 0.1581, |
| "step": 490, |
| "student_loss": 0.0034644226543605328, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14273356401384082, |
| "grad_norm": 12.0, |
| "kd_loss": 0.10091336816549301, |
| "learning_rate": 9.531919072143488e-06, |
| "loss": 0.1283, |
| "step": 495, |
| "student_loss": 0.0010429032845422626, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14417531718569782, |
| "grad_norm": 18.125, |
| "kd_loss": 0.25293171405792236, |
| "learning_rate": 9.520451195004107e-06, |
| "loss": 0.0874, |
| "step": 500, |
| "student_loss": 0.0008772966684773564, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14561707035755478, |
| "grad_norm": 1.453125, |
| "kd_loss": 0.012721131555736065, |
| "learning_rate": 9.508851603086094e-06, |
| "loss": 0.124, |
| "step": 505, |
| "student_loss": 0.0005022344412282109, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14705882352941177, |
| "grad_norm": 14.375, |
| "kd_loss": 0.000858868530485779, |
| "learning_rate": 9.497120634372894e-06, |
| "loss": 0.1335, |
| "step": 510, |
| "student_loss": 0.00032862432999536395, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14850057670126873, |
| "grad_norm": 9.6875, |
| "kd_loss": 0.0011465316638350487, |
| "learning_rate": 9.48525863067595e-06, |
| "loss": 0.0708, |
| "step": 515, |
| "student_loss": 0.008267870172858238, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.14994232987312572, |
| "grad_norm": 10.625, |
| "kd_loss": 0.002247657161206007, |
| "learning_rate": 9.473265937624748e-06, |
| "loss": 0.0503, |
| "step": 520, |
| "student_loss": 0.041273463517427444, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1513840830449827, |
| "grad_norm": 29.25, |
| "kd_loss": 0.0018336102366447449, |
| "learning_rate": 9.461142904656722e-06, |
| "loss": 0.1143, |
| "step": 525, |
| "student_loss": 0.00042416868382133543, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.15282583621683968, |
| "grad_norm": 1.765625, |
| "kd_loss": 0.0032041582744568586, |
| "learning_rate": 9.448889885007105e-06, |
| "loss": 0.0643, |
| "step": 530, |
| "student_loss": 0.00041470438009127975, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.15426758938869667, |
| "grad_norm": 6.125, |
| "kd_loss": 0.0082230931147933, |
| "learning_rate": 9.436507235698613e-06, |
| "loss": 0.0916, |
| "step": 535, |
| "student_loss": 0.0004584086127579212, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.15570934256055363, |
| "grad_norm": 8.6875, |
| "kd_loss": 0.0010036047315225005, |
| "learning_rate": 9.423995317531049e-06, |
| "loss": 0.128, |
| "step": 540, |
| "student_loss": 0.0008725057705305517, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.15715109573241062, |
| "grad_norm": 4.6875, |
| "kd_loss": 0.004125351086258888, |
| "learning_rate": 9.4113544950708e-06, |
| "loss": 0.0659, |
| "step": 545, |
| "student_loss": 0.0022284940350800753, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.15859284890426759, |
| "grad_norm": 5.78125, |
| "kd_loss": 0.0035471171140670776, |
| "learning_rate": 9.398585136640195e-06, |
| "loss": 0.0526, |
| "step": 550, |
| "student_loss": 0.0001974322076421231, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16003460207612458, |
| "grad_norm": 15.0, |
| "kd_loss": 0.6907945871353149, |
| "learning_rate": 9.38568761430679e-06, |
| "loss": 0.1251, |
| "step": 555, |
| "student_loss": 0.025625621899962425, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16147635524798154, |
| "grad_norm": 14.5625, |
| "kd_loss": 0.9356266856193542, |
| "learning_rate": 9.372662303872514e-06, |
| "loss": 0.0997, |
| "step": 560, |
| "student_loss": 0.0429786816239357, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16291810841983853, |
| "grad_norm": 7.59375, |
