| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.19745502413339183, | |
| "eval_steps": 50, | |
| "global_step": 150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0013163668275559457, | |
| "grad_norm": 0.7211491759367259, | |
| "learning_rate": 2e-05, | |
| "loss": 1.9261, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0026327336551118913, | |
| "grad_norm": 0.7486434737524514, | |
| "learning_rate": 1.999777729859618e-05, | |
| "loss": 1.9006, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.003949100482667837, | |
| "grad_norm": 13.325845555145555, | |
| "learning_rate": 1.9991110182465032e-05, | |
| "loss": 2.1757, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.005265467310223783, | |
| "grad_norm": 8.105149289944798, | |
| "learning_rate": 1.9980001615408228e-05, | |
| "loss": 2.1193, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006581834137779728, | |
| "grad_norm": 4.1238351718190644, | |
| "learning_rate": 1.9964456535631287e-05, | |
| "loss": 2.0406, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007898200965335674, | |
| "grad_norm": 2.390019164808198, | |
| "learning_rate": 1.9944481853548335e-05, | |
| "loss": 2.0089, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.009214567792891619, | |
| "grad_norm": 1.6494732405988457, | |
| "learning_rate": 1.9920086448710162e-05, | |
| "loss": 1.9896, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.010530934620447565, | |
| "grad_norm": 2.0568084440045085, | |
| "learning_rate": 1.9891281165856876e-05, | |
| "loss": 1.9538, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01184730144800351, | |
| "grad_norm": 1.746033674255867, | |
| "learning_rate": 1.9858078810097004e-05, | |
| "loss": 1.9822, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.013163668275559455, | |
| "grad_norm": 1.3967713783340578, | |
| "learning_rate": 1.98204941412151e-05, | |
| "loss": 1.9768, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014480035103115402, | |
| "grad_norm": 1.65089292694443, | |
| "learning_rate": 1.9778543867110428e-05, | |
| "loss": 1.9758, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.015796401930671347, | |
| "grad_norm": 1.2861013174507632, | |
| "learning_rate": 1.9732246636369605e-05, | |
| "loss": 1.9786, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.017112768758227294, | |
| "grad_norm": 0.8252394369240796, | |
| "learning_rate": 1.968162302997659e-05, | |
| "loss": 1.98, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.018429135585783237, | |
| "grad_norm": 1.0005195006626915, | |
| "learning_rate": 1.962669555216358e-05, | |
| "loss": 1.9639, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.019745502413339184, | |
| "grad_norm": 1.4352178043541923, | |
| "learning_rate": 1.9567488620406984e-05, | |
| "loss": 1.9808, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02106186924089513, | |
| "grad_norm": 0.8303153298095659, | |
| "learning_rate": 1.9504028554572865e-05, | |
| "loss": 1.9648, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.022378236068451074, | |
| "grad_norm": 1.404505073273737, | |
| "learning_rate": 1.943634356521671e-05, | |
| "loss": 1.9743, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02369460289600702, | |
| "grad_norm": 0.9799958151016589, | |
| "learning_rate": 1.9364463741042694e-05, | |
| "loss": 1.9396, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.025010969723562967, | |
| "grad_norm": 0.7125809437112708, | |
| "learning_rate": 1.928842103552803e-05, | |
| "loss": 1.9493, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02632733655111891, | |
| "grad_norm": 1.0260490332440508, | |
| "learning_rate": 1.920824925271838e-05, | |
| "loss": 1.9657, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.027643703378674857, | |
| "grad_norm": 1.026830841937651, | |
| "learning_rate": 1.9123984032200586e-05, | |
| "loss": 1.9447, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.028960070206230804, | |
| "grad_norm": 0.5967900308007423, | |
| "learning_rate": 1.9035662833259433e-05, | |
| "loss": 1.9623, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.030276437033786747, | |
| "grad_norm": 0.7530918392820164, | |
| "learning_rate": 1.8943324918225495e-05, | |
| "loss": 1.9615, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.031592803861342694, | |
| "grad_norm": 0.8265653626705934, | |
| "learning_rate": 1.8847011335021447e-05, | |
| "loss": 1.9477, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03290917068889864, | |
| "grad_norm": 0.6230744161074221, | |
| "learning_rate": 1.874676489891461e-05, | |
| "loss": 1.9501, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03422553751645459, | |
| "grad_norm": 0.5776206283484728, | |
| "learning_rate": 1.8642630173483832e-05, | |
| "loss": 1.95, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03554190434401053, | |
| "grad_norm": 0.6475828415339461, | |
| "learning_rate": 1.85346534508092e-05, | |
| "loss": 1.954, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.036858271171566474, | |
| "grad_norm": 0.6175950618623995, | |
| "learning_rate": 1.8422882730893323e-05, | |
| "loss": 1.9723, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03817463799912242, | |
| "grad_norm": 0.48955184760547843, | |
| "learning_rate": 1.8307367700323412e-05, | |
| "loss": 1.9689, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03949100482667837, | |
| "grad_norm": 0.5627729339809433, | |
| "learning_rate": 1.8188159710183595e-05, | |
| "loss": 1.9383, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.040807371654234315, | |
| "grad_norm": 0.5843115162113592, | |
| "learning_rate": 1.8065311753227272e-05, | |
| "loss": 1.9839, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04212373848179026, | |
| "grad_norm": 0.46825827290489286, | |
| "learning_rate": 1.7938878440319722e-05, | |
| "loss": 1.9751, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0434401053093462, | |
| "grad_norm": 0.5121356731921046, | |
| "learning_rate": 1.7808915976161364e-05, | |
| "loss": 1.9782, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04475647213690215, | |
| "grad_norm": 0.4946367754339634, | |
| "learning_rate": 1.7675482134302503e-05, | |
| "loss": 1.9586, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.046072838964458095, | |
| "grad_norm": 0.5321421781184499, | |
| "learning_rate": 1.753863623146066e-05, | |
| "loss": 1.9623, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04738920579201404, | |
| "grad_norm": 0.49841500005666933, | |
| "learning_rate": 1.7398439101151908e-05, | |
| "loss": 1.9485, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04870557261956999, | |
| "grad_norm": 0.5015785236055983, | |
| "learning_rate": 1.7254953066647915e-05, | |
| "loss": 1.9528, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.050021939447125935, | |
| "grad_norm": 0.5068614182885935, | |
| "learning_rate": 1.710824191327075e-05, | |
| "loss": 1.9858, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.051338306274681875, | |
| "grad_norm": 0.5343507342806129, | |
| "learning_rate": 1.695837086003772e-05, | |
| "loss": 1.9609, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.05265467310223782, | |
| "grad_norm": 0.5334849711475083, | |
| "learning_rate": 1.680540653066891e-05, | |
| "loss": 1.959, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05397103992979377, | |
| "grad_norm": 0.4554441367354399, | |
| "learning_rate": 1.6649416923970248e-05, | |
| "loss": 1.9596, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.055287406757349715, | |
| "grad_norm": 0.5693121720524362, | |
| "learning_rate": 1.649047138360529e-05, | |
| "loss": 1.9514, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05660377358490566, | |
| "grad_norm": 0.5691191050538144, | |
| "learning_rate": 1.632864056726917e-05, | |
| "loss": 1.9523, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05792014041246161, | |
| "grad_norm": 0.47820348756941083, | |
| "learning_rate": 1.6163996415278423e-05, | |
| "loss": 1.9782, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05923650724001755, | |
| "grad_norm": 0.5159202693438228, | |
| "learning_rate": 1.5996612118590604e-05, | |
| "loss": 1.9582, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.060552874067573495, | |
| "grad_norm": 0.556592049641545, | |
| "learning_rate": 1.5826562086267956e-05, | |
| "loss": 1.9775, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06186924089512944, | |
| "grad_norm": 0.5850524732394254, | |
| "learning_rate": 1.565392191239959e-05, | |
| "loss": 1.9651, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06318560772268539, | |
| "grad_norm": 0.4835939602122527, | |
| "learning_rate": 1.5478768342496872e-05, | |
| "loss": 1.955, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06450197455024133, | |
| "grad_norm": 0.44146549041758676, | |
| "learning_rate": 1.5301179239376936e-05, | |
| "loss": 1.9691, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06581834137779728, | |
| "grad_norm": 0.5280299432517168, | |
| "learning_rate": 1.512123354854955e-05, | |
| "loss": 1.9485, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06713470820535322, | |
| "grad_norm": 0.5524916916858923, | |
| "learning_rate": 1.4939011263122635e-05, | |
| "loss": 1.9493, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06845107503290918, | |
| "grad_norm": 0.47531567433880617, | |
| "learning_rate": 1.4754593388242117e-05, | |
| "loss": 1.9584, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06976744186046512, | |
| "grad_norm": 0.45272463859878564, | |
| "learning_rate": 1.4568061905081874e-05, | |
| "loss": 1.9494, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.07108380868802106, | |
| "grad_norm": 0.4702181357165236, | |
| "learning_rate": 1.4379499734399797e-05, | |
| "loss": 1.9619, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07240017551557701, | |
| "grad_norm": 0.45379341896591296, | |
| "learning_rate": 1.4188990699676186e-05, | |
| "loss": 1.9585, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07371654234313295, | |
| "grad_norm": 0.4775896961874003, | |
| "learning_rate": 1.3996619489850822e-05, | |
| "loss": 1.9713, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0750329091706889, | |
| "grad_norm": 0.4814200193397695, | |
| "learning_rate": 1.3802471621675337e-05, | |
| "loss": 1.954, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07634927599824484, | |
| "grad_norm": 0.5271247311659947, | |
| "learning_rate": 1.3606633401697557e-05, | |
| "loss": 1.9769, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0776656428258008, | |
| "grad_norm": 0.47173817615214364, | |
| "learning_rate": 1.340919188789477e-05, | |
| "loss": 1.957, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07898200965335674, | |
| "grad_norm": 0.4219944844859643, | |
| "learning_rate": 1.3210234850972966e-05, | |
| "loss": 1.9752, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08029837648091268, | |
| "grad_norm": 0.4790011427747401, | |
| "learning_rate": 1.300985073534919e-05, | |
| "loss": 1.9747, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08161474330846863, | |
| "grad_norm": 0.4441961940866282, | |
| "learning_rate": 1.280812861983446e-05, | |
| "loss": 1.9544, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08293111013602457, | |
| "grad_norm": 0.45824396425465297, | |
| "learning_rate": 1.2605158178034656e-05, | |
| "loss": 1.9594, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08424747696358052, | |
| "grad_norm": 0.4355014232337334, | |
| "learning_rate": 1.2401029638486952e-05, | |
| "loss": 1.9369, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08556384379113646, | |
| "grad_norm": 0.4231369314259089, | |
| "learning_rate": 1.219583374454963e-05, | |
| "loss": 1.9738, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0868802106186924, | |
| "grad_norm": 0.45570619643070215, | |
| "learning_rate": 1.1989661714063e-05, | |
| "loss": 1.9912, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08819657744624836, | |
| "grad_norm": 0.37812336844029865, | |
| "learning_rate": 1.1782605198799371e-05, | |
| "loss": 1.9784, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0895129442738043, | |
| "grad_norm": 0.43070668661161077, | |
| "learning_rate": 1.157475624372018e-05, | |
| "loss": 1.951, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09082931110136025, | |
| "grad_norm": 0.4218410596859208, | |
| "learning_rate": 1.1366207246058269e-05, | |
| "loss": 1.9619, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09214567792891619, | |
| "grad_norm": 0.39509398596716727, | |
| "learning_rate": 1.1157050914243614e-05, | |
| "loss": 1.9736, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09346204475647214, | |
| "grad_norm": 0.42796823466878475, | |
| "learning_rate": 1.0947380226690686e-05, | |
| "loss": 1.969, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09477841158402808, | |
| "grad_norm": 0.3830330189798859, | |
| "learning_rate": 1.0737288390465792e-05, | |
| "loss": 1.9592, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09609477841158402, | |
| "grad_norm": 0.4151941986685396, | |
| "learning_rate": 1.0526868799852797e-05, | |
| "loss": 1.9824, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09741114523913998, | |
| "grad_norm": 0.4129906835751094, | |
| "learning_rate": 1.031621499483559e-05, | |
| "loss": 1.9754, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09872751206669592, | |
| "grad_norm": 0.37038541407177233, | |
| "learning_rate": 1.0105420619515798e-05, | |
| "loss": 1.9659, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.10004387889425187, | |
| "grad_norm": 0.4024942738947187, | |
| "learning_rate": 9.894579380484206e-06, | |
| "loss": 1.9821, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.10136024572180781, | |
| "grad_norm": 0.3924522821035196, | |
| "learning_rate": 9.683785005164412e-06, | |
| "loss": 1.9871, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10267661254936375, | |
| "grad_norm": 0.3903897898907002, | |
| "learning_rate": 9.473131200147205e-06, | |
| "loss": 1.9936, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1039929793769197, | |
| "grad_norm": 0.37456390666059425, | |
| "learning_rate": 9.262711609534211e-06, | |
| "loss": 1.9791, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10530934620447564, | |
| "grad_norm": 0.38352789789852887, | |
| "learning_rate": 9.052619773309318e-06, | |
| "loss": 1.9704, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1066257130320316, | |
| "grad_norm": 0.3398793764087006, | |
| "learning_rate": 8.842949085756389e-06, | |
| "loss": 1.9658, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10794207985958754, | |
| "grad_norm": 0.35035212137475186, | |
| "learning_rate": 8.633792753941733e-06, | |
| "loss": 1.9759, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10925844668714349, | |
| "grad_norm": 0.38362703145032767, | |
| "learning_rate": 8.425243756279824e-06, | |
| "loss": 1.9546, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.11057481351469943, | |
| "grad_norm": 0.33953622024923164, | |
| "learning_rate": 8.217394801200632e-06, | |
| "loss": 1.9762, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11189118034225537, | |
| "grad_norm": 0.3529326737966424, | |
| "learning_rate": 8.010338285937006e-06, | |
| "loss": 1.9645, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11320754716981132, | |
| "grad_norm": 0.3589128762503322, | |
| "learning_rate": 7.804166255450372e-06, | |
| "loss": 1.9702, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11452391399736726, | |
| "grad_norm": 0.35199885106641443, | |
| "learning_rate": 7.598970361513052e-06, | |
| "loss": 1.9738, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11584028082492322, | |
| "grad_norm": 0.3303328690896044, | |
| "learning_rate": 7.394841821965345e-06, | |
| "loss": 1.9517, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11715664765247916, | |
| "grad_norm": 0.3070650490072722, | |
| "learning_rate": 7.191871380165538e-06, | |
| "loss": 1.976, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1184730144800351, | |
| "grad_norm": 0.3425188692261718, | |
| "learning_rate": 6.990149264650814e-06, | |
| "loss": 1.9679, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11978938130759105, | |
| "grad_norm": 0.31400541161735, | |
| "learning_rate": 6.789765149027039e-06, | |
| "loss": 1.9744, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.12110574813514699, | |
| "grad_norm": 0.2812642549380601, | |
| "learning_rate": 6.590808112105232e-06, | |
| "loss": 1.991, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12242211496270294, | |
| "grad_norm": 0.3484100272693502, | |
| "learning_rate": 6.3933665983024465e-06, | |
| "loss": 1.9758, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12373848179025888, | |
| "grad_norm": 0.29533510297797094, | |
| "learning_rate": 6.197528378324664e-06, | |
| "loss": 1.9532, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12505484861781482, | |
| "grad_norm": 0.3106735461942925, | |
| "learning_rate": 6.003380510149179e-06, | |
| "loss": 1.9808, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12637121544537078, | |
| "grad_norm": 0.3039339424173133, | |
| "learning_rate": 5.8110093003238175e-06, | |
| "loss": 2.0037, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12768758227292673, | |
| "grad_norm": 0.32068909556214314, | |
| "learning_rate": 5.620500265600206e-06, | |
| "loss": 1.9765, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12900394910048266, | |
| "grad_norm": 0.30235061650814415, | |
| "learning_rate": 5.431938094918132e-06, | |
| "loss": 1.9872, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1303203159280386, | |
| "grad_norm": 0.2806362106759796, | |
| "learning_rate": 5.245406611757882e-06, | |
| "loss": 1.9859, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.13163668275559456, | |
| "grad_norm": 0.27623419463635634, | |
| "learning_rate": 5.060988736877366e-06, | |
| "loss": 1.967, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13295304958315052, | |
| "grad_norm": 0.30416859069440383, | |
| "learning_rate": 4.878766451450451e-06, | |
| "loss": 1.9743, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.13426941641070644, | |
| "grad_norm": 0.28168492270097995, | |
| "learning_rate": 4.698820760623064e-06, | |
| "loss": 1.9683, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1355857832382624, | |
| "grad_norm": 0.25701853953499526, | |
| "learning_rate": 4.5212316575031325e-06, | |
| "loss": 1.9809, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13690215006581835, | |
| "grad_norm": 0.26352997212190776, | |
| "learning_rate": 4.346078087600411e-06, | |
| "loss": 1.9852, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13821851689337428, | |
| "grad_norm": 0.2774470031738603, | |
| "learning_rate": 4.173437913732048e-06, | |
| "loss": 1.9822, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13953488372093023, | |
| "grad_norm": 0.26016253702735714, | |
| "learning_rate": 4.003387881409397e-06, | |
| "loss": 1.9863, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.14085125054848618, | |
| "grad_norm": 0.24520268288962063, | |
| "learning_rate": 3.836003584721577e-06, | |
| "loss": 1.9854, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1421676173760421, | |
| "grad_norm": 0.25962753823152684, | |
| "learning_rate": 3.6713594327308343e-06, | |
| "loss": 1.9708, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14348398420359806, | |
| "grad_norm": 0.2450558999574294, | |
| "learning_rate": 3.509528616394716e-06, | |
| "loss": 1.9883, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.14480035103115402, | |
| "grad_norm": 0.2540644323176429, | |
| "learning_rate": 3.3505830760297543e-06, | |
| "loss": 1.9842, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14611671785870997, | |
| "grad_norm": 0.2511336974868001, | |
| "learning_rate": 3.1945934693310897e-06, | |
| "loss": 1.9907, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.1474330846862659, | |
| "grad_norm": 0.2520194579274834, | |
| "learning_rate": 3.0416291399622834e-06, | |
| "loss": 2.0161, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14874945151382185, | |
| "grad_norm": 0.24030125812599737, | |
| "learning_rate": 2.891758086729253e-06, | |
| "loss": 1.9683, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1500658183413778, | |
| "grad_norm": 0.25198964259356177, | |
| "learning_rate": 2.7450469333520856e-06, | |
| "loss": 1.9744, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15138218516893373, | |
| "grad_norm": 0.24584263977943674, | |
| "learning_rate": 2.6015608988480956e-06, | |
| "loss": 1.9784, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15269855199648968, | |
| "grad_norm": 0.23071754319622353, | |
| "learning_rate": 2.4613637685393433e-06, | |
| "loss": 1.9881, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.15401491882404564, | |
| "grad_norm": 0.22843475214114475, | |
| "learning_rate": 2.324517865697501e-06, | |
| "loss": 1.9619, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.1553312856516016, | |
| "grad_norm": 0.22517499817054293, | |
| "learning_rate": 2.19108402383864e-06, | |
| "loss": 1.9747, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15664765247915752, | |
| "grad_norm": 0.22745976008029595, | |
| "learning_rate": 2.06112155968028e-06, | |
| "loss": 1.994, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.15796401930671347, | |
| "grad_norm": 0.22047239875500865, | |
| "learning_rate": 1.9346882467727323e-06, | |
| "loss": 1.9988, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15928038613426942, | |
| "grad_norm": 0.2288393265643624, | |
| "learning_rate": 1.811840289816409e-06, | |
| "loss": 1.9916, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.16059675296182535, | |
| "grad_norm": 0.210617978293029, | |
| "learning_rate": 1.6926322996765899e-06, | |
| "loss": 1.9925, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.1619131197893813, | |
| "grad_norm": 0.22672890929369188, | |
| "learning_rate": 1.5771172691066793e-06, | |
| "loss": 1.9966, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.16322948661693726, | |
| "grad_norm": 0.2092439044635155, | |
| "learning_rate": 1.4653465491908003e-06, | |
| "loss": 2.0197, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.1645458534444932, | |
| "grad_norm": 0.23171631380591984, | |
| "learning_rate": 1.3573698265161683e-06, | |
| "loss": 1.9757, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16586222027204914, | |
| "grad_norm": 0.21732365123324443, | |
| "learning_rate": 1.2532351010853916e-06, | |
| "loss": 1.9807, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1671785870996051, | |
| "grad_norm": 0.2102223203829434, | |
| "learning_rate": 1.152988664978556e-06, | |
| "loss": 1.9884, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.16849495392716105, | |
| "grad_norm": 0.2068629684474071, | |
| "learning_rate": 1.0566750817745076e-06, | |
| "loss": 1.9954, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16981132075471697, | |
| "grad_norm": 0.20342610599286592, | |
| "learning_rate": 9.6433716674057e-07, | |
| "loss": 1.9875, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.17112768758227292, | |
| "grad_norm": 0.2207294165566657, | |
| "learning_rate": 8.760159677994174e-07, | |
| "loss": 1.9922, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17244405440982888, | |
| "grad_norm": 0.20892064152928483, | |
| "learning_rate": 7.91750747281621e-07, | |
| "loss": 1.9832, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1737604212373848, | |
| "grad_norm": 0.20323723950795747, | |
| "learning_rate": 7.115789644719728e-07, | |
| "loss": 2.0036, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17507678806494076, | |
| "grad_norm": 0.19306149399613967, | |
| "learning_rate": 6.355362589573078e-07, | |
| "loss": 1.9851, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.1763931548924967, | |
| "grad_norm": 0.1996406839141809, | |
| "learning_rate": 5.636564347832907e-07, | |
| "loss": 1.9906, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.17770952172005267, | |
| "grad_norm": 0.1907672927300904, | |
| "learning_rate": 4.95971445427137e-07, | |
| "loss": 1.9917, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1790258885476086, | |
| "grad_norm": 0.1903641108666616, | |
| "learning_rate": 4.3251137959302023e-07, | |
| "loss": 1.979, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.18034225537516455, | |
| "grad_norm": 0.1933990262191582, | |
| "learning_rate": 3.733044478364234e-07, | |
| "loss": 1.9758, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1816586222027205, | |
| "grad_norm": 0.1849617366395776, | |
| "learning_rate": 3.1837697002341293e-07, | |
| "loss": 1.9874, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18297498903027642, | |
| "grad_norm": 0.1896652425818316, | |
| "learning_rate": 2.677533636303964e-07, | |
| "loss": 1.9776, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.18429135585783238, | |
| "grad_norm": 0.19421748901847316, | |
| "learning_rate": 2.214561328895748e-07, | |
| "loss": 1.9825, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18560772268538833, | |
| "grad_norm": 0.1956035221064292, | |
| "learning_rate": 1.7950585878489856e-07, | |
| "loss": 1.9873, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.18692408951294429, | |
| "grad_norm": 0.19967363712410285, | |
| "learning_rate": 1.419211899029971e-07, | |
| "loss": 1.9788, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.1882404563405002, | |
| "grad_norm": 0.18474817640649746, | |
| "learning_rate": 1.0871883414312778e-07, | |
| "loss": 1.9888, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.18955682316805617, | |
| "grad_norm": 0.1864458375859747, | |
| "learning_rate": 7.99135512898408e-08, | |
| "loss": 1.9811, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.19087318999561212, | |
| "grad_norm": 0.18257336160991286, | |
| "learning_rate": 5.55181464516652e-08, | |
| "loss": 1.9833, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.19218955682316805, | |
| "grad_norm": 0.190055973888541, | |
| "learning_rate": 3.554346436871581e-08, | |
| "loss": 1.9782, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.193505923650724, | |
| "grad_norm": 0.18495546875149854, | |
| "learning_rate": 1.9998384591773945e-08, | |
| "loss": 1.9869, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19482229047827995, | |
| "grad_norm": 0.18240518242205553, | |
| "learning_rate": 8.889817534969425e-09, | |
| "loss": 1.9946, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.1961386573058359, | |
| "grad_norm": 0.18579782376852144, | |
| "learning_rate": 2.222701403818972e-09, | |
| "loss": 1.9835, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.19745502413339183, | |
| "grad_norm": 0.20258365648035168, | |
| "learning_rate": 0.0, | |
| "loss": 1.996, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19745502413339183, | |
| "step": 150, | |
| "total_flos": 349677715193856.0, | |
| "train_loss": 1.9757762233416238, | |
| "train_runtime": 5203.4288, | |
| "train_samples_per_second": 58.116, | |
| "train_steps_per_second": 0.029 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 150, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 50, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 349677715193856.0, | |
| "train_batch_size": 42, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |