| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.40210627094303497, | |
| "eval_steps": 500, | |
| "global_step": 210, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0019147917663954045, | |
| "grad_norm": 1.153105616569519, | |
| "learning_rate": 0.0, | |
| "loss": 0.5501, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.003829583532790809, | |
| "grad_norm": 1.2741618156433105, | |
| "learning_rate": 3.846153846153847e-07, | |
| "loss": 0.5444, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0057443752991862135, | |
| "grad_norm": 1.3612818717956543, | |
| "learning_rate": 7.692307692307694e-07, | |
| "loss": 0.5838, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.007659167065581618, | |
| "grad_norm": 1.2917627096176147, | |
| "learning_rate": 1.153846153846154e-06, | |
| "loss": 0.5571, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.009573958831977022, | |
| "grad_norm": 1.1222703456878662, | |
| "learning_rate": 1.5384615384615387e-06, | |
| "loss": 0.494, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.011488750598372427, | |
| "grad_norm": 1.397050142288208, | |
| "learning_rate": 1.9230769230769234e-06, | |
| "loss": 0.531, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.013403542364767831, | |
| "grad_norm": 1.0376505851745605, | |
| "learning_rate": 2.307692307692308e-06, | |
| "loss": 0.5107, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.015318334131163236, | |
| "grad_norm": 1.1001347303390503, | |
| "learning_rate": 2.6923076923076923e-06, | |
| "loss": 0.513, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01723312589755864, | |
| "grad_norm": 0.9904836416244507, | |
| "learning_rate": 3.0769230769230774e-06, | |
| "loss": 0.5334, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.019147917663954045, | |
| "grad_norm": 1.0163137912750244, | |
| "learning_rate": 3.4615384615384617e-06, | |
| "loss": 0.5075, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.02106270943034945, | |
| "grad_norm": 0.9025917053222656, | |
| "learning_rate": 3.846153846153847e-06, | |
| "loss": 0.4907, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.022977501196744854, | |
| "grad_norm": 0.851783037185669, | |
| "learning_rate": 4.230769230769231e-06, | |
| "loss": 0.4954, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.02489229296314026, | |
| "grad_norm": 0.8040263056755066, | |
| "learning_rate": 4.615384615384616e-06, | |
| "loss": 0.5286, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.026807084729535663, | |
| "grad_norm": 0.8646761775016785, | |
| "learning_rate": 5e-06, | |
| "loss": 0.4753, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.028721876495931067, | |
| "grad_norm": 0.7733232975006104, | |
| "learning_rate": 5.384615384615385e-06, | |
| "loss": 0.4566, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.030636668262326472, | |
| "grad_norm": 0.8128564953804016, | |
| "learning_rate": 5.769230769230769e-06, | |
| "loss": 0.4495, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.032551460028721876, | |
| "grad_norm": 0.6713640093803406, | |
| "learning_rate": 6.153846153846155e-06, | |
| "loss": 0.4086, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.03446625179511728, | |
| "grad_norm": 0.7621421813964844, | |
| "learning_rate": 6.538461538461539e-06, | |
| "loss": 0.451, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.036381043561512685, | |
| "grad_norm": 0.7471394538879395, | |
| "learning_rate": 6.923076923076923e-06, | |
| "loss": 0.4663, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.03829583532790809, | |
| "grad_norm": 0.5847220420837402, | |
| "learning_rate": 7.307692307692308e-06, | |
| "loss": 0.4433, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.040210627094303494, | |
| "grad_norm": 0.5838068127632141, | |
| "learning_rate": 7.692307692307694e-06, | |
| "loss": 0.4329, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.0421254188606989, | |
| "grad_norm": 0.548230767250061, | |
| "learning_rate": 8.076923076923077e-06, | |
| "loss": 0.4254, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.0440402106270943, | |
| "grad_norm": 0.6799401044845581, | |
| "learning_rate": 8.461538461538462e-06, | |
| "loss": 0.4389, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.04595500239348971, | |
| "grad_norm": 0.5397069454193115, | |
| "learning_rate": 8.846153846153847e-06, | |
| "loss": 0.4059, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.04786979415988511, | |
| "grad_norm": 0.49979251623153687, | |
| "learning_rate": 9.230769230769232e-06, | |
| "loss": 0.4238, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.04978458592628052, | |
| "grad_norm": 0.6305129528045654, | |
| "learning_rate": 9.615384615384616e-06, | |
| "loss": 0.4301, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.05169937769267592, | |
| "grad_norm": 0.4193238317966461, | |
| "learning_rate": 1e-05, | |
| "loss": 0.3991, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.053614169459071326, | |
| "grad_norm": 0.5012926459312439, | |
| "learning_rate": 9.999899706000774e-06, | |
| "loss": 0.434, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.05552896122546673, | |
| "grad_norm": 0.5148448944091797, | |
| "learning_rate": 9.999598828026644e-06, | |
| "loss": 0.4256, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.057443752991862135, | |
| "grad_norm": 0.45894017815589905, | |
| "learning_rate": 9.999097378148116e-06, | |
| "loss": 0.3914, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.05935854475825754, | |
| "grad_norm": 0.418771892786026, | |
| "learning_rate": 9.998395376482152e-06, | |
| "loss": 0.3818, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.061273336524652944, | |
| "grad_norm": 0.4304119348526001, | |
| "learning_rate": 9.99749285119138e-06, | |
| "loss": 0.3766, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.06318812829104835, | |
| "grad_norm": 0.4879925549030304, | |
| "learning_rate": 9.996389838482942e-06, | |
| "loss": 0.4535, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.06510292005744375, | |
| "grad_norm": 0.4304628372192383, | |
| "learning_rate": 9.995086382607064e-06, | |
| "loss": 0.4199, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.06701771182383916, | |
| "grad_norm": 0.44849488139152527, | |
| "learning_rate": 9.993582535855265e-06, | |
| "loss": 0.3922, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.06893250359023456, | |
| "grad_norm": 0.4242229163646698, | |
| "learning_rate": 9.991878358558267e-06, | |
| "loss": 0.4114, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.07084729535662997, | |
| "grad_norm": 0.42097875475883484, | |
| "learning_rate": 9.989973919083576e-06, | |
| "loss": 0.4003, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.07276208712302537, | |
| "grad_norm": 0.4065902531147003, | |
| "learning_rate": 9.987869293832727e-06, | |
| "loss": 0.4024, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.07467687888942078, | |
| "grad_norm": 0.36274516582489014, | |
| "learning_rate": 9.985564567238237e-06, | |
| "loss": 0.3836, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.07659167065581618, | |
| "grad_norm": 0.45347559452056885, | |
| "learning_rate": 9.983059831760205e-06, | |
| "loss": 0.4214, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.07850646242221158, | |
| "grad_norm": 0.3551877439022064, | |
| "learning_rate": 9.980355187882606e-06, | |
| "loss": 0.379, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.08042125418860699, | |
| "grad_norm": 0.3403424620628357, | |
| "learning_rate": 9.977450744109258e-06, | |
| "loss": 0.3572, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.0823360459550024, | |
| "grad_norm": 0.349159836769104, | |
| "learning_rate": 9.974346616959476e-06, | |
| "loss": 0.3737, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.0842508377213978, | |
| "grad_norm": 0.37900882959365845, | |
| "learning_rate": 9.97104293096339e-06, | |
| "loss": 0.3973, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.0861656294877932, | |
| "grad_norm": 0.3966723084449768, | |
| "learning_rate": 9.967539818656953e-06, | |
| "loss": 0.3876, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.0880804212541886, | |
| "grad_norm": 0.3991048336029053, | |
| "learning_rate": 9.96383742057662e-06, | |
| "loss": 0.4363, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.08999521302058401, | |
| "grad_norm": 0.37847766280174255, | |
| "learning_rate": 9.959935885253715e-06, | |
| "loss": 0.385, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.09191000478697942, | |
| "grad_norm": 0.3453079164028168, | |
| "learning_rate": 9.955835369208475e-06, | |
| "loss": 0.3616, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.09382479655337482, | |
| "grad_norm": 0.33624762296676636, | |
| "learning_rate": 9.951536036943753e-06, | |
| "loss": 0.3595, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.09573958831977022, | |
| "grad_norm": 0.3701552450656891, | |
| "learning_rate": 9.94703806093845e-06, | |
| "loss": 0.3888, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.09765438008616563, | |
| "grad_norm": 0.35528630018234253, | |
| "learning_rate": 9.942341621640558e-06, | |
| "loss": 0.3672, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.09956917185256103, | |
| "grad_norm": 0.36010798811912537, | |
| "learning_rate": 9.937446907459954e-06, | |
| "loss": 0.3839, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.10148396361895644, | |
| "grad_norm": 0.3379530906677246, | |
| "learning_rate": 9.932354114760819e-06, | |
| "loss": 0.3606, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.10339875538535184, | |
| "grad_norm": 0.4284995496273041, | |
| "learning_rate": 9.92706344785377e-06, | |
| "loss": 0.4539, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.10531354715174725, | |
| "grad_norm": 0.3274945616722107, | |
| "learning_rate": 9.921575118987672e-06, | |
| "loss": 0.3436, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.10722833891814265, | |
| "grad_norm": 0.34816086292266846, | |
| "learning_rate": 9.915889348341098e-06, | |
| "loss": 0.3702, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.10914313068453806, | |
| "grad_norm": 0.3399084508419037, | |
| "learning_rate": 9.910006364013522e-06, | |
| "loss": 0.3404, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.11105792245093346, | |
| "grad_norm": 0.35537949204444885, | |
| "learning_rate": 9.903926402016153e-06, | |
| "loss": 0.3564, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.11297271421732887, | |
| "grad_norm": 0.3834598958492279, | |
| "learning_rate": 9.897649706262474e-06, | |
| "loss": 0.4011, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.11488750598372427, | |
| "grad_norm": 0.3632608950138092, | |
| "learning_rate": 9.891176528558451e-06, | |
| "loss": 0.3659, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.11680229775011967, | |
| "grad_norm": 0.359200119972229, | |
| "learning_rate": 9.884507128592435e-06, | |
| "loss": 0.3838, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.11871708951651508, | |
| "grad_norm": 0.33314335346221924, | |
| "learning_rate": 9.877641773924748e-06, | |
| "loss": 0.3539, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.12063188128291048, | |
| "grad_norm": 0.36184924840927124, | |
| "learning_rate": 9.870580739976936e-06, | |
| "loss": 0.352, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.12254667304930589, | |
| "grad_norm": 0.3508659899234772, | |
| "learning_rate": 9.863324310020735e-06, | |
| "loss": 0.3598, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.12446146481570129, | |
| "grad_norm": 0.3793911039829254, | |
| "learning_rate": 9.855872775166696e-06, | |
| "loss": 0.3911, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.1263762565820967, | |
| "grad_norm": 0.35408976674079895, | |
| "learning_rate": 9.848226434352513e-06, | |
| "loss": 0.3641, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.12829104834849211, | |
| "grad_norm": 0.3811071515083313, | |
| "learning_rate": 9.840385594331022e-06, | |
| "loss": 0.3958, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.1302058401148875, | |
| "grad_norm": 0.3835104703903198, | |
| "learning_rate": 9.83235056965791e-06, | |
| "loss": 0.4128, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.13212063188128292, | |
| "grad_norm": 0.3702054023742676, | |
| "learning_rate": 9.824121682679072e-06, | |
| "loss": 0.3629, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.13403542364767831, | |
| "grad_norm": 0.3421335220336914, | |
| "learning_rate": 9.815699263517712e-06, | |
| "loss": 0.3698, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.13595021541407373, | |
| "grad_norm": 0.35186541080474854, | |
| "learning_rate": 9.807083650061063e-06, | |
| "loss": 0.3791, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.13786500718046912, | |
| "grad_norm": 0.35440999269485474, | |
| "learning_rate": 9.798275187946859e-06, | |
| "loss": 0.3663, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.13977979894686454, | |
| "grad_norm": 0.3255431354045868, | |
| "learning_rate": 9.789274230549456e-06, | |
| "loss": 0.3342, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.14169459071325993, | |
| "grad_norm": 0.3124358355998993, | |
| "learning_rate": 9.780081138965663e-06, | |
| "loss": 0.3152, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.14360938247965535, | |
| "grad_norm": 0.37060725688934326, | |
| "learning_rate": 9.770696282000245e-06, | |
| "loss": 0.3773, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.14552417424605074, | |
| "grad_norm": 0.3520626723766327, | |
| "learning_rate": 9.761120036151138e-06, | |
| "loss": 0.3498, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.14743896601244616, | |
| "grad_norm": 0.3260503113269806, | |
| "learning_rate": 9.751352785594337e-06, | |
| "loss": 0.3294, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.14935375777884155, | |
| "grad_norm": 0.3873290717601776, | |
| "learning_rate": 9.741394922168495e-06, | |
| "loss": 0.4018, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.15126854954523697, | |
| "grad_norm": 0.3302907943725586, | |
| "learning_rate": 9.731246845359187e-06, | |
| "loss": 0.3394, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.15318334131163236, | |
| "grad_norm": 0.33078426122665405, | |
| "learning_rate": 9.720908962282893e-06, | |
| "loss": 0.3384, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.15509813307802778, | |
| "grad_norm": 0.3441630005836487, | |
| "learning_rate": 9.710381687670675e-06, | |
| "loss": 0.3601, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.15701292484442317, | |
| "grad_norm": 0.36827200651168823, | |
| "learning_rate": 9.699665443851518e-06, | |
| "loss": 0.3692, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.1589277166108186, | |
| "grad_norm": 0.3212028741836548, | |
| "learning_rate": 9.688760660735403e-06, | |
| "loss": 0.3325, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.16084250837721398, | |
| "grad_norm": 0.32657769322395325, | |
| "learning_rate": 9.677667775796052e-06, | |
| "loss": 0.3324, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.1627573001436094, | |
| "grad_norm": 0.35298746824264526, | |
| "learning_rate": 9.666387234053385e-06, | |
| "loss": 0.3716, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.1646720919100048, | |
| "grad_norm": 0.3585253357887268, | |
| "learning_rate": 9.654919488055656e-06, | |
| "loss": 0.3646, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.1665868836764002, | |
| "grad_norm": 0.34902405738830566, | |
| "learning_rate": 9.643264997861312e-06, | |
| "loss": 0.3424, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.1685016754427956, | |
| "grad_norm": 0.3480973243713379, | |
| "learning_rate": 9.631424231020523e-06, | |
| "loss": 0.3627, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.170416467209191, | |
| "grad_norm": 0.3816223442554474, | |
| "learning_rate": 9.619397662556434e-06, | |
| "loss": 0.374, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1723312589755864, | |
| "grad_norm": 0.41936248540878296, | |
| "learning_rate": 9.607185774946106e-06, | |
| "loss": 0.4321, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.17424605074198182, | |
| "grad_norm": 0.38773012161254883, | |
| "learning_rate": 9.594789058101154e-06, | |
| "loss": 0.3679, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.1761608425083772, | |
| "grad_norm": 0.3591976761817932, | |
| "learning_rate": 9.582208009348104e-06, | |
| "loss": 0.3391, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.17807563427477263, | |
| "grad_norm": 0.38234034180641174, | |
| "learning_rate": 9.569443133408434e-06, | |
| "loss": 0.3795, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.17999042604116802, | |
| "grad_norm": 0.34880128502845764, | |
| "learning_rate": 9.556494942378328e-06, | |
| "loss": 0.3641, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.18190521780756344, | |
| "grad_norm": 0.37424835562705994, | |
| "learning_rate": 9.543363955708124e-06, | |
| "loss": 0.3632, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.18382000957395883, | |
| "grad_norm": 0.3154820203781128, | |
| "learning_rate": 9.530050700181499e-06, | |
| "loss": 0.3049, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.18573480134035425, | |
| "grad_norm": 0.3233638107776642, | |
| "learning_rate": 9.5165557098943e-06, | |
| "loss": 0.3424, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.18764959310674964, | |
| "grad_norm": 0.3317395746707916, | |
| "learning_rate": 9.502879526233151e-06, | |
| "loss": 0.3302, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.18956438487314506, | |
| "grad_norm": 0.33342453837394714, | |
| "learning_rate": 9.48902269785371e-06, | |
| "loss": 0.3493, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.19147917663954045, | |
| "grad_norm": 0.3376815915107727, | |
| "learning_rate": 9.47498578065867e-06, | |
| "loss": 0.3374, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.19339396840593587, | |
| "grad_norm": 0.33345019817352295, | |
| "learning_rate": 9.460769337775461e-06, | |
| "loss": 0.3545, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.19530876017233126, | |
| "grad_norm": 0.3201693594455719, | |
| "learning_rate": 9.446373939533642e-06, | |
| "loss": 0.3247, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.19722355193872668, | |
| "grad_norm": 0.34736594557762146, | |
| "learning_rate": 9.431800163442043e-06, | |
| "loss": 0.3432, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.19913834370512207, | |
| "grad_norm": 0.34281763434410095, | |
| "learning_rate": 9.417048594165572e-06, | |
| "loss": 0.3728, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.20105313547151749, | |
| "grad_norm": 0.36040249466896057, | |
| "learning_rate": 9.402119823501787e-06, | |
| "loss": 0.3629, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.20296792723791288, | |
| "grad_norm": 0.35351505875587463, | |
| "learning_rate": 9.387014450357128e-06, | |
| "loss": 0.345, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.2048827190043083, | |
| "grad_norm": 0.3892531991004944, | |
| "learning_rate": 9.371733080722911e-06, | |
| "loss": 0.3921, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.20679751077070369, | |
| "grad_norm": 0.36659398674964905, | |
| "learning_rate": 9.356276327651006e-06, | |
| "loss": 0.3645, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.2087123025370991, | |
| "grad_norm": 0.3415011763572693, | |
| "learning_rate": 9.340644811229243e-06, | |
| "loss": 0.333, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.2106270943034945, | |
| "grad_norm": 0.3052555322647095, | |
| "learning_rate": 9.324839158556542e-06, | |
| "loss": 0.3087, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.2125418860698899, | |
| "grad_norm": 0.3534447252750397, | |
| "learning_rate": 9.308860003717748e-06, | |
| "loss": 0.3516, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.2144566778362853, | |
| "grad_norm": 0.3732927143573761, | |
| "learning_rate": 9.292707987758202e-06, | |
| "loss": 0.3647, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.21637146960268072, | |
| "grad_norm": 0.37197062373161316, | |
| "learning_rate": 9.27638375865801e-06, | |
| "loss": 0.3606, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.2182862613690761, | |
| "grad_norm": 0.3826271593570709, | |
| "learning_rate": 9.259887971306064e-06, | |
| "loss": 0.4033, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.22020105313547153, | |
| "grad_norm": 0.38200679421424866, | |
| "learning_rate": 9.243221287473755e-06, | |
| "loss": 0.3698, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.22211584490186692, | |
| "grad_norm": 0.3601812720298767, | |
| "learning_rate": 9.226384375788435e-06, | |
| "loss": 0.3462, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.22403063666826234, | |
| "grad_norm": 0.341447114944458, | |
| "learning_rate": 9.209377911706585e-06, | |
| "loss": 0.3348, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.22594542843465773, | |
| "grad_norm": 0.34483444690704346, | |
| "learning_rate": 9.192202577486725e-06, | |
| "loss": 0.3198, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.22786022020105315, | |
| "grad_norm": 0.34377017617225647, | |
| "learning_rate": 9.174859062162037e-06, | |
| "loss": 0.3286, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.22977501196744854, | |
| "grad_norm": 0.3510438799858093, | |
| "learning_rate": 9.157348061512728e-06, | |
| "loss": 0.3626, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.23168980373384396, | |
| "grad_norm": 0.37636810541152954, | |
| "learning_rate": 9.139670278038109e-06, | |
| "loss": 0.37, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.23360459550023935, | |
| "grad_norm": 0.3157687485218048, | |
| "learning_rate": 9.121826420928421e-06, | |
| "loss": 0.3043, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.23551938726663477, | |
| "grad_norm": 0.3494786322116852, | |
| "learning_rate": 9.103817206036383e-06, | |
| "loss": 0.3325, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.23743417903303016, | |
| "grad_norm": 0.3598034083843231, | |
| "learning_rate": 9.085643355848468e-06, | |
| "loss": 0.3652, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.23934897079942558, | |
| "grad_norm": 0.34662720561027527, | |
| "learning_rate": 9.06730559945592e-06, | |
| "loss": 0.3261, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.24126376256582097, | |
| "grad_norm": 0.34236112236976624, | |
| "learning_rate": 9.048804672525513e-06, | |
| "loss": 0.3418, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.24317855433221638, | |
| "grad_norm": 0.33544981479644775, | |
| "learning_rate": 9.030141317270026e-06, | |
| "loss": 0.3259, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.24509334609861178, | |
| "grad_norm": 0.3610815107822418, | |
| "learning_rate": 9.011316282418474e-06, | |
| "loss": 0.3399, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.2470081378650072, | |
| "grad_norm": 0.3822128176689148, | |
| "learning_rate": 8.992330323186069e-06, | |
| "loss": 0.3667, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.24892292963140258, | |
| "grad_norm": 0.3826322853565216, | |
| "learning_rate": 8.973184201243922e-06, | |
| "loss": 0.4064, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.250837721397798, | |
| "grad_norm": 0.3710707724094391, | |
| "learning_rate": 8.953878684688492e-06, | |
| "loss": 0.3503, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.2527525131641934, | |
| "grad_norm": 0.3488466143608093, | |
| "learning_rate": 8.934414548010764e-06, | |
| "loss": 0.3351, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.2546673049305888, | |
| "grad_norm": 0.35479050874710083, | |
| "learning_rate": 8.914792572065178e-06, | |
| "loss": 0.3376, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.25658209669698423, | |
| "grad_norm": 0.39882269501686096, | |
| "learning_rate": 8.89501354403831e-06, | |
| "loss": 0.3814, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.2584968884633796, | |
| "grad_norm": 0.38730108737945557, | |
| "learning_rate": 8.875078257417294e-06, | |
| "loss": 0.3824, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.260411680229775, | |
| "grad_norm": 0.35698166489601135, | |
| "learning_rate": 8.854987511957974e-06, | |
| "loss": 0.3599, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.26232647199617043, | |
| "grad_norm": 0.3598966598510742, | |
| "learning_rate": 8.834742113652835e-06, | |
| "loss": 0.353, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.26424126376256585, | |
| "grad_norm": 0.3241833448410034, | |
| "learning_rate": 8.81434287469866e-06, | |
| "loss": 0.3076, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.2661560555289612, | |
| "grad_norm": 0.33494773507118225, | |
| "learning_rate": 8.793790613463956e-06, | |
| "loss": 0.3325, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.26807084729535663, | |
| "grad_norm": 0.341126412153244, | |
| "learning_rate": 8.773086154456106e-06, | |
| "loss": 0.3425, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.26998563906175205, | |
| "grad_norm": 0.3854438364505768, | |
| "learning_rate": 8.752230328288314e-06, | |
| "loss": 0.3902, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.27190043082814747, | |
| "grad_norm": 0.3441813290119171, | |
| "learning_rate": 8.731223971646261e-06, | |
| "loss": 0.3196, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.27381522259454283, | |
| "grad_norm": 0.32438111305236816, | |
| "learning_rate": 8.710067927254555e-06, | |
| "loss": 0.3152, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.27573001436093825, | |
| "grad_norm": 0.34742826223373413, | |
| "learning_rate": 8.688763043842916e-06, | |
| "loss": 0.3312, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.27764480612733367, | |
| "grad_norm": 0.3526897132396698, | |
| "learning_rate": 8.66731017611213e-06, | |
| "loss": 0.3591, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.2795595978937291, | |
| "grad_norm": 0.3407156467437744, | |
| "learning_rate": 8.645710184699756e-06, | |
| "loss": 0.3501, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.28147438966012445, | |
| "grad_norm": 0.3403429388999939, | |
| "learning_rate": 8.6239639361456e-06, | |
| "loss": 0.3367, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.28338918142651986, | |
| "grad_norm": 0.32284489274024963, | |
| "learning_rate": 8.602072302856961e-06, | |
| "loss": 0.3362, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.2853039731929153, | |
| "grad_norm": 0.35054418444633484, | |
| "learning_rate": 8.580036163073615e-06, | |
| "loss": 0.3556, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.2872187649593107, | |
| "grad_norm": 0.3371654152870178, | |
| "learning_rate": 8.5578564008326e-06, | |
| "loss": 0.3243, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.28913355672570606, | |
| "grad_norm": 0.33571740984916687, | |
| "learning_rate": 8.535533905932739e-06, | |
| "loss": 0.3352, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.2910483484921015, | |
| "grad_norm": 0.36411982774734497, | |
| "learning_rate": 8.513069573898944e-06, | |
| "loss": 0.3423, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.2929631402584969, | |
| "grad_norm": 0.3553025722503662, | |
| "learning_rate": 8.490464305946296e-06, | |
| "loss": 0.3486, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.2948779320248923, | |
| "grad_norm": 0.34129369258880615, | |
| "learning_rate": 8.467719008943886e-06, | |
| "loss": 0.3355, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.2967927237912877, | |
| "grad_norm": 0.34735849499702454, | |
| "learning_rate": 8.444834595378434e-06, | |
| "loss": 0.3432, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.2987075155576831, | |
| "grad_norm": 0.3442891836166382, | |
| "learning_rate": 8.421811983317682e-06, | |
| "loss": 0.3303, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.3006223073240785, | |
| "grad_norm": 0.3469332754611969, | |
| "learning_rate": 8.398652096373566e-06, | |
| "loss": 0.3455, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.30253709909047394, | |
| "grad_norm": 0.3474913239479065, | |
| "learning_rate": 8.375355863665155e-06, | |
| "loss": 0.3317, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.3044518908568693, | |
| "grad_norm": 0.34943217039108276, | |
| "learning_rate": 8.351924219781393e-06, | |
| "loss": 0.3632, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.3063666826232647, | |
| "grad_norm": 0.35648858547210693, | |
| "learning_rate": 8.328358104743588e-06, | |
| "loss": 0.3409, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.30828147438966014, | |
| "grad_norm": 0.32513663172721863, | |
| "learning_rate": 8.304658463967705e-06, | |
| "loss": 0.3387, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.31019626615605556, | |
| "grad_norm": 0.31542882323265076, | |
| "learning_rate": 8.28082624822645e-06, | |
| "loss": 0.31, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.3121110579224509, | |
| "grad_norm": 0.3230755031108856, | |
| "learning_rate": 8.256862413611113e-06, | |
| "loss": 0.3177, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.31402584968884634, | |
| "grad_norm": 0.3628579080104828, | |
| "learning_rate": 8.232767921493216e-06, | |
| "loss": 0.3574, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.31594064145524176, | |
| "grad_norm": 0.3684355914592743, | |
| "learning_rate": 8.20854373848595e-06, | |
| "loss": 0.3701, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.3178554332216372, | |
| "grad_norm": 0.3106248378753662, | |
| "learning_rate": 8.184190836405394e-06, | |
| "loss": 0.3071, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.31977022498803254, | |
| "grad_norm": 0.35116639733314514, | |
| "learning_rate": 8.15971019223152e-06, | |
| "loss": 0.3386, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.32168501675442795, | |
| "grad_norm": 0.32329127192497253, | |
| "learning_rate": 8.135102788069015e-06, | |
| "loss": 0.2877, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.3235998085208234, | |
| "grad_norm": 0.32711708545684814, | |
| "learning_rate": 8.110369611107869e-06, | |
| "loss": 0.3049, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.3255146002872188, | |
| "grad_norm": 0.3359197676181793, | |
| "learning_rate": 8.085511653583772e-06, | |
| "loss": 0.3127, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.32742939205361415, | |
| "grad_norm": 0.3889077305793762, | |
| "learning_rate": 8.060529912738316e-06, | |
| "loss": 0.37, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.3293441838200096, | |
| "grad_norm": 0.39840781688690186, | |
| "learning_rate": 8.035425390778975e-06, | |
| "loss": 0.3799, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.331258975586405, | |
| "grad_norm": 0.34368225932121277, | |
| "learning_rate": 8.010199094838915e-06, | |
| "loss": 0.3429, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.3331737673528004, | |
| "grad_norm": 0.3298027515411377, | |
| "learning_rate": 7.984852036936578e-06, | |
| "loss": 0.3141, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.3350885591191958, | |
| "grad_norm": 0.3519516587257385, | |
| "learning_rate": 7.959385233935087e-06, | |
| "loss": 0.3655, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.3370033508855912, | |
| "grad_norm": 0.31407034397125244, | |
| "learning_rate": 7.933799707501448e-06, | |
| "loss": 0.289, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.3389181426519866, | |
| "grad_norm": 0.3381166458129883, | |
| "learning_rate": 7.908096484065569e-06, | |
| "loss": 0.3389, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.340832934418382, | |
| "grad_norm": 0.3414974510669708, | |
| "learning_rate": 7.88227659477908e-06, | |
| "loss": 0.3394, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.3427477261847774, | |
| "grad_norm": 0.3472573161125183, | |
| "learning_rate": 7.856341075473963e-06, | |
| "loss": 0.318, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.3446625179511728, | |
| "grad_norm": 0.3200587332248688, | |
| "learning_rate": 7.830290966620997e-06, | |
| "loss": 0.3046, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.3465773097175682, | |
| "grad_norm": 0.33979329466819763, | |
| "learning_rate": 7.804127313288023e-06, | |
| "loss": 0.3293, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.34849210148396365, | |
| "grad_norm": 0.352670282125473, | |
| "learning_rate": 7.777851165098012e-06, | |
| "loss": 0.3446, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.350406893250359, | |
| "grad_norm": 0.3321717381477356, | |
| "learning_rate": 7.751463576186957e-06, | |
| "loss": 0.3144, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.3523216850167544, | |
| "grad_norm": 0.3698163628578186, | |
| "learning_rate": 7.72496560516159e-06, | |
| "loss": 0.3421, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.35423647678314985, | |
| "grad_norm": 0.3798256516456604, | |
| "learning_rate": 7.6983583150569e-06, | |
| "loss": 0.3472, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.35615126854954526, | |
| "grad_norm": 0.33373215794563293, | |
| "learning_rate": 7.671642773293506e-06, | |
| "loss": 0.3191, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.3580660603159406, | |
| "grad_norm": 0.37389230728149414, | |
| "learning_rate": 7.644820051634813e-06, | |
| "loss": 0.3496, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.35998085208233604, | |
| "grad_norm": 0.3754700720310211, | |
| "learning_rate": 7.617891226144034e-06, | |
| "loss": 0.394, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.36189564384873146, | |
| "grad_norm": 0.32576683163642883, | |
| "learning_rate": 7.59085737714101e-06, | |
| "loss": 0.316, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.3638104356151269, | |
| "grad_norm": 0.39560186862945557, | |
| "learning_rate": 7.563719589158874e-06, | |
| "loss": 0.3461, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.36572522738152224, | |
| "grad_norm": 0.3524802625179291, | |
| "learning_rate": 7.536478950900537e-06, | |
| "loss": 0.3406, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.36764001914791766, | |
| "grad_norm": 0.33458060026168823, | |
| "learning_rate": 7.509136555195025e-06, | |
| "loss": 0.326, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.3695548109143131, | |
| "grad_norm": 0.33656662702560425, | |
| "learning_rate": 7.481693498953621e-06, | |
| "loss": 0.3441, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.3714696026807085, | |
| "grad_norm": 0.3443593680858612, | |
| "learning_rate": 7.4541508831258695e-06, | |
| "loss": 0.3201, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.37338439444710386, | |
| "grad_norm": 0.33111077547073364, | |
| "learning_rate": 7.4265098126554065e-06, | |
| "loss": 0.3144, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.3752991862134993, | |
| "grad_norm": 0.37307754158973694, | |
| "learning_rate": 7.3987713964356335e-06, | |
| "loss": 0.3721, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.3772139779798947, | |
| "grad_norm": 0.4052973687648773, | |
| "learning_rate": 7.370936747265226e-06, | |
| "loss": 0.381, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.3791287697462901, | |
| "grad_norm": 0.34146490693092346, | |
| "learning_rate": 7.3430069818035e-06, | |
| "loss": 0.3244, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.3810435615126855, | |
| "grad_norm": 0.3338208794593811, | |
| "learning_rate": 7.314983220525604e-06, | |
| "loss": 0.3263, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.3829583532790809, | |
| "grad_norm": 0.31547433137893677, | |
| "learning_rate": 7.286866587677576e-06, | |
| "loss": 0.3121, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.3848731450454763, | |
| "grad_norm": 0.3688790798187256, | |
| "learning_rate": 7.2586582112312355e-06, | |
| "loss": 0.3507, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.38678793681187174, | |
| "grad_norm": 0.33696216344833374, | |
| "learning_rate": 7.230359222838939e-06, | |
| "loss": 0.319, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.3887027285782671, | |
| "grad_norm": 0.3736443817615509, | |
| "learning_rate": 7.201970757788172e-06, | |
| "loss": 0.3762, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.3906175203446625, | |
| "grad_norm": 0.3320814371109009, | |
| "learning_rate": 7.173493954956012e-06, | |
| "loss": 0.3178, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.39253231211105793, | |
| "grad_norm": 0.3501550555229187, | |
| "learning_rate": 7.144929956763438e-06, | |
| "loss": 0.3352, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.39444710387745335, | |
| "grad_norm": 0.3848736882209778, | |
| "learning_rate": 7.116279909129492e-06, | |
| "loss": 0.3741, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.3963618956438487, | |
| "grad_norm": 0.35347843170166016, | |
| "learning_rate": 7.087544961425317e-06, | |
| "loss": 0.3326, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.39827668741024413, | |
| "grad_norm": 0.34179723262786865, | |
| "learning_rate": 7.058726266428042e-06, | |
| "loss": 0.3322, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.40019147917663955, | |
| "grad_norm": 0.3124852478504181, | |
| "learning_rate": 7.029824980274536e-06, | |
| "loss": 0.3096, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.40210627094303497, | |
| "grad_norm": 0.350807249546051, | |
| "learning_rate": 7.0008422624150285e-06, | |
| "loss": 0.3431, | |
| "step": 210 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 522, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 105, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.1029906388628275e+17, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |