| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 1473, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.00679001867255135, | |
| "grad_norm": 1.8175065517425537, | |
| "learning_rate": 6.081081081081082e-06, | |
| "loss": 0.1857, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.0135800373451027, | |
| "grad_norm": 0.9356499910354614, | |
| "learning_rate": 1.2837837837837838e-05, | |
| "loss": 0.1512, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02037005601765405, | |
| "grad_norm": 0.19530220329761505, | |
| "learning_rate": 1.9594594594594595e-05, | |
| "loss": 0.1038, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.0271600746902054, | |
| "grad_norm": 0.1939963847398758, | |
| "learning_rate": 2.635135135135135e-05, | |
| "loss": 0.082, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.03395009336275675, | |
| "grad_norm": 0.25467729568481445, | |
| "learning_rate": 3.310810810810811e-05, | |
| "loss": 0.0825, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.0407401120353081, | |
| "grad_norm": 0.24394609034061432, | |
| "learning_rate": 3.986486486486487e-05, | |
| "loss": 0.0841, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.04753013070785945, | |
| "grad_norm": 0.18351706862449646, | |
| "learning_rate": 4.662162162162162e-05, | |
| "loss": 0.0795, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.0543201493804108, | |
| "grad_norm": 0.20120970904827118, | |
| "learning_rate": 5.337837837837838e-05, | |
| "loss": 0.072, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.06111016805296215, | |
| "grad_norm": 0.30063867568969727, | |
| "learning_rate": 6.013513513513514e-05, | |
| "loss": 0.0752, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.0679001867255135, | |
| "grad_norm": 0.35980555415153503, | |
| "learning_rate": 6.68918918918919e-05, | |
| "loss": 0.0673, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.07469020539806484, | |
| "grad_norm": 0.37400639057159424, | |
| "learning_rate": 7.364864864864865e-05, | |
| "loss": 0.069, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.0814802240706162, | |
| "grad_norm": 0.27199000120162964, | |
| "learning_rate": 8.040540540540541e-05, | |
| "loss": 0.0601, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.08827024274316754, | |
| "grad_norm": 0.3759700655937195, | |
| "learning_rate": 8.716216216216216e-05, | |
| "loss": 0.0655, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0950602614157189, | |
| "grad_norm": 0.4591917395591736, | |
| "learning_rate": 9.391891891891892e-05, | |
| "loss": 0.0461, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.10185028008827024, | |
| "grad_norm": 0.2534618675708771, | |
| "learning_rate": 9.999985945746134e-05, | |
| "loss": 0.0515, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1086402987608216, | |
| "grad_norm": 0.23862995207309723, | |
| "learning_rate": 9.998299530880787e-05, | |
| "loss": 0.054, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.11543031743337294, | |
| "grad_norm": 0.3287966847419739, | |
| "learning_rate": 9.993803351513094e-05, | |
| "loss": 0.041, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.1222203361059243, | |
| "grad_norm": 0.3776935636997223, | |
| "learning_rate": 9.986499935143679e-05, | |
| "loss": 0.0355, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.12901035477847564, | |
| "grad_norm": 0.503021776676178, | |
| "learning_rate": 9.976393387344834e-05, | |
| "loss": 0.0417, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.135800373451027, | |
| "grad_norm": 0.27203965187072754, | |
| "learning_rate": 9.963489389452596e-05, | |
| "loss": 0.0312, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.14259039212357835, | |
| "grad_norm": 0.37624090909957886, | |
| "learning_rate": 9.947795195373016e-05, | |
| "loss": 0.0379, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.14938041079612968, | |
| "grad_norm": 0.45090368390083313, | |
| "learning_rate": 9.92931962750442e-05, | |
| "loss": 0.0354, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.15617042946868104, | |
| "grad_norm": 0.26543551683425903, | |
| "learning_rate": 9.908073071777954e-05, | |
| "loss": 0.0274, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.1629604481412324, | |
| "grad_norm": 0.5471091270446777, | |
| "learning_rate": 9.884067471819184e-05, | |
| "loss": 0.0226, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.16975046681378375, | |
| "grad_norm": 0.5838286876678467, | |
| "learning_rate": 9.857316322234067e-05, | |
| "loss": 0.0288, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.17654048548633508, | |
| "grad_norm": 0.16055133938789368, | |
| "learning_rate": 9.827834661023034e-05, | |
| "loss": 0.0262, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.18333050415888644, | |
| "grad_norm": 0.33942335844039917, | |
| "learning_rate": 9.795639061127468e-05, | |
| "loss": 0.0233, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1901205228314378, | |
| "grad_norm": 0.5522304177284241, | |
| "learning_rate": 9.760747621113325e-05, | |
| "loss": 0.0168, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.19691054150398912, | |
| "grad_norm": 0.6882764101028442, | |
| "learning_rate": 9.723179954997125e-05, | |
| "loss": 0.0184, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.20370056017654048, | |
| "grad_norm": 0.30410072207450867, | |
| "learning_rate": 9.682957181220062e-05, | |
| "loss": 0.0203, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.21049057884909184, | |
| "grad_norm": 0.30174189805984497, | |
| "learning_rate": 9.640101910776381e-05, | |
| "loss": 0.0191, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.2172805975216432, | |
| "grad_norm": 0.14357490837574005, | |
| "learning_rate": 9.594638234502753e-05, | |
| "loss": 0.0268, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.22407061619419452, | |
| "grad_norm": 0.19478002190589905, | |
| "learning_rate": 9.546591709535751e-05, | |
| "loss": 0.0218, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.23086063486674588, | |
| "grad_norm": 0.22178731858730316, | |
| "learning_rate": 9.495989344945056e-05, | |
| "loss": 0.0121, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.23765065353929724, | |
| "grad_norm": 0.5312681794166565, | |
| "learning_rate": 9.442859586550468e-05, | |
| "loss": 0.0096, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.2444406722118486, | |
| "grad_norm": 0.43574365973472595, | |
| "learning_rate": 9.387232300931255e-05, | |
| "loss": 0.0129, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.2512306908843999, | |
| "grad_norm": 0.029588760808110237, | |
| "learning_rate": 9.32913875863684e-05, | |
| "loss": 0.0193, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.2580207095569513, | |
| "grad_norm": 0.41019147634506226, | |
| "learning_rate": 9.268611616608236e-05, | |
| "loss": 0.0143, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.26481072822950263, | |
| "grad_norm": 0.13079501688480377, | |
| "learning_rate": 9.205684899820147e-05, | |
| "loss": 0.0197, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.271600746902054, | |
| "grad_norm": 0.04413610324263573, | |
| "learning_rate": 9.140393982154026e-05, | |
| "loss": 0.0163, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.27839076557460535, | |
| "grad_norm": 0.33774223923683167, | |
| "learning_rate": 9.072775566512863e-05, | |
| "loss": 0.0126, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.2851807842471567, | |
| "grad_norm": 0.11442425101995468, | |
| "learning_rate": 9.002867664188858e-05, | |
| "loss": 0.0049, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.291970802919708, | |
| "grad_norm": 0.24148213863372803, | |
| "learning_rate": 8.930709573495602e-05, | |
| "loss": 0.0061, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.29876082159225936, | |
| "grad_norm": 0.49706923961639404, | |
| "learning_rate": 8.856341857676758e-05, | |
| "loss": 0.0069, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.3055508402648107, | |
| "grad_norm": 0.8686341047286987, | |
| "learning_rate": 8.779806322103682e-05, | |
| "loss": 0.0056, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.3123408589373621, | |
| "grad_norm": 0.3294643461704254, | |
| "learning_rate": 8.701145990774775e-05, | |
| "loss": 0.0074, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.31913087760991343, | |
| "grad_norm": 0.14896813035011292, | |
| "learning_rate": 8.620405082129798e-05, | |
| "loss": 0.0098, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.3259208962824648, | |
| "grad_norm": 0.009663172997534275, | |
| "learning_rate": 8.537628984192736e-05, | |
| "loss": 0.0058, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.33271091495501615, | |
| "grad_norm": 0.03118247166275978, | |
| "learning_rate": 8.452864229057191e-05, | |
| "loss": 0.0066, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.3395009336275675, | |
| "grad_norm": 0.2592410743236542, | |
| "learning_rate": 8.366158466728644e-05, | |
| "loss": 0.0038, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.3462909523001188, | |
| "grad_norm": 0.006336142774671316, | |
| "learning_rate": 8.277560438338278e-05, | |
| "loss": 0.0063, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.35308097097267016, | |
| "grad_norm": 0.9991777539253235, | |
| "learning_rate": 8.18711994874345e-05, | |
| "loss": 0.008, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.3598709896452215, | |
| "grad_norm": 0.03354354575276375, | |
| "learning_rate": 8.094887838530174e-05, | |
| "loss": 0.0023, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.3666610083177729, | |
| "grad_norm": 0.1517736166715622, | |
| "learning_rate": 8.000915955433396e-05, | |
| "loss": 0.0072, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.37345102699032423, | |
| "grad_norm": 0.3716999888420105, | |
| "learning_rate": 7.905257125191094e-05, | |
| "loss": 0.0072, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.3802410456628756, | |
| "grad_norm": 0.15485776960849762, | |
| "learning_rate": 7.807965121848597e-05, | |
| "loss": 0.0061, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.38703106433542694, | |
| "grad_norm": 0.015828760340809822, | |
| "learning_rate": 7.70909463752983e-05, | |
| "loss": 0.0031, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.39382108300797825, | |
| "grad_norm": 0.472591370344162, | |
| "learning_rate": 7.608701251692457e-05, | |
| "loss": 0.0061, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.4006111016805296, | |
| "grad_norm": 0.49039226770401, | |
| "learning_rate": 7.506841399884217e-05, | |
| "loss": 0.0032, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.40740112035308096, | |
| "grad_norm": 0.013108485378324986, | |
| "learning_rate": 7.403572342018021e-05, | |
| "loss": 0.0083, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.4141911390256323, | |
| "grad_norm": 0.014069960452616215, | |
| "learning_rate": 7.29895213018362e-05, | |
| "loss": 0.0076, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.4209811576981837, | |
| "grad_norm": 0.016907596960663795, | |
| "learning_rate": 7.193039576013976e-05, | |
| "loss": 0.0035, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.42777117637073503, | |
| "grad_norm": 0.02205067314207554, | |
| "learning_rate": 7.085894217624645e-05, | |
| "loss": 0.0035, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.4345611950432864, | |
| "grad_norm": 0.0153163131326437, | |
| "learning_rate": 6.977576286144784e-05, | |
| "loss": 0.0008, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.44135121371583774, | |
| "grad_norm": 0.0036254250444471836, | |
| "learning_rate": 6.868146671858567e-05, | |
| "loss": 0.0061, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.44814123238838904, | |
| "grad_norm": 0.026128802448511124, | |
| "learning_rate": 6.757666889976085e-05, | |
| "loss": 0.0004, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.4549312510609404, | |
| "grad_norm": 0.009841466322541237, | |
| "learning_rate": 6.646199046052921e-05, | |
| "loss": 0.0054, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.46172126973349176, | |
| "grad_norm": 0.1286161094903946, | |
| "learning_rate": 6.533805801077899e-05, | |
| "loss": 0.0083, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.4685112884060431, | |
| "grad_norm": 0.1420389711856842, | |
| "learning_rate": 6.420550336248558e-05, | |
| "loss": 0.0011, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.47530130707859447, | |
| "grad_norm": 0.019207162782549858, | |
| "learning_rate": 6.306496317454227e-05, | |
| "loss": 0.0014, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.4820913257511458, | |
| "grad_norm": 0.02628934569656849, | |
| "learning_rate": 6.191707859486622e-05, | |
| "loss": 0.0071, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.4888813444236972, | |
| "grad_norm": 0.07600024342536926, | |
| "learning_rate": 6.076249489998097e-05, | |
| "loss": 0.0007, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.49567136309624854, | |
| "grad_norm": 0.057020459324121475, | |
| "learning_rate": 5.9601861132278e-05, | |
| "loss": 0.0029, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.5024613817687998, | |
| "grad_norm": 0.18588444590568542, | |
| "learning_rate": 5.8435829735161416e-05, | |
| "loss": 0.0015, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.5092514004413512, | |
| "grad_norm": 0.08587218075990677, | |
| "learning_rate": 5.7265056186280754e-05, | |
| "loss": 0.0064, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.5160414191139026, | |
| "grad_norm": 0.027101216837763786, | |
| "learning_rate": 5.6090198629058134e-05, | |
| "loss": 0.0002, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.5228314377864539, | |
| "grad_norm": 0.008036206476390362, | |
| "learning_rate": 5.491191750271677e-05, | |
| "loss": 0.0096, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.5296214564590053, | |
| "grad_norm": 0.02794210985302925, | |
| "learning_rate": 5.373087517101899e-05, | |
| "loss": 0.0005, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.5364114751315566, | |
| "grad_norm": 0.015519050881266594, | |
| "learning_rate": 5.254773554992242e-05, | |
| "loss": 0.0012, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.543201493804108, | |
| "grad_norm": 0.5680004954338074, | |
| "learning_rate": 5.136316373436343e-05, | |
| "loss": 0.0011, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.5499915124766593, | |
| "grad_norm": 0.01568940281867981, | |
| "learning_rate": 5.0177825624378114e-05, | |
| "loss": 0.0006, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.5567815311492107, | |
| "grad_norm": 0.15695148706436157, | |
| "learning_rate": 4.8992387550770373e-05, | |
| "loss": 0.0064, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.563571549821762, | |
| "grad_norm": 0.24050813913345337, | |
| "learning_rate": 4.780751590053813e-05, | |
| "loss": 0.0054, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.5703615684943134, | |
| "grad_norm": 0.008626952767372131, | |
| "learning_rate": 4.662387674226771e-05, | |
| "loss": 0.0032, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.5771515871668647, | |
| "grad_norm": 0.007616691756993532, | |
| "learning_rate": 4.544213545170741e-05, | |
| "loss": 0.0039, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.583941605839416, | |
| "grad_norm": 0.014656963758170605, | |
| "learning_rate": 4.4262956337730385e-05, | |
| "loss": 0.0012, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.5907316245119674, | |
| "grad_norm": 0.2970845401287079, | |
| "learning_rate": 4.3087002268897434e-05, | |
| "loss": 0.0011, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.5975216431845187, | |
| "grad_norm": 0.010836818255484104, | |
| "learning_rate": 4.191493430082929e-05, | |
| "loss": 0.0059, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.6043116618570701, | |
| "grad_norm": 0.016124485060572624, | |
| "learning_rate": 4.074741130459813e-05, | |
| "loss": 0.0051, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.6111016805296214, | |
| "grad_norm": 0.008550368249416351, | |
| "learning_rate": 3.9585089596347094e-05, | |
| "loss": 0.0006, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.6178916992021728, | |
| "grad_norm": 0.007996040396392345, | |
| "learning_rate": 3.842862256834613e-05, | |
| "loss": 0.0013, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.6246817178747242, | |
| "grad_norm": 0.01987188495695591, | |
| "learning_rate": 3.727866032169127e-05, | |
| "loss": 0.002, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.6314717365472755, | |
| "grad_norm": 0.006365388166159391, | |
| "learning_rate": 3.6135849300854225e-05, | |
| "loss": 0.0003, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.6382617552198269, | |
| "grad_norm": 0.029804030433297157, | |
| "learning_rate": 3.500083193028741e-05, | |
| "loss": 0.0052, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.6450517738923782, | |
| "grad_norm": 0.009540177881717682, | |
| "learning_rate": 3.387424625328892e-05, | |
| "loss": 0.0004, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.6518417925649296, | |
| "grad_norm": 0.007210243493318558, | |
| "learning_rate": 3.275672557333015e-05, | |
| "loss": 0.0038, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.6586318112374809, | |
| "grad_norm": 0.024687010794878006, | |
| "learning_rate": 3.164889809804808e-05, | |
| "loss": 0.0022, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.6654218299100323, | |
| "grad_norm": 0.005847941618412733, | |
| "learning_rate": 3.055138658610202e-05, | |
| "loss": 0.0035, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.6722118485825836, | |
| "grad_norm": 0.005948640406131744, | |
| "learning_rate": 2.946480799709358e-05, | |
| "loss": 0.0015, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.679001867255135, | |
| "grad_norm": 0.018232271075248718, | |
| "learning_rate": 2.83897731447464e-05, | |
| "loss": 0.0056, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.6857918859276863, | |
| "grad_norm": 0.4337656795978546, | |
| "learning_rate": 2.7326886353540916e-05, | |
| "loss": 0.0027, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.6925819046002376, | |
| "grad_norm": 0.4810737371444702, | |
| "learning_rate": 2.627674511899678e-05, | |
| "loss": 0.0042, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.699371923272789, | |
| "grad_norm": 0.024683628231287003, | |
| "learning_rate": 2.5239939771794563e-05, | |
| "loss": 0.0006, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.7061619419453403, | |
| "grad_norm": 0.3206782639026642, | |
| "learning_rate": 2.421705314592452e-05, | |
| "loss": 0.0038, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.7129519606178917, | |
| "grad_norm": 0.194807767868042, | |
| "learning_rate": 2.3208660251050158e-05, | |
| "loss": 0.0012, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.719741979290443, | |
| "grad_norm": 0.009538485668599606, | |
| "learning_rate": 2.2215327949269886e-05, | |
| "loss": 0.0004, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.7265319979629944, | |
| "grad_norm": 0.00484565319493413, | |
| "learning_rate": 2.123761463645878e-05, | |
| "loss": 0.0003, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.7333220166355457, | |
| "grad_norm": 0.0036626686342060566, | |
| "learning_rate": 2.0276069928369746e-05, | |
| "loss": 0.0002, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.7401120353080971, | |
| "grad_norm": 0.006346094887703657, | |
| "learning_rate": 1.933123435167018e-05, | |
| "loss": 0.0008, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.7469020539806485, | |
| "grad_norm": 0.0050748297944664955, | |
| "learning_rate": 1.840363904008819e-05, | |
| "loss": 0.0002, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.7536920726531998, | |
| "grad_norm": 0.0035511874593794346, | |
| "learning_rate": 1.7493805435838874e-05, | |
| "loss": 0.0002, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.7604820913257512, | |
| "grad_norm": 0.01719609834253788, | |
| "learning_rate": 1.660224499649861e-05, | |
| "loss": 0.0012, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.7672721099983025, | |
| "grad_norm": 0.5908659100532532, | |
| "learning_rate": 1.5729458907492327e-05, | |
| "loss": 0.0021, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.7740621286708539, | |
| "grad_norm": 0.0023723470512777567, | |
| "learning_rate": 1.4875937800354988e-05, | |
| "loss": 0.0002, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.7808521473434052, | |
| "grad_norm": 0.003473122837021947, | |
| "learning_rate": 1.404216147692598e-05, | |
| "loss": 0.0001, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.7876421660159565, | |
| "grad_norm": 0.0032885069958865643, | |
| "learning_rate": 1.3228598639631418e-05, | |
| "loss": 0.0046, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.7944321846885078, | |
| "grad_norm": 0.004905693233013153, | |
| "learning_rate": 1.2435706628005767e-05, | |
| "loss": 0.0016, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.8012222033610592, | |
| "grad_norm": 0.0058155064471066, | |
| "learning_rate": 1.1663931161601188e-05, | |
| "loss": 0.0027, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.8080122220336106, | |
| "grad_norm": 0.007166674826294184, | |
| "learning_rate": 1.0913706089428932e-05, | |
| "loss": 0.003, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.8148022407061619, | |
| "grad_norm": 0.002422675024718046, | |
| "learning_rate": 1.0185453146073604e-05, | |
| "loss": 0.0002, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.8215922593787133, | |
| "grad_norm": 0.5606914758682251, | |
| "learning_rate": 9.479581714617668e-06, | |
| "loss": 0.0043, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.8283822780512646, | |
| "grad_norm": 0.018740450963377953, | |
| "learning_rate": 8.796488596509133e-06, | |
| "loss": 0.0011, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.835172296723816, | |
| "grad_norm": 0.0064006103202700615, | |
| "learning_rate": 8.136557788501903e-06, | |
| "loss": 0.0023, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.8419623153963673, | |
| "grad_norm": 0.01389048807322979, | |
| "learning_rate": 7.500160266794371e-06, | |
| "loss": 0.0003, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.8487523340689187, | |
| "grad_norm": 0.0191491711884737, | |
| "learning_rate": 6.887653778487307e-06, | |
| "loss": 0.0037, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.8555423527414701, | |
| "grad_norm": 0.010205433703958988, | |
| "learning_rate": 6.2993826404783965e-06, | |
| "loss": 0.0004, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.8623323714140214, | |
| "grad_norm": 0.005938360001891851, | |
| "learning_rate": 5.735677545906626e-06, | |
| "loss": 0.0002, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.8691223900865728, | |
| "grad_norm": 0.42641371488571167, | |
| "learning_rate": 5.196855378254989e-06, | |
| "loss": 0.0019, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.8759124087591241, | |
| "grad_norm": 0.008787781931459904, | |
| "learning_rate": 4.683219033216402e-06, | |
| "loss": 0.0006, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.8827024274316755, | |
| "grad_norm": 0.001734372926875949, | |
| "learning_rate": 4.1950572484226345e-06, | |
| "loss": 0.0002, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.8894924461042268, | |
| "grad_norm": 0.13424113392829895, | |
| "learning_rate": 3.7326444411321547e-06, | |
| "loss": 0.0003, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.8962824647767781, | |
| "grad_norm": 0.0027601660694926977, | |
| "learning_rate": 3.2962405539681217e-06, | |
| "loss": 0.0002, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.9030724834493294, | |
| "grad_norm": 0.011034502647817135, | |
| "learning_rate": 2.8860909087931543e-06, | |
| "loss": 0.0002, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.9098625021218808, | |
| "grad_norm": 0.00510817626491189, | |
| "learning_rate": 2.5024260688030987e-06, | |
| "loss": 0.0002, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.9166525207944322, | |
| "grad_norm": 0.006623192224651575, | |
| "learning_rate": 2.145461708917312e-06, | |
| "loss": 0.0019, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.9234425394669835, | |
| "grad_norm": 0.008591300807893276, | |
| "learning_rate": 1.8153984945382452e-06, | |
| "loss": 0.0002, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.9302325581395349, | |
| "grad_norm": 0.0064789485186338425, | |
| "learning_rate": 1.512421968748623e-06, | |
| "loss": 0.0006, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.9370225768120862, | |
| "grad_norm": 0.009375182911753654, | |
| "learning_rate": 1.2367024480094691e-06, | |
| "loss": 0.0026, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.9438125954846376, | |
| "grad_norm": 0.005188928451389074, | |
| "learning_rate": 9.883949264176962e-07, | |
| "loss": 0.0003, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.9506026141571889, | |
| "grad_norm": 0.006225943099707365, | |
| "learning_rate": 7.676389885771518e-07, | |
| "loss": 0.0015, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.9573926328297403, | |
| "grad_norm": 0.0153355086222291, | |
| "learning_rate": 5.745587311318968e-07, | |
| "loss": 0.0002, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.9641826515022917, | |
| "grad_norm": 0.0042914156801998615, | |
| "learning_rate": 4.0926269300603503e-07, | |
| "loss": 0.0002, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.970972670174843, | |
| "grad_norm": 0.011045188643038273, | |
| "learning_rate": 2.71843794389226e-07, | |
| "loss": 0.0002, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.9777626888473944, | |
| "grad_norm": 0.00516318716108799, | |
| "learning_rate": 1.623792845020955e-07, | |
| "loss": 0.0004, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.9845527075199457, | |
| "grad_norm": 0.008642218075692654, | |
| "learning_rate": 8.093069817109445e-08, | |
| "loss": 0.0002, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.9913427261924971, | |
| "grad_norm": 0.010795416310429573, | |
| "learning_rate": 2.7543821237030475e-08, | |
| "loss": 0.0002, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.9981327448650483, | |
| "grad_norm": 0.012065191753208637, | |
| "learning_rate": 2.2486648168396075e-09, | |
| "loss": 0.0002, | |
| "step": 1470 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 1473, | |
| "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": 6.075096721816289e+17, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |