| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 5.0, |
| "eval_steps": 500, |
| "global_step": 1230, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.04069175991861648, |
| "grad_norm": 59.704185485839844, |
| "learning_rate": 7.317073170731707e-07, |
| "loss": 1.3787, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.08138351983723296, |
| "grad_norm": 50.97575378417969, |
| "learning_rate": 1.5447154471544717e-06, |
| "loss": 1.2265, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.12207527975584945, |
| "grad_norm": 30.238018035888672, |
| "learning_rate": 2.3577235772357727e-06, |
| "loss": 0.837, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.16276703967446593, |
| "grad_norm": 14.13187026977539, |
| "learning_rate": 3.1707317073170736e-06, |
| "loss": 0.6985, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.2034587995930824, |
| "grad_norm": 17.53030014038086, |
| "learning_rate": 3.983739837398374e-06, |
| "loss": 0.5767, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.2441505595116989, |
| "grad_norm": 31.861324310302734, |
| "learning_rate": 4.796747967479675e-06, |
| "loss": 0.7292, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.28484231943031535, |
| "grad_norm": 21.95199203491211, |
| "learning_rate": 5.609756097560977e-06, |
| "loss": 0.7324, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.32553407934893186, |
| "grad_norm": 19.857227325439453, |
| "learning_rate": 6.422764227642278e-06, |
| "loss": 0.5797, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.3662258392675483, |
| "grad_norm": 18.20450782775879, |
| "learning_rate": 7.2357723577235786e-06, |
| "loss": 0.6719, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.4069175991861648, |
| "grad_norm": 14.949827194213867, |
| "learning_rate": 8.048780487804879e-06, |
| "loss": 0.5985, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.4476093591047813, |
| "grad_norm": 19.118457794189453, |
| "learning_rate": 8.86178861788618e-06, |
| "loss": 0.5965, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.4883011190233978, |
| "grad_norm": 13.308806419372559, |
| "learning_rate": 9.67479674796748e-06, |
| "loss": 0.7175, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.5289928789420142, |
| "grad_norm": 18.49018669128418, |
| "learning_rate": 9.999275169399614e-06, |
| "loss": 0.6086, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.5696846388606307, |
| "grad_norm": 18.213380813598633, |
| "learning_rate": 9.994846410084447e-06, |
| "loss": 0.6615, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.6103763987792472, |
| "grad_norm": 11.54091739654541, |
| "learning_rate": 9.986395137421607e-06, |
| "loss": 0.6652, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.6510681586978637, |
| "grad_norm": 14.71457576751709, |
| "learning_rate": 9.973928157497675e-06, |
| "loss": 0.7393, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.6917599186164801, |
| "grad_norm": 17.259702682495117, |
| "learning_rate": 9.957455510379753e-06, |
| "loss": 0.6728, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.7324516785350966, |
| "grad_norm": 9.521153450012207, |
| "learning_rate": 9.93699046202989e-06, |
| "loss": 0.5985, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.7731434384537131, |
| "grad_norm": 16.928251266479492, |
| "learning_rate": 9.912549493621555e-06, |
| "loss": 0.6867, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.8138351983723296, |
| "grad_norm": 17.1862850189209, |
| "learning_rate": 9.88415228826682e-06, |
| "loss": 0.5762, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.854526958290946, |
| "grad_norm": 21.559711456298828, |
| "learning_rate": 9.851821715164891e-06, |
| "loss": 0.5887, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.8952187182095626, |
| "grad_norm": 13.549447059631348, |
| "learning_rate": 9.815583811184809e-06, |
| "loss": 0.6607, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.9359104781281791, |
| "grad_norm": 11.863809585571289, |
| "learning_rate": 9.775467759897092e-06, |
| "loss": 0.6735, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.9766022380467956, |
| "grad_norm": 16.73202896118164, |
| "learning_rate": 9.731505868071262e-06, |
| "loss": 0.6576, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.0162767039674465, |
| "grad_norm": 13.482553482055664, |
| "learning_rate": 9.68373353965814e-06, |
| "loss": 0.464, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.056968463886063, |
| "grad_norm": 13.539304733276367, |
| "learning_rate": 9.632189247277885e-06, |
| "loss": 0.2511, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.0976602238046795, |
| "grad_norm": 19.008159637451172, |
| "learning_rate": 9.576914501236734e-06, |
| "loss": 0.3155, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.138351983723296, |
| "grad_norm": 9.890944480895996, |
| "learning_rate": 9.517953816097396e-06, |
| "loss": 0.2411, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.1790437436419126, |
| "grad_norm": 12.655500411987305, |
| "learning_rate": 9.455354674830016e-06, |
| "loss": 0.3769, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.219735503560529, |
| "grad_norm": 13.159302711486816, |
| "learning_rate": 9.38916749057258e-06, |
| "loss": 0.2914, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.2604272634791456, |
| "grad_norm": 13.500814437866211, |
| "learning_rate": 9.31944556603157e-06, |
| "loss": 0.3151, |
| "step": 310 |
| }, |
| { |
| "epoch": 1.3011190233977619, |
| "grad_norm": 13.991122245788574, |
| "learning_rate": 9.24624505055555e-06, |
| "loss": 0.2463, |
| "step": 320 |
| }, |
| { |
| "epoch": 1.3418107833163784, |
| "grad_norm": 16.40860939025879, |
| "learning_rate": 9.169624894916252e-06, |
| "loss": 0.4099, |
| "step": 330 |
| }, |
| { |
| "epoch": 1.3825025432349949, |
| "grad_norm": 12.705772399902344, |
| "learning_rate": 9.089646803833589e-06, |
| "loss": 0.3211, |
| "step": 340 |
| }, |
| { |
| "epoch": 1.4231943031536114, |
| "grad_norm": 10.286760330200195, |
| "learning_rate": 9.006375186282832e-06, |
| "loss": 0.3521, |
| "step": 350 |
| }, |
| { |
| "epoch": 1.463886063072228, |
| "grad_norm": 7.008501052856445, |
| "learning_rate": 8.919877103623949e-06, |
| "loss": 0.2947, |
| "step": 360 |
| }, |
| { |
| "epoch": 1.5045778229908442, |
| "grad_norm": 19.253416061401367, |
| "learning_rate": 8.83022221559489e-06, |
| "loss": 0.3433, |
| "step": 370 |
| }, |
| { |
| "epoch": 1.545269582909461, |
| "grad_norm": 10.227717399597168, |
| "learning_rate": 8.737482724212331e-06, |
| "loss": 0.3676, |
| "step": 380 |
| }, |
| { |
| "epoch": 1.5859613428280772, |
| "grad_norm": 10.90295124053955, |
| "learning_rate": 8.641733315625014e-06, |
| "loss": 0.3761, |
| "step": 390 |
| }, |
| { |
| "epoch": 1.626653102746694, |
| "grad_norm": 12.347325325012207, |
| "learning_rate": 8.543051099966558e-06, |
| "loss": 0.3742, |
| "step": 400 |
| }, |
| { |
| "epoch": 1.6673448626653102, |
| "grad_norm": 11.799545288085938, |
| "learning_rate": 8.441515549256134e-06, |
| "loss": 0.3786, |
| "step": 410 |
| }, |
| { |
| "epoch": 1.7080366225839267, |
| "grad_norm": 12.328984260559082, |
| "learning_rate": 8.337208433397051e-06, |
| "loss": 0.3095, |
| "step": 420 |
| }, |
| { |
| "epoch": 1.7487283825025433, |
| "grad_norm": 14.553093910217285, |
| "learning_rate": 8.230213754324773e-06, |
| "loss": 0.309, |
| "step": 430 |
| }, |
| { |
| "epoch": 1.7894201424211598, |
| "grad_norm": 22.428647994995117, |
| "learning_rate": 8.120617678357415e-06, |
| "loss": 0.342, |
| "step": 440 |
| }, |
| { |
| "epoch": 1.8301119023397763, |
| "grad_norm": 15.604973793029785, |
| "learning_rate": 8.00850846680318e-06, |
| "loss": 0.3279, |
| "step": 450 |
| }, |
| { |
| "epoch": 1.8708036622583926, |
| "grad_norm": 12.865342140197754, |
| "learning_rate": 7.893976404880643e-06, |
| "loss": 0.3104, |
| "step": 460 |
| }, |
| { |
| "epoch": 1.9114954221770093, |
| "grad_norm": 8.454442977905273, |
| "learning_rate": 7.7771137290091e-06, |
| "loss": 0.3436, |
| "step": 470 |
| }, |
| { |
| "epoch": 1.9521871820956256, |
| "grad_norm": 11.887304306030273, |
| "learning_rate": 7.658014552527572e-06, |
| "loss": 0.3077, |
| "step": 480 |
| }, |
| { |
| "epoch": 1.992878942014242, |
| "grad_norm": 12.78626823425293, |
| "learning_rate": 7.536774789902246e-06, |
| "loss": 0.3412, |
| "step": 490 |
| }, |
| { |
| "epoch": 2.032553407934893, |
| "grad_norm": 7.850743293762207, |
| "learning_rate": 7.413492079483405e-06, |
| "loss": 0.1631, |
| "step": 500 |
| }, |
| { |
| "epoch": 2.0732451678535098, |
| "grad_norm": 17.043455123901367, |
| "learning_rate": 7.288265704874089e-06, |
| "loss": 0.0888, |
| "step": 510 |
| }, |
| { |
| "epoch": 2.113936927772126, |
| "grad_norm": 11.06017017364502, |
| "learning_rate": 7.161196514973735e-06, |
| "loss": 0.1366, |
| "step": 520 |
| }, |
| { |
| "epoch": 2.154628687690743, |
| "grad_norm": 9.118733406066895, |
| "learning_rate": 7.032386842761282e-06, |
| "loss": 0.0993, |
| "step": 530 |
| }, |
| { |
| "epoch": 2.195320447609359, |
| "grad_norm": 10.076557159423828, |
| "learning_rate": 6.9019404228830465e-06, |
| "loss": 0.1687, |
| "step": 540 |
| }, |
| { |
| "epoch": 2.236012207527976, |
| "grad_norm": 16.989974975585938, |
| "learning_rate": 6.769962308111839e-06, |
| "loss": 0.1505, |
| "step": 550 |
| }, |
| { |
| "epoch": 2.276703967446592, |
| "grad_norm": 6.469771385192871, |
| "learning_rate": 6.636558784744507e-06, |
| "loss": 0.0678, |
| "step": 560 |
| }, |
| { |
| "epoch": 2.3173957273652084, |
| "grad_norm": 5.974874496459961, |
| "learning_rate": 6.501837287006112e-06, |
| "loss": 0.116, |
| "step": 570 |
| }, |
| { |
| "epoch": 2.358087487283825, |
| "grad_norm": 14.892068862915039, |
| "learning_rate": 6.365906310529631e-06, |
| "loss": 0.1232, |
| "step": 580 |
| }, |
| { |
| "epoch": 2.3987792472024414, |
| "grad_norm": 13.642632484436035, |
| "learning_rate": 6.228875324980862e-06, |
| "loss": 0.1505, |
| "step": 590 |
| }, |
| { |
| "epoch": 2.439471007121058, |
| "grad_norm": 7.214609622955322, |
| "learning_rate": 6.090854685898928e-06, |
| "loss": 0.1282, |
| "step": 600 |
| }, |
| { |
| "epoch": 2.4801627670396744, |
| "grad_norm": 15.701526641845703, |
| "learning_rate": 5.951955545823342e-06, |
| "loss": 0.1105, |
| "step": 610 |
| }, |
| { |
| "epoch": 2.520854526958291, |
| "grad_norm": 14.56447696685791, |
| "learning_rate": 5.812289764779232e-06, |
| "loss": 0.1085, |
| "step": 620 |
| }, |
| { |
| "epoch": 2.5615462868769074, |
| "grad_norm": 8.728901863098145, |
| "learning_rate": 5.671969820192794e-06, |
| "loss": 0.1116, |
| "step": 630 |
| }, |
| { |
| "epoch": 2.6022380467955237, |
| "grad_norm": 12.29587459564209, |
| "learning_rate": 5.5311087163095475e-06, |
| "loss": 0.0912, |
| "step": 640 |
| }, |
| { |
| "epoch": 2.6429298067141405, |
| "grad_norm": 21.944841384887695, |
| "learning_rate": 5.389819893188304e-06, |
| "loss": 0.1691, |
| "step": 650 |
| }, |
| { |
| "epoch": 2.6836215666327567, |
| "grad_norm": 10.609075546264648, |
| "learning_rate": 5.248217135344191e-06, |
| "loss": 0.0925, |
| "step": 660 |
| }, |
| { |
| "epoch": 2.7243133265513735, |
| "grad_norm": 13.140278816223145, |
| "learning_rate": 5.106414480114238e-06, |
| "loss": 0.1029, |
| "step": 670 |
| }, |
| { |
| "epoch": 2.7650050864699898, |
| "grad_norm": 12.165359497070312, |
| "learning_rate": 4.96452612581939e-06, |
| "loss": 0.0797, |
| "step": 680 |
| }, |
| { |
| "epoch": 2.8056968463886065, |
| "grad_norm": 10.018705368041992, |
| "learning_rate": 4.82266633979685e-06, |
| "loss": 0.1326, |
| "step": 690 |
| }, |
| { |
| "epoch": 2.846388606307223, |
| "grad_norm": 6.573582172393799, |
| "learning_rate": 4.680949366376858e-06, |
| "loss": 0.1534, |
| "step": 700 |
| }, |
| { |
| "epoch": 2.887080366225839, |
| "grad_norm": 15.606719970703125, |
| "learning_rate": 4.539489334877992e-06, |
| "loss": 0.1322, |
| "step": 710 |
| }, |
| { |
| "epoch": 2.927772126144456, |
| "grad_norm": 12.944083213806152, |
| "learning_rate": 4.3984001676950875e-06, |
| "loss": 0.1422, |
| "step": 720 |
| }, |
| { |
| "epoch": 2.968463886063072, |
| "grad_norm": 10.922980308532715, |
| "learning_rate": 4.2577954885537985e-06, |
| "loss": 0.1413, |
| "step": 730 |
| }, |
| { |
| "epoch": 3.0081383519837233, |
| "grad_norm": 5.96065092086792, |
| "learning_rate": 4.1177885310057045e-06, |
| "loss": 0.0939, |
| "step": 740 |
| }, |
| { |
| "epoch": 3.04883011190234, |
| "grad_norm": 9.311906814575195, |
| "learning_rate": 3.978492047237608e-06, |
| "loss": 0.0392, |
| "step": 750 |
| }, |
| { |
| "epoch": 3.0895218718209563, |
| "grad_norm": 4.063051700592041, |
| "learning_rate": 3.840018217268527e-06, |
| "loss": 0.0306, |
| "step": 760 |
| }, |
| { |
| "epoch": 3.1302136317395726, |
| "grad_norm": 8.086892127990723, |
| "learning_rate": 3.702478558607429e-06, |
| "loss": 0.0329, |
| "step": 770 |
| }, |
| { |
| "epoch": 3.1709053916581893, |
| "grad_norm": 6.210553169250488, |
| "learning_rate": 3.5659838364445505e-06, |
| "loss": 0.0363, |
| "step": 780 |
| }, |
| { |
| "epoch": 3.2115971515768056, |
| "grad_norm": 5.711456298828125, |
| "learning_rate": 3.4306439744485453e-06, |
| "loss": 0.0465, |
| "step": 790 |
| }, |
| { |
| "epoch": 3.2522889114954223, |
| "grad_norm": 10.935355186462402, |
| "learning_rate": 3.2965679662413772e-06, |
| "loss": 0.0437, |
| "step": 800 |
| }, |
| { |
| "epoch": 3.2929806714140386, |
| "grad_norm": 1.5802483558654785, |
| "learning_rate": 3.163863787622162e-06, |
| "loss": 0.03, |
| "step": 810 |
| }, |
| { |
| "epoch": 3.333672431332655, |
| "grad_norm": 15.58889389038086, |
| "learning_rate": 3.0326383096107424e-06, |
| "loss": 0.0425, |
| "step": 820 |
| }, |
| { |
| "epoch": 3.3743641912512716, |
| "grad_norm": 5.607379913330078, |
| "learning_rate": 2.9029972123809425e-06, |
| "loss": 0.0338, |
| "step": 830 |
| }, |
| { |
| "epoch": 3.415055951169888, |
| "grad_norm": 2.1293985843658447, |
| "learning_rate": 2.775044900152873e-06, |
| "loss": 0.0469, |
| "step": 840 |
| }, |
| { |
| "epoch": 3.4557477110885046, |
| "grad_norm": 5.509940147399902, |
| "learning_rate": 2.6488844171127903e-06, |
| "loss": 0.0349, |
| "step": 850 |
| }, |
| { |
| "epoch": 3.496439471007121, |
| "grad_norm": 1.3028367757797241, |
| "learning_rate": 2.52461736442824e-06, |
| "loss": 0.0416, |
| "step": 860 |
| }, |
| { |
| "epoch": 3.5371312309257377, |
| "grad_norm": 4.432685852050781, |
| "learning_rate": 2.4023438184253115e-06, |
| "loss": 0.0294, |
| "step": 870 |
| }, |
| { |
| "epoch": 3.577822990844354, |
| "grad_norm": 10.347676277160645, |
| "learning_rate": 2.282162249993895e-06, |
| "loss": 0.0334, |
| "step": 880 |
| }, |
| { |
| "epoch": 3.6185147507629702, |
| "grad_norm": 8.725708961486816, |
| "learning_rate": 2.1641694452858486e-06, |
| "loss": 0.0331, |
| "step": 890 |
| }, |
| { |
| "epoch": 3.659206510681587, |
| "grad_norm": 2.433056354522705, |
| "learning_rate": 2.0484604277699437e-06, |
| "loss": 0.04, |
| "step": 900 |
| }, |
| { |
| "epoch": 3.6998982706002037, |
| "grad_norm": 4.625722885131836, |
| "learning_rate": 1.935128381706355e-06, |
| "loss": 0.0387, |
| "step": 910 |
| }, |
| { |
| "epoch": 3.74059003051882, |
| "grad_norm": 4.8643083572387695, |
| "learning_rate": 1.8242645771023205e-06, |
| "loss": 0.0436, |
| "step": 920 |
| }, |
| { |
| "epoch": 3.7812817904374363, |
| "grad_norm": 5.704628944396973, |
| "learning_rate": 1.7159582962094224e-06, |
| "loss": 0.0348, |
| "step": 930 |
| }, |
| { |
| "epoch": 3.821973550356053, |
| "grad_norm": 4.647428512573242, |
| "learning_rate": 1.610296761621662e-06, |
| "loss": 0.0273, |
| "step": 940 |
| }, |
| { |
| "epoch": 3.8626653102746693, |
| "grad_norm": 3.7586710453033447, |
| "learning_rate": 1.5073650660322509e-06, |
| "loss": 0.0387, |
| "step": 950 |
| }, |
| { |
| "epoch": 3.9033570701932856, |
| "grad_norm": 6.614203929901123, |
| "learning_rate": 1.4072461037056806e-06, |
| "loss": 0.034, |
| "step": 960 |
| }, |
| { |
| "epoch": 3.9440488301119023, |
| "grad_norm": 2.0238308906555176, |
| "learning_rate": 1.310020503720254e-06, |
| "loss": 0.0572, |
| "step": 970 |
| }, |
| { |
| "epoch": 3.984740590030519, |
| "grad_norm": 1.2515283823013306, |
| "learning_rate": 1.2157665650348516e-06, |
| "loss": 0.0425, |
| "step": 980 |
| }, |
| { |
| "epoch": 4.02441505595117, |
| "grad_norm": 1.190735936164856, |
| "learning_rate": 1.1245601934322148e-06, |
| "loss": 0.0422, |
| "step": 990 |
| }, |
| { |
| "epoch": 4.065106815869786, |
| "grad_norm": 4.267083644866943, |
| "learning_rate": 1.0364748403895368e-06, |
| "loss": 0.0177, |
| "step": 1000 |
| }, |
| { |
| "epoch": 4.105798575788403, |
| "grad_norm": 1.938965916633606, |
| "learning_rate": 9.51581443925576e-07, |
| "loss": 0.0088, |
| "step": 1010 |
| }, |
| { |
| "epoch": 4.1464903357070195, |
| "grad_norm": 2.2659451961517334, |
| "learning_rate": 8.699483714719547e-07, |
| "loss": 0.0073, |
| "step": 1020 |
| }, |
| { |
| "epoch": 4.187182095625635, |
| "grad_norm": 0.5305855870246887, |
| "learning_rate": 7.916413648146282e-07, |
| "loss": 0.0065, |
| "step": 1030 |
| }, |
| { |
| "epoch": 4.227873855544252, |
| "grad_norm": 1.4761003255844116, |
| "learning_rate": 7.167234871498646e-07, |
| "loss": 0.0071, |
| "step": 1040 |
| }, |
| { |
| "epoch": 4.268565615462869, |
| "grad_norm": 0.35723185539245605, |
| "learning_rate": 6.452550722973927e-07, |
| "loss": 0.0079, |
| "step": 1050 |
| }, |
| { |
| "epoch": 4.309257375381486, |
| "grad_norm": 0.4571964144706726, |
| "learning_rate": 5.772936761116027e-07, |
| "loss": 0.0073, |
| "step": 1060 |
| }, |
| { |
| "epoch": 4.349949135300101, |
| "grad_norm": 0.5277753472328186, |
| "learning_rate": 5.128940301299334e-07, |
| "loss": 0.0043, |
| "step": 1070 |
| }, |
| { |
| "epoch": 4.390640895218718, |
| "grad_norm": 2.183068037033081, |
| "learning_rate": 4.5210799749576815e-07, |
| "loss": 0.0076, |
| "step": 1080 |
| }, |
| { |
| "epoch": 4.431332655137335, |
| "grad_norm": 3.6342573165893555, |
| "learning_rate": 3.9498453119134917e-07, |
| "loss": 0.014, |
| "step": 1090 |
| }, |
| { |
| "epoch": 4.472024415055952, |
| "grad_norm": 1.6003706455230713, |
| "learning_rate": 3.4156963461434156e-07, |
| "loss": 0.009, |
| "step": 1100 |
| }, |
| { |
| "epoch": 4.5127161749745675, |
| "grad_norm": 2.1982743740081787, |
| "learning_rate": 2.9190632452978706e-07, |
| "loss": 0.0187, |
| "step": 1110 |
| }, |
| { |
| "epoch": 4.553407934893184, |
| "grad_norm": 1.6797971725463867, |
| "learning_rate": 2.4603459642729867e-07, |
| "loss": 0.0039, |
| "step": 1120 |
| }, |
| { |
| "epoch": 4.594099694811801, |
| "grad_norm": 0.580377459526062, |
| "learning_rate": 2.0399139231137731e-07, |
| "loss": 0.0134, |
| "step": 1130 |
| }, |
| { |
| "epoch": 4.634791454730417, |
| "grad_norm": 0.8677375912666321, |
| "learning_rate": 1.6581057095081288e-07, |
| "loss": 0.0047, |
| "step": 1140 |
| }, |
| { |
| "epoch": 4.6754832146490335, |
| "grad_norm": 8.13941478729248, |
| "learning_rate": 1.3152288061110518e-07, |
| "loss": 0.0078, |
| "step": 1150 |
| }, |
| { |
| "epoch": 4.71617497456765, |
| "grad_norm": 14.699440002441406, |
| "learning_rate": 1.0115593429187942e-07, |
| "loss": 0.0109, |
| "step": 1160 |
| }, |
| { |
| "epoch": 4.756866734486266, |
| "grad_norm": 0.6142909526824951, |
| "learning_rate": 7.473418748923545e-08, |
| "loss": 0.0091, |
| "step": 1170 |
| }, |
| { |
| "epoch": 4.797558494404883, |
| "grad_norm": 9.127274513244629, |
| "learning_rate": 5.227891850093314e-08, |
| "loss": 0.0103, |
| "step": 1180 |
| }, |
| { |
| "epoch": 4.8382502543234995, |
| "grad_norm": 0.5334455966949463, |
| "learning_rate": 3.3808211290284886e-08, |
| "loss": 0.0097, |
| "step": 1190 |
| }, |
| { |
| "epoch": 4.878942014242116, |
| "grad_norm": 1.6610623598098755, |
| "learning_rate": 1.933694092254801e-08, |
| "loss": 0.0149, |
| "step": 1200 |
| }, |
| { |
| "epoch": 4.919633774160732, |
| "grad_norm": 1.1210153102874756, |
| "learning_rate": 8.87676158554507e-09, |
| "loss": 0.0048, |
| "step": 1210 |
| }, |
| { |
| "epoch": 4.960325534079349, |
| "grad_norm": 0.8069121241569519, |
| "learning_rate": 2.436097204167043e-09, |
| "loss": 0.0045, |
| "step": 1220 |
| }, |
| { |
| "epoch": 5.0, |
| "grad_norm": 1.9714659452438354, |
| "learning_rate": 2.0134656311676658e-11, |
| "loss": 0.0076, |
| "step": 1230 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1230, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 5, |
| "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": 1.553356054069248e+16, |
| "train_batch_size": 1, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