| "kd_loss": 0.0022694910876452923, |
| "learning_rate": 9.359509584862735e-06, |
| "loss": 0.1102, |
| "step": 565, |
| "student_loss": 0.038102827966213226, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1643598615916955, |
| "grad_norm": 24.5, |
| "kd_loss": 0.002567964605987072, |
| "learning_rate": 9.346229840515186e-06, |
| "loss": 0.1791, |
| "step": 570, |
| "student_loss": 0.010701650753617287, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16580161476355249, |
| "grad_norm": 47.75, |
| "kd_loss": 0.0009870093781501055, |
| "learning_rate": 9.332823457768805e-06, |
| "loss": 0.0795, |
| "step": 575, |
| "student_loss": 0.0004428588435985148, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16724336793540945, |
| "grad_norm": 5.625, |
| "kd_loss": 0.018184596672654152, |
| "learning_rate": 9.31929082725246e-06, |
| "loss": 0.1173, |
| "step": 580, |
| "student_loss": 0.0014106971211731434, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.16868512110726644, |
| "grad_norm": 5.46875, |
| "kd_loss": 0.10436643660068512, |
| "learning_rate": 9.305632343273567e-06, |
| "loss": 0.0741, |
| "step": 585, |
| "student_loss": 0.0011741393245756626, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1701268742791234, |
| "grad_norm": 21.375, |
| "kd_loss": 0.008095277473330498, |
| "learning_rate": 9.291848403806604e-06, |
| "loss": 0.1016, |
| "step": 590, |
| "student_loss": 0.00045060788397677243, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1715686274509804, |
| "grad_norm": 8.0625, |
| "kd_loss": 0.004174170084297657, |
| "learning_rate": 9.277939410481507e-06, |
| "loss": 0.0639, |
| "step": 595, |
| "student_loss": 0.00036720611387863755, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.17301038062283736, |
| "grad_norm": 31.375, |
| "kd_loss": 0.0980350524187088, |
| "learning_rate": 9.263905768571977e-06, |
| "loss": 0.1305, |
| "step": 600, |
| "student_loss": 0.039228975772857666, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.17445213379469435, |
| "grad_norm": 7.84375, |
| "kd_loss": 0.0019628205336630344, |
| "learning_rate": 9.249747886983664e-06, |
| "loss": 0.1027, |
| "step": 605, |
| "student_loss": 0.10567551851272583, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.17589388696655134, |
| "grad_norm": 10.8125, |
| "kd_loss": 0.15832534432411194, |
| "learning_rate": 9.235466178242255e-06, |
| "loss": 0.0776, |
| "step": 610, |
| "student_loss": 0.008307134732604027, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1773356401384083, |
| "grad_norm": 7.4375, |
| "kd_loss": 0.0004231311031617224, |
| "learning_rate": 9.221061058481456e-06, |
| "loss": 0.0726, |
| "step": 615, |
| "student_loss": 0.00020960887195542455, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1787773933102653, |
| "grad_norm": 2.9375, |
| "kd_loss": 0.000930448470171541, |
| "learning_rate": 9.206532947430865e-06, |
| "loss": 0.0953, |
| "step": 620, |
| "student_loss": 0.012740620411932468, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.18021914648212226, |
| "grad_norm": 10.9375, |
| "kd_loss": 0.00044887661351822317, |
| "learning_rate": 9.191882268403743e-06, |
| "loss": 0.0502, |
| "step": 625, |
| "student_loss": 0.006197873968631029, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.18166089965397925, |
| "grad_norm": 39.75, |
| "kd_loss": 2.8654651641845703, |
| "learning_rate": 9.177109448284673e-06, |
| "loss": 0.1582, |
| "step": 630, |
| "student_loss": 0.05481014773249626, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1831026528258362, |
| "grad_norm": 5.15625, |
| "kd_loss": 0.0009848499903455377, |
| "learning_rate": 9.162214917517139e-06, |
| "loss": 0.1122, |
| "step": 635, |
| "student_loss": 0.0012388643808662891, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1845444059976932, |
| "grad_norm": 8.8125, |
| "kd_loss": 0.0020470728632062674, |
| "learning_rate": 9.14719911009096e-06, |
| "loss": 0.0638, |
| "step": 640, |
| "student_loss": 0.00041219103150069714, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.18598615916955016, |
| "grad_norm": 8.8125, |
| "kd_loss": 0.003420033724978566, |
| "learning_rate": 9.132062463529666e-06, |
| "loss": 0.0673, |
| "step": 645, |
| "student_loss": 0.0004483861557673663, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.18742791234140715, |
| "grad_norm": 16.375, |
| "kd_loss": 0.0017254488775506616, |
| "learning_rate": 9.11680541887774e-06, |
| "loss": 0.088, |
| "step": 650, |
| "student_loss": 0.0007219853578135371, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.18886966551326412, |
| "grad_norm": 12.4375, |
| "kd_loss": 0.004693194758147001, |
| "learning_rate": 9.101428420687759e-06, |
| "loss": 0.0916, |
| "step": 655, |
| "student_loss": 0.0004059349885210395, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.1903114186851211, |
| "grad_norm": 17.125, |
| "kd_loss": 0.0011971323983743787, |
| "learning_rate": 9.085931917007464e-06, |
| "loss": 0.088, |
| "step": 660, |
| "student_loss": 0.0004724525206256658, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.19175317185697807, |
| "grad_norm": 7.90625, |
| "kd_loss": 0.01377791166305542, |
| "learning_rate": 9.070316359366681e-06, |
| "loss": 0.0904, |
| "step": 665, |
| "student_loss": 0.00045861524995416403, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.19319492502883506, |
| "grad_norm": 36.5, |
| "kd_loss": 1.8900201320648193, |
| "learning_rate": 9.054582202764175e-06, |
| "loss": 0.0963, |
| "step": 670, |
| "student_loss": 0.07230479270219803, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.19463667820069205, |
| "grad_norm": 4.0, |
| "kd_loss": 0.0026326398365199566, |
| "learning_rate": 9.038729905654393e-06, |
| "loss": 0.1095, |
| "step": 675, |
| "student_loss": 0.0004619761893991381, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.19607843137254902, |
| "grad_norm": 10.5, |
| "kd_loss": 0.18342895805835724, |
| "learning_rate": 9.022759929934109e-06, |
| "loss": 0.0867, |
| "step": 680, |
| "student_loss": 0.07520744204521179, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.197520184544406, |
| "grad_norm": 18.5, |
| "kd_loss": 0.0016399289015680552, |
| "learning_rate": 9.006672740928952e-06, |
| "loss": 0.1277, |
| "step": 685, |
| "student_loss": 0.041107166558504105, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.19896193771626297, |
| "grad_norm": 22.5, |
| "kd_loss": 0.8545010089874268, |
| "learning_rate": 8.990468807379861e-06, |
| "loss": 0.0946, |
| "step": 690, |
| "student_loss": 0.014823048375546932, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20040369088811996, |
| "grad_norm": 6.1875, |
| "kd_loss": 0.0027170374523848295, |
| "learning_rate": 8.974148601429426e-06, |
| "loss": 0.1075, |
| "step": 695, |
| "student_loss": 0.005867379251867533, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20184544405997693, |
| "grad_norm": 13.6875, |
| "kd_loss": 0.005027920473366976, |
| "learning_rate": 8.957712598608123e-06, |
| "loss": 0.1523, |
| "step": 700, |
| "student_loss": 0.0006474690162576735, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20328719723183392, |
| "grad_norm": 10.5625, |
| "kd_loss": 0.018207697197794914, |
| "learning_rate": 8.941161277820462e-06, |
| "loss": 0.0628, |
| "step": 705, |
| "student_loss": 0.0015416023088619113, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20472895040369088, |
| "grad_norm": 5.15625, |
| "kd_loss": 0.31215962767601013, |
| "learning_rate": 8.924495121331038e-06, |
| "loss": 0.099, |
| "step": 710, |
| "student_loss": 0.007331720553338528, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20617070357554787, |
| "grad_norm": 15.875, |
| "kd_loss": 0.0036377429496496916, |
| "learning_rate": 8.907714614750473e-06, |
| "loss": 0.0935, |
| "step": 715, |
| "student_loss": 0.008250439539551735, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20761245674740483, |
| "grad_norm": 3.171875, |
| "kd_loss": 0.04166782274842262, |
| "learning_rate": 8.890820247021263e-06, |
| "loss": 0.1256, |
| "step": 720, |
| "student_loss": 0.0010231786873191595, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.20905420991926182, |
| "grad_norm": 10.8125, |
| "kd_loss": 0.002339277183637023, |
| "learning_rate": 8.873812510403545e-06, |
| "loss": 0.0612, |
| "step": 725, |
| "student_loss": 0.00030273603624664247, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2104959630911188, |
| "grad_norm": 20.75, |
| "kd_loss": 0.07314825803041458, |
| "learning_rate": 8.85669190046074e-06, |
| "loss": 0.1104, |
| "step": 730, |
| "student_loss": 0.021018339321017265, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.21193771626297578, |
| "grad_norm": 5.53125, |
| "kd_loss": 0.0017747314414009452, |
| "learning_rate": 8.839458916045117e-06, |
| "loss": 0.0881, |
| "step": 735, |
| "student_loss": 0.00029675825498998165, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.21337946943483277, |
| "grad_norm": 4.6875, |
| "kd_loss": 0.001755792647600174, |
| "learning_rate": 8.822114059283267e-06, |
| "loss": 0.0907, |
| "step": 740, |
| "student_loss": 0.004126196261495352, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.21482122260668973, |
| "grad_norm": 20.625, |
| "kd_loss": 0.001788739929907024, |
| "learning_rate": 8.804657835561456e-06, |
| "loss": 0.1042, |
| "step": 745, |
| "student_loss": 0.0402885265648365, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.21626297577854672, |
| "grad_norm": 7.28125, |
| "kd_loss": 0.7170178294181824, |
| "learning_rate": 8.787090753510917e-06, |
| "loss": 0.0756, |
| "step": 750, |
| "student_loss": 0.022320456802845, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2177047289504037, |
| "grad_norm": 12.875, |
| "kd_loss": 0.0030846295412629843, |
| "learning_rate": 8.76941332499301e-06, |
| "loss": 0.1132, |
| "step": 755, |
| "student_loss": 0.03327662870287895, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.21914648212226068, |
| "grad_norm": 12.4375, |
| "kd_loss": 0.0014672816032543778, |
| "learning_rate": 8.751626065084328e-06, |
| "loss": 0.1147, |
| "step": 760, |
| "student_loss": 0.00921363290399313, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.22058823529411764, |
| "grad_norm": 14.125, |
| "kd_loss": 0.005435326602309942, |
| "learning_rate": 8.733729492061675e-06, |
| "loss": 0.0476, |
| "step": 765, |
| "student_loss": 0.004831620957702398, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.22202998846597463, |
| "grad_norm": 4.0, |
| "kd_loss": 0.03225019574165344, |
| "learning_rate": 8.715724127386971e-06, |
| "loss": 0.103, |
| "step": 770, |
| "student_loss": 0.0014119932893663645, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2234717416378316, |
| "grad_norm": 3.390625, |
| "kd_loss": 0.10136047750711441, |
| "learning_rate": 8.697610495692055e-06, |
| "loss": 0.0591, |
| "step": 775, |
| "student_loss": 0.0035214233212172985, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.22491349480968859, |
| "grad_norm": 21.25, |
| "kd_loss": 0.0011202533496543765, |
| "learning_rate": 8.679389124763396e-06, |
| "loss": 0.1158, |
| "step": 780, |
| "student_loss": 0.0014409847790375352, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.22635524798154555, |
