| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 0.7917374080362196, |
| "eval_steps": 500, |
| "global_step": 1399, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.005659309564233163, |
| "grad_norm": 744.6958561130482, |
| "learning_rate": 8.240909090909092e-06, |
| "loss": 1.440648078918457, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.011318619128466326, |
| "grad_norm": 17.48279794016324, |
| "learning_rate": 1.728636363636364e-05, |
| "loss": 0.7085334777832031, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.01697792869269949, |
| "grad_norm": 86.07861482821112, |
| "learning_rate": 2e-05, |
| "loss": 0.6137940883636475, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.022637238256932653, |
| "grad_norm": 157.19160579458676, |
| "learning_rate": 2e-05, |
| "loss": 0.5798912048339844, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.028296547821165818, |
| "grad_norm": 32.724114045514916, |
| "learning_rate": 2e-05, |
| "loss": 0.5812548637390137, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.03395585738539898, |
| "grad_norm": 107.34079516623102, |
| "learning_rate": 2e-05, |
| "loss": 0.5763686180114747, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.039615166949632144, |
| "grad_norm": 170.53323192528686, |
| "learning_rate": 2e-05, |
| "loss": 0.5897064208984375, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.045274476513865305, |
| "grad_norm": 12.93590651750023, |
| "learning_rate": 2e-05, |
| "loss": 0.5879075527191162, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.050933786078098474, |
| "grad_norm": 5.4235068968091165, |
| "learning_rate": 2e-05, |
| "loss": 0.5804771423339844, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.056593095642331635, |
| "grad_norm": 2.4857016216612546, |
| "learning_rate": 2e-05, |
| "loss": 0.5606931209564209, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.0622524052065648, |
| "grad_norm": 10.660553013604694, |
| "learning_rate": 2e-05, |
| "loss": 0.5512332439422607, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.06791171477079797, |
| "grad_norm": 3.4384625931751436, |
| "learning_rate": 2e-05, |
| "loss": 0.5416618347167969, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.07357102433503113, |
| "grad_norm": 2.117791360157757, |
| "learning_rate": 2e-05, |
| "loss": 0.5654376029968262, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.07923033389926429, |
| "grad_norm": 17.54985314509229, |
| "learning_rate": 2e-05, |
| "loss": 0.5611878871917725, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.08488964346349745, |
| "grad_norm": 5.113064762862566, |
| "learning_rate": 2e-05, |
| "loss": 0.5519698143005372, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.09054895302773061, |
| "grad_norm": 6.499369801100052, |
| "learning_rate": 2e-05, |
| "loss": 0.5497481822967529, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.09620826259196379, |
| "grad_norm": 6.755673272588835, |
| "learning_rate": 2e-05, |
| "loss": 0.534516429901123, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.10186757215619695, |
| "grad_norm": 73.69118025858943, |
| "learning_rate": 2e-05, |
| "loss": 0.5282778263092041, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.10752688172043011, |
| "grad_norm": 4.756927243710702, |
| "learning_rate": 2e-05, |
| "loss": 0.5542384147644043, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.11318619128466327, |
| "grad_norm": 2.7730655651433205, |
| "learning_rate": 2e-05, |
| "loss": 0.5678863525390625, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.11884550084889643, |
| "grad_norm": 3.2192407132264984, |
| "learning_rate": 2e-05, |
| "loss": 0.5510906219482422, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.1245048104131296, |
| "grad_norm": 7.006627374817867, |
| "learning_rate": 2e-05, |
| "loss": 0.5367250442504883, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.13016411997736277, |
| "grad_norm": 7.157530546935823, |
| "learning_rate": 2e-05, |
| "loss": 0.5382485866546631, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.13582342954159593, |
| "grad_norm": 37.70011856458165, |
| "learning_rate": 2e-05, |
| "loss": 0.5416894435882569, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.1414827391058291, |
| "grad_norm": 2.951278644932636, |
| "learning_rate": 2e-05, |
| "loss": 0.5506397247314453, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.14714204867006225, |
| "grad_norm": 93.81696688025806, |
| "learning_rate": 2e-05, |
| "loss": 0.5345812797546386, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.15280135823429541, |
| "grad_norm": 5.57362465944681, |
| "learning_rate": 2e-05, |
| "loss": 0.5400020122528076, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.15846066779852858, |
| "grad_norm": 4.914063909235412, |
| "learning_rate": 2e-05, |
| "loss": 0.5373981475830079, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.16411997736276174, |
| "grad_norm": 4.666262118175498, |
| "learning_rate": 2e-05, |
| "loss": 0.5310977458953857, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.1697792869269949, |
| "grad_norm": 5.248995451196224, |
| "learning_rate": 2e-05, |
| "loss": 0.542457103729248, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.17543859649122806, |
| "grad_norm": 2.910218284953411, |
| "learning_rate": 2e-05, |
| "loss": 0.5338900566101075, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.18109790605546122, |
| "grad_norm": 3.509128211188278, |
| "learning_rate": 2e-05, |
| "loss": 0.5293301582336426, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.1867572156196944, |
| "grad_norm": 25.69752157756616, |
| "learning_rate": 2e-05, |
| "loss": 0.5202945709228516, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.19241652518392757, |
| "grad_norm": 2.3594081938212117, |
| "learning_rate": 2e-05, |
| "loss": 0.5250492095947266, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.19807583474816073, |
| "grad_norm": 1.11935102001747, |
| "learning_rate": 2e-05, |
| "loss": 0.5368710994720459, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2037351443123939, |
| "grad_norm": 1.192612217147343, |
| "learning_rate": 2e-05, |
| "loss": 0.5384870052337647, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.20939445387662706, |
| "grad_norm": 1.4696082379739595, |
| "learning_rate": 2e-05, |
| "loss": 0.537387752532959, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.21505376344086022, |
| "grad_norm": 0.7859626291409186, |
| "learning_rate": 2e-05, |
| "loss": 0.5352621608310275, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.22071307300509338, |
| "grad_norm": 0.7841049279370906, |
| "learning_rate": 2e-05, |
| "loss": 0.5250838756561279, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.22637238256932654, |
| "grad_norm": 0.9563604693764177, |
| "learning_rate": 2e-05, |
| "loss": 0.5247737407684326, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.2320316921335597, |
| "grad_norm": 5.240248825021192, |
| "learning_rate": 2e-05, |
| "loss": 0.5204930782318116, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.23769100169779286, |
| "grad_norm": 8.857184939031043, |
| "learning_rate": 2e-05, |
| "loss": 0.5226087093353271, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.24335031126202603, |
| "grad_norm": 22.418904301264167, |
| "learning_rate": 2e-05, |
| "loss": 0.5341937939325968, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.2490096208262592, |
| "grad_norm": 7.835851273738778, |
| "learning_rate": 2e-05, |
| "loss": 0.5139634132385253, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.2546689303904924, |
| "grad_norm": 16.49609548803437, |
| "learning_rate": 2e-05, |
| "loss": 0.5270539283752441, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.26032823995472554, |
| "grad_norm": 1.0759613256297782, |
| "learning_rate": 2e-05, |
| "loss": 0.5301344394683838, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.2659875495189587, |
| "grad_norm": 0.9878440106310662, |
| "learning_rate": 2e-05, |
| "loss": 0.525136661529541, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.27164685908319186, |
| "grad_norm": 0.5697186684602807, |
| "learning_rate": 2e-05, |
| "loss": 0.520656156539917, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.277306168647425, |
| "grad_norm": 1.934528822541238, |
| "learning_rate": 2e-05, |
| "loss": 0.528661060333252, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.2829654782116582, |
| "grad_norm": 1.0020714377947892, |
| "learning_rate": 2e-05, |
| "loss": 0.512585973739624, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.28862478777589134, |
| "grad_norm": 2.593668088509703, |
| "learning_rate": 2e-05, |
| "loss": 0.5176972389221192, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.2942840973401245, |
| "grad_norm": 2.1122510918063107, |
| "learning_rate": 2e-05, |
| "loss": 0.5184473514556884, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.29994340690435767, |
| "grad_norm": 10.30593354350407, |
| "learning_rate": 2e-05, |
| "loss": 0.5069304943084717, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.30560271646859083, |
| "grad_norm": 4.212501373740993, |
| "learning_rate": 2e-05, |
| "loss": 0.5282890319824218, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.311262026032824, |
| "grad_norm": 0.9464912723573213, |
| "learning_rate": 2e-05, |
| "loss": 0.5245438098907471, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.31692133559705715, |
| "grad_norm": 0.5845031124464878, |
| "learning_rate": 2e-05, |
| "loss": 0.5293712615966797, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.3225806451612903, |
| "grad_norm": 1.4076062699369296, |
| "learning_rate": 2e-05, |
| "loss": 0.5228209972381592, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.3282399547255235, |
| "grad_norm": 4.508342493184027, |
| "learning_rate": 2e-05, |
| "loss": 0.5179474830627442, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.33389926428975664, |
| "grad_norm": 0.5720914840137702, |
| "learning_rate": 2e-05, |
| "loss": 0.516933012008667, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.3395585738539898, |
| "grad_norm": 1.8430897412288034, |
| "learning_rate": 2e-05, |
| "loss": 0.5143716335296631, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.34521788341822296, |
| "grad_norm": 1.118320609555216, |
| "learning_rate": 2e-05, |
| "loss": 0.5236732006072998, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.3508771929824561, |
| "grad_norm": 0.8979781579023246, |
| "learning_rate": 2e-05, |
| "loss": 0.5099713802337646, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.3565365025466893, |
| "grad_norm": 1.1592160810439176, |
| "learning_rate": 2e-05, |
| "loss": 0.5082225799560547, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.36219581211092244, |
| "grad_norm": 0.766890404533274, |
| "learning_rate": 2e-05, |
| "loss": 0.505848503112793, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.3678551216751556, |
| "grad_norm": 1.6376848466786926, |
| "learning_rate": 2e-05, |
| "loss": 0.5080933570861816, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.3735144312393888, |
| "grad_norm": 1.7392890620498735, |
| "learning_rate": 2e-05, |
| "loss": 0.5076011657714844, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.379173740803622, |
| "grad_norm": 1.4529546486266038, |
| "learning_rate": 2e-05, |
| "loss": 0.5106268405914307, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.38483305036785515, |
| "grad_norm": 0.9734931895867416, |
| "learning_rate": 2e-05, |
| "loss": 0.5103437900543213, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.3904923599320883, |
| "grad_norm": 0.6914274328589272, |
| "learning_rate": 2e-05, |
| "loss": 0.5173460960388183, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.39615166949632147, |
| "grad_norm": 1.7290932161763068, |
| "learning_rate": 2e-05, |
| "loss": 0.5062826156616211, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.40181097906055463, |
| "grad_norm": 2.100161887785846, |
| "learning_rate": 2e-05, |
| "loss": 0.5085488319396972, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.4074702886247878, |
| "grad_norm": 1.5778477154432393, |
| "learning_rate": 2e-05, |
| "loss": 0.4917932033538818, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.41312959818902095, |
| "grad_norm": 6.504004809363307, |
| "learning_rate": 2e-05, |
| "loss": 0.5028924942016602, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.4187889077532541, |
| "grad_norm": 2.3903000103786773, |
| "learning_rate": 2e-05, |
| "loss": 0.5029994964599609, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.4244482173174873, |
| "grad_norm": 1.8935488043625464, |
| "learning_rate": 2e-05, |
| "loss": 0.508833360671997, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.43010752688172044, |
| "grad_norm": 0.9820912768131386, |
| "learning_rate": 2e-05, |
| "loss": 0.5064041137695312, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.4357668364459536, |
| "grad_norm": 0.7627396067485848, |
| "learning_rate": 2e-05, |
| "loss": 0.5121818065643311, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.44142614601018676, |
| "grad_norm": 4.757596638422343, |
| "learning_rate": 2e-05, |
| "loss": 0.5105973720550537, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.4470854555744199, |
| "grad_norm": 1.1861037116325448, |
| "learning_rate": 2e-05, |
| "loss": 0.5016807079315185, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.4527447651386531, |
| "grad_norm": 1.3529926335099773, |
| "learning_rate": 2e-05, |
| "loss": 0.49750475883483886, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.45840407470288624, |
| "grad_norm": 1.5151684382033037, |
| "learning_rate": 2e-05, |
| "loss": 0.5030447483062744, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.4640633842671194, |
| "grad_norm": 1.2068881443148534, |
| "learning_rate": 2e-05, |
| "loss": 0.5149904251098633, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.46972269383135257, |
| "grad_norm": 2.825100271078658, |
| "learning_rate": 2e-05, |
| "loss": 0.5099709033966064, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.47538200339558573, |
| "grad_norm": 1.7497159418720338, |
| "learning_rate": 2e-05, |
| "loss": 0.5079097270965576, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.4810413129598189, |
| "grad_norm": 1.1299988546641022, |
| "learning_rate": 2e-05, |
| "loss": 0.506084394454956, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.48670062252405205, |
| "grad_norm": 2.6872075335974235, |
| "learning_rate": 2e-05, |
| "loss": 0.5105088233947754, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.4923599320882852, |
| "grad_norm": 3.063389539777061, |
| "learning_rate": 2e-05, |
| "loss": 0.5057406425476074, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.4980192416525184, |
| "grad_norm": 4.150307544078542, |
| "learning_rate": 2e-05, |
| "loss": 0.5026792526245117, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.5036785512167515, |
| "grad_norm": 2.873287947619104, |
| "learning_rate": 2e-05, |
| "loss": 0.5136403560638427, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.5093378607809848, |
| "grad_norm": 1.6426401998876443, |
| "learning_rate": 2e-05, |
| "loss": 0.5109119415283203, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.5149971703452179, |
| "grad_norm": 3.5388592650813675, |
| "learning_rate": 2e-05, |
| "loss": 0.5001020431518555, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.5206564799094511, |
| "grad_norm": 1.704676559198332, |
| "learning_rate": 2e-05, |
| "loss": 0.5094768524169921, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.5263157894736842, |
| "grad_norm": 1.605344381343744, |
| "learning_rate": 2e-05, |
| "loss": 0.49990320205688477, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.5319750990379174, |
| "grad_norm": 0.7374565936911701, |
| "learning_rate": 2e-05, |
| "loss": 0.514597225189209, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.5376344086021505, |
| "grad_norm": 0.9883431603589855, |
| "learning_rate": 2e-05, |
| "loss": 0.5054900646209717, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.5432937181663837, |
| "grad_norm": 1.984621659847055, |
| "learning_rate": 2e-05, |
| "loss": 0.5039671421051025, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.5489530277306168, |
| "grad_norm": 1.3710572996747166, |
| "learning_rate": 2e-05, |
| "loss": 0.5205180168151855, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.55461233729485, |
| "grad_norm": 1.2670249611982882, |
| "learning_rate": 2e-05, |
| "loss": 0.49703097343444824, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.5602716468590832, |
| "grad_norm": 4.783741391685528, |
| "learning_rate": 2e-05, |
| "loss": 0.5074984550476074, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.5659309564233164, |
| "grad_norm": 1.8376255696712758, |
| "learning_rate": 2e-05, |
| "loss": 0.5061185836791993, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.5715902659875495, |
| "grad_norm": 1.333054330915451, |
| "learning_rate": 2e-05, |
| "loss": 0.5070064544677735, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.5772495755517827, |
| "grad_norm": 1.2127463728652028, |
| "learning_rate": 2e-05, |
| "loss": 0.5048279285430908, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.5829088851160158, |
| "grad_norm": 0.5786969457237549, |
| "learning_rate": 2e-05, |
| "loss": 0.4991403102874756, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.588568194680249, |
| "grad_norm": 1.6337441254695233, |
| "learning_rate": 2e-05, |
| "loss": 0.5029354095458984, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.5942275042444821, |
| "grad_norm": 0.6596247281982506, |
| "learning_rate": 2e-05, |
| "loss": 0.4930370330810547, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.5998868138087153, |
| "grad_norm": 0.918792010554476, |
| "learning_rate": 2e-05, |
| "loss": 0.49286723136901855, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.6055461233729486, |
| "grad_norm": 0.729701235920584, |
| "learning_rate": 2e-05, |
| "loss": 0.500383996963501, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.6112054329371817, |
| "grad_norm": 1.1175698516708477, |
| "learning_rate": 2e-05, |
| "loss": 0.48552513122558594, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.6168647425014149, |
| "grad_norm": 1.3339103436735449, |
| "learning_rate": 2e-05, |
| "loss": 0.4872349739074707, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.622524052065648, |
| "grad_norm": 1.2346983705110355, |
| "learning_rate": 2e-05, |
| "loss": 0.4872392177581787, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.6281833616298812, |
| "grad_norm": 1.1260907657525638, |
| "learning_rate": 2e-05, |
| "loss": 0.4961869239807129, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.6338426711941143, |
| "grad_norm": 0.8748901105410148, |
| "learning_rate": 2e-05, |
| "loss": 0.4913774013519287, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.6395019807583475, |
| "grad_norm": 0.7879741555047347, |
| "learning_rate": 2e-05, |
| "loss": 0.48564891815185546, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.6451612903225806, |
| "grad_norm": 0.8856235339776327, |
| "learning_rate": 2e-05, |
| "loss": 0.4924321174621582, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.6508205998868138, |
| "grad_norm": 0.5381037582804371, |
| "learning_rate": 2e-05, |
| "loss": 0.4967320442199707, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.656479909451047, |
| "grad_norm": 0.830947391810387, |
| "learning_rate": 2e-05, |
| "loss": 0.4909396171569824, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.6621392190152802, |
| "grad_norm": 1.9059911967468053, |
| "learning_rate": 2e-05, |
| "loss": 0.49750614166259766, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.6677985285795133, |
| "grad_norm": 1.8286192060038517, |
| "learning_rate": 2e-05, |
| "loss": 0.49876904487609863, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.6734578381437465, |
| "grad_norm": 1.3859061620749373, |
| "learning_rate": 2e-05, |
| "loss": 0.4920955181121826, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.6791171477079796, |
| "grad_norm": 3.5643991692726624, |
| "learning_rate": 2e-05, |
| "loss": 0.49145846366882323, |
| "step": 1200 |
| }, |
| { |
| "epoch": 0.6847764572722128, |
| "grad_norm": 2.0352089345510103, |
| "learning_rate": 2e-05, |
| "loss": 0.48438310623168945, |
| "step": 1210 |
| }, |
| { |
| "epoch": 0.6904357668364459, |
| "grad_norm": 9.035623506275542, |
| "learning_rate": 2e-05, |
| "loss": 0.4872251510620117, |
| "step": 1220 |
| }, |
| { |
| "epoch": 0.6960950764006791, |
| "grad_norm": 3.210092622192876, |
| "learning_rate": 2e-05, |
| "loss": 0.4969188213348389, |
| "step": 1230 |
| }, |
| { |
| "epoch": 0.7017543859649122, |
| "grad_norm": 1.6601081953764236, |
| "learning_rate": 2e-05, |
| "loss": 0.48545141220092775, |
| "step": 1240 |
| }, |
| { |
| "epoch": 0.7074136955291455, |
| "grad_norm": 1.5054115950802458, |
| "learning_rate": 2e-05, |
| "loss": 0.4934364318847656, |
| "step": 1250 |
| }, |
| { |
| "epoch": 0.7130730050933786, |
| "grad_norm": 1.1857008917699552, |
| "learning_rate": 2e-05, |
| "loss": 0.49532232284545896, |
| "step": 1260 |
| }, |
| { |
| "epoch": 0.7187323146576118, |
| "grad_norm": 2.4227566470294737, |
| "learning_rate": 2e-05, |
| "loss": 0.4930436611175537, |
| "step": 1270 |
| }, |
| { |
| "epoch": 0.7243916242218449, |
| "grad_norm": 1.3594449924948224, |
| "learning_rate": 1.985120236000677e-05, |
| "loss": 0.4916706085205078, |
| "step": 1280 |
| }, |
| { |
| "epoch": 0.7300509337860781, |
| "grad_norm": 0.6110965406930631, |
| "learning_rate": 1.9133090868227833e-05, |
| "loss": 0.48916125297546387, |
| "step": 1290 |
| }, |
| { |
| "epoch": 0.7357102433503112, |
| "grad_norm": 1.8747901595509369, |
| "learning_rate": 1.786199834829156e-05, |
| "loss": 0.4878711223602295, |
| "step": 1300 |
| }, |
| { |
| "epoch": 0.7413695529145444, |
| "grad_norm": 0.6709150232616968, |
| "learning_rate": 1.6115309256519543e-05, |
| "loss": 0.48683719635009765, |
| "step": 1310 |
| }, |
| { |
| "epoch": 0.7470288624787776, |
| "grad_norm": 0.65539635519017, |
| "learning_rate": 1.399936249712973e-05, |
| "loss": 0.4867414474487305, |
| "step": 1320 |
| }, |
| { |
| "epoch": 0.7526881720430108, |
| "grad_norm": 1.7751575692660173, |
| "learning_rate": 1.1642977480993187e-05, |
| "loss": 0.48627915382385256, |
| "step": 1330 |
| }, |
| { |
| "epoch": 0.758347481607244, |
| "grad_norm": 1.621602390427581, |
| "learning_rate": 9.189611565067083e-06, |
| "loss": 0.4905862808227539, |
| "step": 1340 |
| }, |
| { |
| "epoch": 0.7640067911714771, |
| "grad_norm": 0.6855138308905224, |
| "learning_rate": 6.788626329711741e-06, |
| "loss": 0.48851566314697265, |
| "step": 1350 |
| }, |
| { |
| "epoch": 0.7696661007357103, |
| "grad_norm": 0.8023388696855852, |
| "learning_rate": 4.5861944052904325e-06, |
| "loss": 0.4804098129272461, |
| "step": 1360 |
| }, |
| { |
| "epoch": 0.7753254102999434, |
| "grad_norm": 1.0107512381048729, |
| "learning_rate": 2.7164004429072312e-06, |
| "loss": 0.48613681793212893, |
| "step": 1370 |
| }, |
| { |
| "epoch": 0.7809847198641766, |
| "grad_norm": 0.6292657020838206, |
| "learning_rate": 1.2930780045908777e-06, |
| "loss": 0.4785513401031494, |
| "step": 1380 |
| }, |
| { |
| "epoch": 0.7866440294284097, |
| "grad_norm": 1.5981312599351896, |
| "learning_rate": 4.028793452574461e-07, |
| "loss": 0.4815356731414795, |
| "step": 1390 |
| }, |
| { |
| "epoch": 0.7917374080362196, |
| "step": 1399, |
| "total_flos": 9636328589230080.0, |
| "train_loss": 0.044585333967992795, |
| "train_runtime": 5284.1263, |
| "train_samples_per_second": 25.417, |
| "train_steps_per_second": 0.265 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1399, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 1304, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 9636328589230080.0, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
