| { | |
| "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": 1.65111437852943, | |
| "learning_rate": 2e-05, | |
| "loss": 1.7474, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0026327336551118913, | |
| "grad_norm": 1.7295359235133803, | |
| "learning_rate": 1.999777729859618e-05, | |
| "loss": 1.712, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.003949100482667837, | |
| "grad_norm": 10.447342724724326, | |
| "learning_rate": 1.9991110182465032e-05, | |
| "loss": 2.0566, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.005265467310223783, | |
| "grad_norm": 8.439236569082253, | |
| "learning_rate": 1.9980001615408228e-05, | |
| "loss": 2.0446, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006581834137779728, | |
| "grad_norm": 5.502178376287468, | |
| "learning_rate": 1.9964456535631287e-05, | |
| "loss": 1.9864, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007898200965335674, | |
| "grad_norm": 3.4244942197147945, | |
| "learning_rate": 1.9944481853548335e-05, | |
| "loss": 1.9393, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.009214567792891619, | |
| "grad_norm": 2.2479839558307013, | |
| "learning_rate": 1.9920086448710162e-05, | |
| "loss": 1.8865, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.010530934620447565, | |
| "grad_norm": 1.7351334611523515, | |
| "learning_rate": 1.9891281165856876e-05, | |
| "loss": 1.8423, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.01184730144800351, | |
| "grad_norm": 1.4656029971843716, | |
| "learning_rate": 1.9858078810097004e-05, | |
| "loss": 1.8586, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.013163668275559455, | |
| "grad_norm": 1.3773721079335524, | |
| "learning_rate": 1.98204941412151e-05, | |
| "loss": 1.8524, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.014480035103115402, | |
| "grad_norm": 1.1578401907098703, | |
| "learning_rate": 1.9778543867110428e-05, | |
| "loss": 1.8525, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.015796401930671347, | |
| "grad_norm": 1.0088566042734737, | |
| "learning_rate": 1.9732246636369605e-05, | |
| "loss": 1.8441, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.017112768758227294, | |
| "grad_norm": 0.9601192905672813, | |
| "learning_rate": 1.968162302997659e-05, | |
| "loss": 1.8396, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.018429135585783237, | |
| "grad_norm": 0.9248937703572266, | |
| "learning_rate": 1.962669555216358e-05, | |
| "loss": 1.8251, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.019745502413339184, | |
| "grad_norm": 0.9073184114547326, | |
| "learning_rate": 1.9567488620406984e-05, | |
| "loss": 1.8546, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02106186924089513, | |
| "grad_norm": 0.7850875076115653, | |
| "learning_rate": 1.9504028554572865e-05, | |
| "loss": 1.8354, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.022378236068451074, | |
| "grad_norm": 0.8835800418613018, | |
| "learning_rate": 1.943634356521671e-05, | |
| "loss": 1.8499, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.02369460289600702, | |
| "grad_norm": 0.763280002887791, | |
| "learning_rate": 1.9364463741042694e-05, | |
| "loss": 1.8137, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.025010969723562967, | |
| "grad_norm": 0.7586586554363934, | |
| "learning_rate": 1.928842103552803e-05, | |
| "loss": 1.8228, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.02632733655111891, | |
| "grad_norm": 0.7075808706792394, | |
| "learning_rate": 1.920824925271838e-05, | |
| "loss": 1.8425, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.027643703378674857, | |
| "grad_norm": 0.7601165553708226, | |
| "learning_rate": 1.9123984032200586e-05, | |
| "loss": 1.8197, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.028960070206230804, | |
| "grad_norm": 0.7193254428686906, | |
| "learning_rate": 1.9035662833259433e-05, | |
| "loss": 1.8407, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.030276437033786747, | |
| "grad_norm": 0.7215706441009537, | |
| "learning_rate": 1.8943324918225495e-05, | |
| "loss": 1.8395, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.031592803861342694, | |
| "grad_norm": 0.8203870414871476, | |
| "learning_rate": 1.8847011335021447e-05, | |
| "loss": 1.8271, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03290917068889864, | |
| "grad_norm": 0.7431586249459948, | |
| "learning_rate": 1.874676489891461e-05, | |
| "loss": 1.8337, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03422553751645459, | |
| "grad_norm": 0.7737992048688279, | |
| "learning_rate": 1.8642630173483832e-05, | |
| "loss": 1.8356, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.03554190434401053, | |
| "grad_norm": 0.7804596966065455, | |
| "learning_rate": 1.85346534508092e-05, | |
| "loss": 1.8412, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.036858271171566474, | |
| "grad_norm": 0.6931515626982303, | |
| "learning_rate": 1.8422882730893323e-05, | |
| "loss": 1.8614, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.03817463799912242, | |
| "grad_norm": 0.8084803042078094, | |
| "learning_rate": 1.8307367700323412e-05, | |
| "loss": 1.8593, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.03949100482667837, | |
| "grad_norm": 0.626512906169853, | |
| "learning_rate": 1.8188159710183595e-05, | |
| "loss": 1.831, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.040807371654234315, | |
| "grad_norm": 0.7813065255670536, | |
| "learning_rate": 1.8065311753227272e-05, | |
| "loss": 1.8762, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.04212373848179026, | |
| "grad_norm": 0.6706823503135386, | |
| "learning_rate": 1.7938878440319722e-05, | |
| "loss": 1.8683, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0434401053093462, | |
| "grad_norm": 0.6364652672451495, | |
| "learning_rate": 1.7808915976161364e-05, | |
| "loss": 1.8733, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04475647213690215, | |
| "grad_norm": 0.7368025681738634, | |
| "learning_rate": 1.7675482134302503e-05, | |
| "loss": 1.8544, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.046072838964458095, | |
| "grad_norm": 0.5548763381698079, | |
| "learning_rate": 1.753863623146066e-05, | |
| "loss": 1.8598, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04738920579201404, | |
| "grad_norm": 0.7780814291300651, | |
| "learning_rate": 1.7398439101151908e-05, | |
| "loss": 1.8469, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04870557261956999, | |
| "grad_norm": 0.6521757251300341, | |
| "learning_rate": 1.7254953066647915e-05, | |
| "loss": 1.8524, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.050021939447125935, | |
| "grad_norm": 0.7722078567672159, | |
| "learning_rate": 1.710824191327075e-05, | |
| "loss": 1.8855, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.051338306274681875, | |
| "grad_norm": 0.7298655195609485, | |
| "learning_rate": 1.695837086003772e-05, | |
| "loss": 1.8635, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.05265467310223782, | |
| "grad_norm": 0.6468602911451591, | |
| "learning_rate": 1.680540653066891e-05, | |
| "loss": 1.863, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05397103992979377, | |
| "grad_norm": 0.7916429438105029, | |
| "learning_rate": 1.6649416923970248e-05, | |
| "loss": 1.8646, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.055287406757349715, | |
| "grad_norm": 0.5474264565504703, | |
| "learning_rate": 1.649047138360529e-05, | |
| "loss": 1.8576, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05660377358490566, | |
| "grad_norm": 0.7518985692585881, | |
| "learning_rate": 1.632864056726917e-05, | |
| "loss": 1.8606, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05792014041246161, | |
| "grad_norm": 0.5585165468856192, | |
| "learning_rate": 1.6163996415278423e-05, | |
| "loss": 1.8872, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.05923650724001755, | |
| "grad_norm": 0.7084850146213678, | |
| "learning_rate": 1.5996612118590604e-05, | |
| "loss": 1.8692, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.060552874067573495, | |
| "grad_norm": 0.6998461172771332, | |
| "learning_rate": 1.5826562086267956e-05, | |
| "loss": 1.8885, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.06186924089512944, | |
| "grad_norm": 0.7300033950295419, | |
| "learning_rate": 1.565392191239959e-05, | |
| "loss": 1.8772, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.06318560772268539, | |
| "grad_norm": 0.7899786680966636, | |
| "learning_rate": 1.5478768342496872e-05, | |
| "loss": 1.8685, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.06450197455024133, | |
| "grad_norm": 0.6561757387110828, | |
| "learning_rate": 1.5301179239376936e-05, | |
| "loss": 1.8842, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06581834137779728, | |
| "grad_norm": 0.7117935357444856, | |
| "learning_rate": 1.512123354854955e-05, | |
| "loss": 1.865, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.06713470820535322, | |
| "grad_norm": 0.5646089364283718, | |
| "learning_rate": 1.4939011263122635e-05, | |
| "loss": 1.8651, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06845107503290918, | |
| "grad_norm": 0.6673985877657005, | |
| "learning_rate": 1.4754593388242117e-05, | |
| "loss": 1.8768, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06976744186046512, | |
| "grad_norm": 0.5092546824434766, | |
| "learning_rate": 1.4568061905081874e-05, | |
| "loss": 1.868, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.07108380868802106, | |
| "grad_norm": 0.6140848791437851, | |
| "learning_rate": 1.4379499734399797e-05, | |
| "loss": 1.8817, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.07240017551557701, | |
| "grad_norm": 0.5766820950125316, | |
| "learning_rate": 1.4188990699676186e-05, | |
| "loss": 1.8799, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.07371654234313295, | |
| "grad_norm": 0.6127504628805722, | |
| "learning_rate": 1.3996619489850822e-05, | |
| "loss": 1.8936, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0750329091706889, | |
| "grad_norm": 0.6266059301264697, | |
| "learning_rate": 1.3802471621675337e-05, | |
| "loss": 1.8784, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07634927599824484, | |
| "grad_norm": 0.5534750017049672, | |
| "learning_rate": 1.3606633401697557e-05, | |
| "loss": 1.9011, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.0776656428258008, | |
| "grad_norm": 0.6175892828039317, | |
| "learning_rate": 1.340919188789477e-05, | |
| "loss": 1.8837, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.07898200965335674, | |
| "grad_norm": 0.5258068472184501, | |
| "learning_rate": 1.3210234850972966e-05, | |
| "loss": 1.9028, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.08029837648091268, | |
| "grad_norm": 0.5803172257476928, | |
| "learning_rate": 1.300985073534919e-05, | |
| "loss": 1.9027, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.08161474330846863, | |
| "grad_norm": 0.46576381858780536, | |
| "learning_rate": 1.280812861983446e-05, | |
| "loss": 1.8829, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.08293111013602457, | |
| "grad_norm": 0.5243424882503339, | |
| "learning_rate": 1.2605158178034656e-05, | |
| "loss": 1.8897, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.08424747696358052, | |
| "grad_norm": 0.44275509950840103, | |
| "learning_rate": 1.2401029638486952e-05, | |
| "loss": 1.8689, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.08556384379113646, | |
| "grad_norm": 0.47692124746423276, | |
| "learning_rate": 1.219583374454963e-05, | |
| "loss": 1.9078, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0868802106186924, | |
| "grad_norm": 0.423870884057128, | |
| "learning_rate": 1.1989661714063e-05, | |
| "loss": 1.9246, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08819657744624836, | |
| "grad_norm": 0.4344116733448028, | |
| "learning_rate": 1.1782605198799371e-05, | |
| "loss": 1.9138, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.0895129442738043, | |
| "grad_norm": 0.44365257109026207, | |
| "learning_rate": 1.157475624372018e-05, | |
| "loss": 1.8872, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.09082931110136025, | |
| "grad_norm": 0.45705186620536725, | |
| "learning_rate": 1.1366207246058269e-05, | |
| "loss": 1.8984, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.09214567792891619, | |
| "grad_norm": 0.40175081883984465, | |
| "learning_rate": 1.1157050914243614e-05, | |
| "loss": 1.9134, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.09346204475647214, | |
| "grad_norm": 0.42726401125065505, | |
| "learning_rate": 1.0947380226690686e-05, | |
| "loss": 1.9097, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.09477841158402808, | |
| "grad_norm": 0.3900674410637252, | |
| "learning_rate": 1.0737288390465792e-05, | |
| "loss": 1.9008, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.09609477841158402, | |
| "grad_norm": 0.41225255645420644, | |
| "learning_rate": 1.0526868799852797e-05, | |
| "loss": 1.9243, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09741114523913998, | |
| "grad_norm": 0.3838092023163431, | |
| "learning_rate": 1.031621499483559e-05, | |
| "loss": 1.9182, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09872751206669592, | |
| "grad_norm": 0.3599445898903389, | |
| "learning_rate": 1.0105420619515798e-05, | |
| "loss": 1.9112, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.10004387889425187, | |
| "grad_norm": 0.3909645250507488, | |
| "learning_rate": 9.894579380484206e-06, | |
| "loss": 1.9276, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.10136024572180781, | |
| "grad_norm": 0.3640880651891862, | |
| "learning_rate": 9.683785005164412e-06, | |
| "loss": 1.9343, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.10267661254936375, | |
| "grad_norm": 0.3520227157576893, | |
| "learning_rate": 9.473131200147205e-06, | |
| "loss": 1.9412, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.1039929793769197, | |
| "grad_norm": 0.36960565249242594, | |
| "learning_rate": 9.262711609534211e-06, | |
| "loss": 1.9272, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.10530934620447564, | |
| "grad_norm": 0.35697589279129394, | |
| "learning_rate": 9.052619773309318e-06, | |
| "loss": 1.9205, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.1066257130320316, | |
| "grad_norm": 0.3539289322099185, | |
| "learning_rate": 8.842949085756389e-06, | |
| "loss": 1.9167, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10794207985958754, | |
| "grad_norm": 0.377830507323211, | |
| "learning_rate": 8.633792753941733e-06, | |
| "loss": 1.9283, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10925844668714349, | |
| "grad_norm": 0.32762948772243283, | |
| "learning_rate": 8.425243756279824e-06, | |
| "loss": 1.9078, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.11057481351469943, | |
| "grad_norm": 0.3218390246795396, | |
| "learning_rate": 8.217394801200632e-06, | |
| "loss": 1.9303, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.11189118034225537, | |
| "grad_norm": 0.36785562529934085, | |
| "learning_rate": 8.010338285937006e-06, | |
| "loss": 1.9196, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.11320754716981132, | |
| "grad_norm": 0.33474377063086946, | |
| "learning_rate": 7.804166255450372e-06, | |
| "loss": 1.9252, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.11452391399736726, | |
| "grad_norm": 0.3276135559258542, | |
| "learning_rate": 7.598970361513052e-06, | |
| "loss": 1.931, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.11584028082492322, | |
| "grad_norm": 0.34019785956502163, | |
| "learning_rate": 7.394841821965345e-06, | |
| "loss": 1.9104, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.11715664765247916, | |
| "grad_norm": 0.3038814019019217, | |
| "learning_rate": 7.191871380165538e-06, | |
| "loss": 1.936, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.1184730144800351, | |
| "grad_norm": 0.2934690304673656, | |
| "learning_rate": 6.990149264650814e-06, | |
| "loss": 1.9285, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11978938130759105, | |
| "grad_norm": 0.3241397790607444, | |
| "learning_rate": 6.789765149027039e-06, | |
| "loss": 1.9357, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.12110574813514699, | |
| "grad_norm": 0.28870444952258223, | |
| "learning_rate": 6.590808112105232e-06, | |
| "loss": 1.954, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.12242211496270294, | |
| "grad_norm": 0.29485920601939425, | |
| "learning_rate": 6.3933665983024465e-06, | |
| "loss": 1.9394, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.12373848179025888, | |
| "grad_norm": 0.29142694778611705, | |
| "learning_rate": 6.197528378324664e-06, | |
| "loss": 1.9168, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.12505484861781482, | |
| "grad_norm": 0.2832369769565699, | |
| "learning_rate": 6.003380510149179e-06, | |
| "loss": 1.9466, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.12637121544537078, | |
| "grad_norm": 0.2730019638072709, | |
| "learning_rate": 5.8110093003238175e-06, | |
| "loss": 1.9697, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.12768758227292673, | |
| "grad_norm": 0.29650610897698637, | |
| "learning_rate": 5.620500265600206e-06, | |
| "loss": 1.9431, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.12900394910048266, | |
| "grad_norm": 0.28216205906101904, | |
| "learning_rate": 5.431938094918132e-06, | |
| "loss": 1.9555, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.1303203159280386, | |
| "grad_norm": 0.26142115592109544, | |
| "learning_rate": 5.245406611757882e-06, | |
| "loss": 1.9546, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.13163668275559456, | |
| "grad_norm": 0.2663430873063966, | |
| "learning_rate": 5.060988736877366e-06, | |
| "loss": 1.9371, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.13295304958315052, | |
| "grad_norm": 0.27203256121384983, | |
| "learning_rate": 4.878766451450451e-06, | |
| "loss": 1.9442, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.13426941641070644, | |
| "grad_norm": 0.2472440118588447, | |
| "learning_rate": 4.698820760623064e-06, | |
| "loss": 1.9397, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.1355857832382624, | |
| "grad_norm": 0.24129115289522723, | |
| "learning_rate": 4.5212316575031325e-06, | |
| "loss": 1.9537, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.13690215006581835, | |
| "grad_norm": 0.24834249614222498, | |
| "learning_rate": 4.346078087600411e-06, | |
| "loss": 1.9582, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.13821851689337428, | |
| "grad_norm": 0.2343448970716797, | |
| "learning_rate": 4.173437913732048e-06, | |
| "loss": 1.9564, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.13953488372093023, | |
| "grad_norm": 0.22977461940610383, | |
| "learning_rate": 4.003387881409397e-06, | |
| "loss": 1.9612, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.14085125054848618, | |
| "grad_norm": 0.2327764341813503, | |
| "learning_rate": 3.836003584721577e-06, | |
| "loss": 1.9613, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1421676173760421, | |
| "grad_norm": 0.2361882693796626, | |
| "learning_rate": 3.6713594327308343e-06, | |
| "loss": 1.9468, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.14348398420359806, | |
| "grad_norm": 0.23347458789212797, | |
| "learning_rate": 3.509528616394716e-06, | |
| "loss": 1.9658, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.14480035103115402, | |
| "grad_norm": 0.22028219685938538, | |
| "learning_rate": 3.3505830760297543e-06, | |
| "loss": 1.9621, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.14611671785870997, | |
| "grad_norm": 0.21636698387704192, | |
| "learning_rate": 3.1945934693310897e-06, | |
| "loss": 1.9693, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.1474330846862659, | |
| "grad_norm": 0.22272439284871326, | |
| "learning_rate": 3.0416291399622834e-06, | |
| "loss": 1.9952, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.14874945151382185, | |
| "grad_norm": 0.2240698735197302, | |
| "learning_rate": 2.891758086729253e-06, | |
| "loss": 1.9484, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1500658183413778, | |
| "grad_norm": 0.23983098594113533, | |
| "learning_rate": 2.7450469333520856e-06, | |
| "loss": 1.9546, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.15138218516893373, | |
| "grad_norm": 0.210678907298948, | |
| "learning_rate": 2.6015608988480956e-06, | |
| "loss": 1.9598, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.15269855199648968, | |
| "grad_norm": 0.2094708658742722, | |
| "learning_rate": 2.4613637685393433e-06, | |
| "loss": 1.9693, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.15401491882404564, | |
| "grad_norm": 0.2086143485980995, | |
| "learning_rate": 2.324517865697501e-06, | |
| "loss": 1.9438, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.1553312856516016, | |
| "grad_norm": 0.21377991371943222, | |
| "learning_rate": 2.19108402383864e-06, | |
| "loss": 1.9576, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.15664765247915752, | |
| "grad_norm": 0.205753248097387, | |
| "learning_rate": 2.06112155968028e-06, | |
| "loss": 1.9772, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.15796401930671347, | |
| "grad_norm": 0.21192191289793974, | |
| "learning_rate": 1.9346882467727323e-06, | |
| "loss": 1.9826, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.15928038613426942, | |
| "grad_norm": 0.21524883648102194, | |
| "learning_rate": 1.811840289816409e-06, | |
| "loss": 1.976, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.16059675296182535, | |
| "grad_norm": 0.20625990022421095, | |
| "learning_rate": 1.6926322996765899e-06, | |
| "loss": 1.9776, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.1619131197893813, | |
| "grad_norm": 0.20310346421796105, | |
| "learning_rate": 1.5771172691066793e-06, | |
| "loss": 1.9817, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.16322948661693726, | |
| "grad_norm": 0.208622897847129, | |
| "learning_rate": 1.4653465491908003e-06, | |
| "loss": 2.0059, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.1645458534444932, | |
| "grad_norm": 0.21517659681508963, | |
| "learning_rate": 1.3573698265161683e-06, | |
| "loss": 1.9617, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.16586222027204914, | |
| "grad_norm": 0.20363550281699117, | |
| "learning_rate": 1.2532351010853916e-06, | |
| "loss": 1.9674, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.1671785870996051, | |
| "grad_norm": 0.2108213644456224, | |
| "learning_rate": 1.152988664978556e-06, | |
| "loss": 1.9756, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.16849495392716105, | |
| "grad_norm": 0.199640605228958, | |
| "learning_rate": 1.0566750817745076e-06, | |
| "loss": 1.9827, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.16981132075471697, | |
| "grad_norm": 0.19250372450550102, | |
| "learning_rate": 9.6433716674057e-07, | |
| "loss": 1.9753, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.17112768758227292, | |
| "grad_norm": 0.2075516417455384, | |
| "learning_rate": 8.760159677994174e-07, | |
| "loss": 1.98, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.17244405440982888, | |
| "grad_norm": 0.20463184448298366, | |
| "learning_rate": 7.91750747281621e-07, | |
| "loss": 1.9715, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1737604212373848, | |
| "grad_norm": 0.20224889675782617, | |
| "learning_rate": 7.115789644719728e-07, | |
| "loss": 1.9926, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.17507678806494076, | |
| "grad_norm": 0.19807622313564802, | |
| "learning_rate": 6.355362589573078e-07, | |
| "loss": 1.9738, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.1763931548924967, | |
| "grad_norm": 0.20715058257872104, | |
| "learning_rate": 5.636564347832907e-07, | |
| "loss": 1.9795, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.17770952172005267, | |
| "grad_norm": 0.1987868248478531, | |
| "learning_rate": 4.95971445427137e-07, | |
| "loss": 1.9809, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.1790258885476086, | |
| "grad_norm": 0.19761659136898976, | |
| "learning_rate": 4.3251137959302023e-07, | |
| "loss": 1.9689, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.18034225537516455, | |
| "grad_norm": 0.1979731497305122, | |
| "learning_rate": 3.733044478364234e-07, | |
| "loss": 1.9649, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1816586222027205, | |
| "grad_norm": 0.19563118558862347, | |
| "learning_rate": 3.1837697002341293e-07, | |
| "loss": 1.9773, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.18297498903027642, | |
| "grad_norm": 0.19291494254497765, | |
| "learning_rate": 2.677533636303964e-07, | |
| "loss": 1.9676, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.18429135585783238, | |
| "grad_norm": 0.195403305387489, | |
| "learning_rate": 2.214561328895748e-07, | |
| "loss": 1.9723, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.18560772268538833, | |
| "grad_norm": 0.20351297594650425, | |
| "learning_rate": 1.7950585878489856e-07, | |
| "loss": 1.9773, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.18692408951294429, | |
| "grad_norm": 0.20075344659988453, | |
| "learning_rate": 1.419211899029971e-07, | |
| "loss": 1.969, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.1882404563405002, | |
| "grad_norm": 0.19473013664735456, | |
| "learning_rate": 1.0871883414312778e-07, | |
| "loss": 1.9798, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.18955682316805617, | |
| "grad_norm": 0.19013570239010694, | |
| "learning_rate": 7.99135512898408e-08, | |
| "loss": 1.9716, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.19087318999561212, | |
| "grad_norm": 0.19366203327357978, | |
| "learning_rate": 5.55181464516652e-08, | |
| "loss": 1.9733, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.19218955682316805, | |
| "grad_norm": 0.19674377568766022, | |
| "learning_rate": 3.554346436871581e-08, | |
| "loss": 1.969, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.193505923650724, | |
| "grad_norm": 0.19587314824937796, | |
| "learning_rate": 1.9998384591773945e-08, | |
| "loss": 1.9775, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.19482229047827995, | |
| "grad_norm": 0.1926937742324079, | |
| "learning_rate": 8.889817534969425e-09, | |
| "loss": 1.9853, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.1961386573058359, | |
| "grad_norm": 0.19720577027547917, | |
| "learning_rate": 2.222701403818972e-09, | |
| "loss": 1.9742, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.19745502413339183, | |
| "grad_norm": 0.2111140853212857, | |
| "learning_rate": 0.0, | |
| "loss": 1.9861, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.19745502413339183, | |
| "step": 150, | |
| "total_flos": 349677715193856.0, | |
| "train_loss": 1.9152453955014546, | |
| "train_runtime": 5213.1367, | |
| "train_samples_per_second": 58.007, | |
| "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 | |
| } | |