| "grad_norm": 9.0, |
| "kd_loss": 0.000793208135291934, |
| "learning_rate": 8.661060545526724e-06, |
| "loss": 0.0603, |
| "step": 785, |
| "student_loss": 0.0002954969822894782, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.22779700115340254, |
| "grad_norm": 7.5, |
| "kd_loss": 0.0026095786597579718, |
| "learning_rate": 8.64262529203155e-06, |
| "loss": 0.0422, |
| "step": 790, |
| "student_loss": 0.002638905541971326, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2292387543252595, |
| "grad_norm": 3.28125, |
| "kd_loss": 0.002780693117529154, |
| "learning_rate": 8.624083901435613e-06, |
| "loss": 0.0766, |
| "step": 795, |
| "student_loss": 0.000536055420525372, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2306805074971165, |
| "grad_norm": 50.0, |
| "kd_loss": 0.13559553027153015, |
| "learning_rate": 8.60543691398922e-06, |
| "loss": 0.1005, |
| "step": 800, |
| "student_loss": 0.06979604810476303, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.23212226066897348, |
| "grad_norm": 35.75, |
| "kd_loss": 0.00036787468707188964, |
| "learning_rate": 8.586684873019513e-06, |
| "loss": 0.1055, |
| "step": 805, |
| "student_loss": 0.0003365253796800971, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.23356401384083045, |
| "grad_norm": 19.875, |
| "kd_loss": 0.00150456337723881, |
| "learning_rate": 8.567828324914633e-06, |
| "loss": 0.0539, |
| "step": 810, |
| "student_loss": 0.0004286724142730236, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.23500576701268744, |
| "grad_norm": 7.0625, |
| "kd_loss": 0.007362797390669584, |
| "learning_rate": 8.548867819107803e-06, |
| "loss": 0.0949, |
| "step": 815, |
| "student_loss": 0.001179418177343905, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2364475201845444, |
| "grad_norm": 6.6875, |
| "kd_loss": 0.001827954431064427, |
| "learning_rate": 8.52980390806131e-06, |
| "loss": 0.0525, |
| "step": 820, |
| "student_loss": 0.004497961141169071, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2378892733564014, |
| "grad_norm": 10.1875, |
| "kd_loss": 0.0011187309864908457, |
| "learning_rate": 8.51063714725042e-06, |
| "loss": 0.1075, |
| "step": 825, |
| "student_loss": 0.000924907042644918, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.23933102652825836, |
| "grad_norm": 12.0625, |
| "kd_loss": 0.12930721044540405, |
| "learning_rate": 8.49136809514719e-06, |
| "loss": 0.101, |
| "step": 830, |
| "student_loss": 0.012681624852120876, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.24077277970011535, |
| "grad_norm": 22.125, |
| "kd_loss": 0.0007281019352376461, |
| "learning_rate": 8.471997313204183e-06, |
| "loss": 0.0645, |
| "step": 835, |
| "student_loss": 0.01504411082714796, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2422145328719723, |
| "grad_norm": 12.0, |
| "kd_loss": 0.0006879305001348257, |
| "learning_rate": 8.452525365838123e-06, |
| "loss": 0.0615, |
| "step": 840, |
| "student_loss": 0.0004384240601211786, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2436562860438293, |
| "grad_norm": 11.125, |
| "kd_loss": 0.001608243677765131, |
| "learning_rate": 8.432952820413447e-06, |
| "loss": 0.0914, |
| "step": 845, |
| "student_loss": 0.0005417144275270402, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.24509803921568626, |
| "grad_norm": 33.25, |
| "kd_loss": 0.1373538076877594, |
| "learning_rate": 8.41328024722577e-06, |
| "loss": 0.1133, |
| "step": 850, |
| "student_loss": 0.01142928283661604, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.24653979238754326, |
| "grad_norm": 10.3125, |
| "kd_loss": 0.0016314623644575477, |
| "learning_rate": 8.393508219485263e-06, |
| "loss": 0.1024, |
| "step": 855, |
| "student_loss": 0.00022962426010053605, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.24798154555940022, |
| "grad_norm": 5.90625, |
| "kd_loss": 0.0019791526719927788, |
| "learning_rate": 8.373637313299966e-06, |
| "loss": 0.1224, |
| "step": 860, |
| "student_loss": 0.0005864988779649138, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2494232987312572, |
| "grad_norm": 9.875, |
| "kd_loss": 0.000920691411010921, |
| "learning_rate": 8.353668107658984e-06, |
| "loss": 0.0633, |
| "step": 865, |
| "student_loss": 0.0002618279540911317, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2508650519031142, |
| "grad_norm": 75.0, |
| "kd_loss": 0.41142162680625916, |
| "learning_rate": 8.333601184415632e-06, |
| "loss": 0.0753, |
| "step": 870, |
| "student_loss": 0.008613992482423782, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.25230680507497116, |
| "grad_norm": 24.125, |
| "kd_loss": 0.0006664465181529522, |
| "learning_rate": 8.31343712827047e-06, |
| "loss": 0.0742, |
| "step": 875, |
| "student_loss": 0.00016053368744906038, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2537485582468281, |
| "grad_norm": 5.84375, |
| "kd_loss": 0.006322797387838364, |
| "learning_rate": 8.293176526754274e-06, |
| "loss": 0.1238, |
| "step": 880, |
| "student_loss": 0.000416541617596522, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.25519031141868515, |
| "grad_norm": 12.75, |
| "kd_loss": 0.19639839231967926, |
| "learning_rate": 8.27281997021091e-06, |
| "loss": 0.0568, |
| "step": 885, |
| "student_loss": 0.007913337089121342, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2566320645905421, |
| "grad_norm": 7.15625, |
| "kd_loss": 0.001160801388323307, |
| "learning_rate": 8.25236805178014e-06, |
| "loss": 0.0727, |
| "step": 890, |
| "student_loss": 0.0003532449482008815, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.25807381776239907, |
| "grad_norm": 12.6875, |
| "kd_loss": 0.19797247648239136, |
| "learning_rate": 8.231821367380335e-06, |
| "loss": 0.128, |
| "step": 895, |
| "student_loss": 0.003061032621189952, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.25951557093425603, |
| "grad_norm": 8.0, |
| "kd_loss": 0.007570919580757618, |
| "learning_rate": 8.211180515691108e-06, |
| "loss": 0.0792, |
| "step": 900, |
| "student_loss": 0.033695392310619354, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.26095732410611305, |
| "grad_norm": 11.3125, |
| "kd_loss": 0.38567981123924255, |
| "learning_rate": 8.190446098135883e-06, |
| "loss": 0.0922, |
| "step": 905, |
| "student_loss": 0.028850680217146873, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.26239907727797, |
| "grad_norm": 10.9375, |
| "kd_loss": 0.0007043812656775117, |
| "learning_rate": 8.16961871886435e-06, |
| "loss": 0.0954, |
| "step": 910, |
| "student_loss": 0.0002878632803913206, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.263840830449827, |
| "grad_norm": 8.1875, |
| "kd_loss": 0.14511993527412415, |
| "learning_rate": 8.148698984734887e-06, |
| "loss": 0.0477, |
| "step": 915, |
| "student_loss": 0.0008710387046448886, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.26528258362168394, |
| "grad_norm": 8.5, |
| "kd_loss": 0.0384184792637825, |
| "learning_rate": 8.127687505296852e-06, |
| "loss": 0.0797, |
| "step": 920, |
| "student_loss": 0.016883913427591324, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.26672433679354096, |
| "grad_norm": 4.4375, |
| "kd_loss": 0.00027013575891032815, |
| "learning_rate": 8.106584892772844e-06, |
| "loss": 0.0756, |
| "step": 925, |
| "student_loss": 0.0002693244314286858, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2681660899653979, |
| "grad_norm": 10.0, |
| "kd_loss": 0.0005406018462963402, |
| "learning_rate": 8.08539176204085e-06, |
| "loss": 0.0681, |
| "step": 930, |
| "student_loss": 0.025998031720519066, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2696078431372549, |
| "grad_norm": 7.84375, |
| "kd_loss": 0.07199026644229889, |
| "learning_rate": 8.064108730616336e-06, |
| "loss": 0.0594, |
| "step": 935, |
| "student_loss": 0.0006435631075873971, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2710495963091119, |
| "grad_norm": 6.0625, |
| "kd_loss": 0.006361419800668955, |
| "learning_rate": 8.042736418634252e-06, |
| "loss": 0.0734, |
| "step": 940, |
| "student_loss": 0.012877441942691803, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.27249134948096887, |
| "grad_norm": 25.25, |
| "kd_loss": 0.009065474383533001, |
| "learning_rate": 8.02127544883096e-06, |
| "loss": 0.0763, |
| "step": 945, |
| "student_loss": 0.007113016210496426, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.27393310265282583, |
| "grad_norm": 1.890625, |
| "kd_loss": 0.0012218133779242635, |
| "learning_rate": 7.999726446526095e-06, |
| "loss": 0.0542, |
| "step": 950, |
| "student_loss": 0.00028497542371042073, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2753748558246828, |
| "grad_norm": 8.8125, |
| "kd_loss": 0.0014024431584402919, |
| "learning_rate": 7.978090039604342e-06, |
| "loss": 0.0548, |
| "step": 955, |
| "student_loss": 0.0005670087411999702, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2768166089965398, |
| "grad_norm": 26.5, |
| "kd_loss": 0.006449506618082523, |
| "learning_rate": 7.956366858497138e-06, |
| "loss": 0.0734, |
| "step": 960, |
| "student_loss": 0.0003615773457568139, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2782583621683968, |
| "grad_norm": 16.625, |
| "kd_loss": 0.001059827976860106, |
| "learning_rate": 7.934557536164304e-06, |
| "loss": 0.1002, |
| "step": 965, |
| "student_loss": 0.012110190466046333, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.27970011534025374, |
| "grad_norm": 23.375, |
| "kd_loss": 0.0003945316420868039, |
| "learning_rate": 7.91266270807561e-06, |
| "loss": 0.0942, |
| "step": 970, |
| "student_loss": 0.00027471804060041904, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2811418685121107, |
| "grad_norm": 12.0, |
| "kd_loss": 0.006996847223490477, |
| "learning_rate": 7.890683012192249e-06, |
| "loss": 0.1016, |
| "step": 975, |
| "student_loss": 0.020073270425200462, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2825836216839677, |
| "grad_norm": 2.34375, |
| "kd_loss": 0.0010718721896409988, |
| "learning_rate": 7.868619088948246e-06, |
| "loss": 0.0668, |
| "step": 980, |
| "student_loss": 0.000354983116267249, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2840253748558247, |
| "grad_norm": 7.9375, |
| "kd_loss": 0.000552264042198658, |
| "learning_rate": 7.846471581231814e-06, |
| "loss": 0.1767, |
| "step": 985, |
| "student_loss": 0.000291457399725914, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.28546712802768165, |
| "grad_norm": 21.375, |
| "kd_loss": 0.0026221515145152807, |
| "learning_rate": 7.824241134366604e-06, |
| "loss": 0.0533, |
| "step": 990, |
| "student_loss": 0.00033598451409488916, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.2869088811995386, |
| "grad_norm": 16.375, |
| "kd_loss": 0.014461357146501541, |
| "learning_rate": 7.801928396092908e-06, |
| "loss": 0.0847, |
| "step": 995, |
| "student_loss": 0.006032541859894991, |
| "teacher_loss": 0.0 |
| }, |
| { |
| "epoch": 0.28835063437139563, |
| "grad_norm": 3.90625, |
| "kd_loss": 0.005895162001252174, |
| "learning_rate": 7.779534016548791e-06, |
| "loss": 0.0785, |
| "step": 1000, |
| "student_loss": 0.005668016150593758, |
| "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 |
| } |
|
|
