diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,8440 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 0.999954960229883, + "eval_steps": 100, + "global_step": 3469, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0008647635862466558, + "grad_norm": 27.898378372192383, + "learning_rate": 1.7241379310344828e-07, + "loss": 2.2211, + "step": 3 + }, + { + "epoch": 0.0017295271724933115, + "grad_norm": 25.436155319213867, + "learning_rate": 3.4482758620689656e-07, + "loss": 2.1928, + "step": 6 + }, + { + "epoch": 0.0025942907587399675, + "grad_norm": 25.42555046081543, + "learning_rate": 5.172413793103449e-07, + "loss": 2.006, + "step": 9 + }, + { + "epoch": 0.003459054344986623, + "grad_norm": 26.934585571289062, + "learning_rate": 6.896551724137931e-07, + "loss": 2.0768, + "step": 12 + }, + { + "epoch": 0.004323817931233279, + "grad_norm": 25.911136627197266, + "learning_rate": 8.620689655172415e-07, + "loss": 1.8806, + "step": 15 + }, + { + "epoch": 0.005188581517479935, + "grad_norm": 26.93269157409668, + "learning_rate": 1.0344827586206898e-06, + "loss": 2.2425, + "step": 18 + }, + { + "epoch": 0.006053345103726591, + "grad_norm": 29.343605041503906, + "learning_rate": 1.2068965517241381e-06, + "loss": 2.1498, + "step": 21 + }, + { + "epoch": 0.006918108689973246, + "grad_norm": 22.887393951416016, + "learning_rate": 1.3793103448275862e-06, + "loss": 1.8932, + "step": 24 + }, + { + "epoch": 0.0077828722762199026, + "grad_norm": 24.839616775512695, + "learning_rate": 1.5517241379310346e-06, + "loss": 1.8092, + "step": 27 + }, + { + "epoch": 0.008647635862466557, + "grad_norm": 23.600284576416016, + "learning_rate": 1.724137931034483e-06, + "loss": 1.6473, + "step": 30 + }, + { + "epoch": 0.009512399448713214, + "grad_norm": 20.898502349853516, + "learning_rate": 1.896551724137931e-06, + "loss": 1.5059, + "step": 33 + }, + { + "epoch": 0.01037716303495987, + "grad_norm": 20.16602897644043, + "learning_rate": 2.0689655172413796e-06, + "loss": 1.3399, + "step": 36 + }, + { + "epoch": 0.011241926621206525, + "grad_norm": 20.394386291503906, + "learning_rate": 2.241379310344828e-06, + "loss": 1.276, + "step": 39 + }, + { + "epoch": 0.012106690207453181, + "grad_norm": 17.962718963623047, + "learning_rate": 2.4137931034482762e-06, + "loss": 1.2166, + "step": 42 + }, + { + "epoch": 0.012971453793699838, + "grad_norm": 16.842355728149414, + "learning_rate": 2.5862068965517246e-06, + "loss": 1.2867, + "step": 45 + }, + { + "epoch": 0.013836217379946492, + "grad_norm": 16.04344940185547, + "learning_rate": 2.7586206896551725e-06, + "loss": 1.2473, + "step": 48 + }, + { + "epoch": 0.014700980966193149, + "grad_norm": 10.40717887878418, + "learning_rate": 2.931034482758621e-06, + "loss": 1.1712, + "step": 51 + }, + { + "epoch": 0.015565744552439805, + "grad_norm": 9.352970123291016, + "learning_rate": 3.103448275862069e-06, + "loss": 0.8918, + "step": 54 + }, + { + "epoch": 0.01643050813868646, + "grad_norm": 15.546124458312988, + "learning_rate": 3.2758620689655175e-06, + "loss": 0.8335, + "step": 57 + }, + { + "epoch": 0.017295271724933114, + "grad_norm": 1.752021074295044, + "learning_rate": 3.448275862068966e-06, + "loss": 0.747, + "step": 60 + }, + { + "epoch": 0.01816003531117977, + "grad_norm": 4.960226058959961, + "learning_rate": 3.620689655172414e-06, + "loss": 0.7935, + "step": 63 + }, + { + "epoch": 0.019024798897426427, + "grad_norm": 2.807642698287964, + "learning_rate": 3.793103448275862e-06, + "loss": 0.6656, + "step": 66 + }, + { + "epoch": 0.019889562483673084, + "grad_norm": 8.289933204650879, + "learning_rate": 3.96551724137931e-06, + "loss": 0.6971, + "step": 69 + }, + { + "epoch": 0.02075432606991974, + "grad_norm": 3.8299474716186523, + "learning_rate": 4.137931034482759e-06, + "loss": 0.5833, + "step": 72 + }, + { + "epoch": 0.021619089656166397, + "grad_norm": 3.540727138519287, + "learning_rate": 4.310344827586207e-06, + "loss": 0.5504, + "step": 75 + }, + { + "epoch": 0.02248385324241305, + "grad_norm": 4.521378517150879, + "learning_rate": 4.482758620689656e-06, + "loss": 0.5376, + "step": 78 + }, + { + "epoch": 0.023348616828659706, + "grad_norm": 1.7125041484832764, + "learning_rate": 4.655172413793104e-06, + "loss": 0.5059, + "step": 81 + }, + { + "epoch": 0.024213380414906362, + "grad_norm": 4.4612555503845215, + "learning_rate": 4.8275862068965525e-06, + "loss": 0.5802, + "step": 84 + }, + { + "epoch": 0.02507814400115302, + "grad_norm": 2.3113410472869873, + "learning_rate": 5e-06, + "loss": 0.5996, + "step": 87 + }, + { + "epoch": 0.025942907587399675, + "grad_norm": 2.3540542125701904, + "learning_rate": 5.172413793103449e-06, + "loss": 0.6472, + "step": 90 + }, + { + "epoch": 0.026807671173646328, + "grad_norm": 2.9415371417999268, + "learning_rate": 5.344827586206896e-06, + "loss": 0.4013, + "step": 93 + }, + { + "epoch": 0.027672434759892985, + "grad_norm": 3.8324429988861084, + "learning_rate": 5.517241379310345e-06, + "loss": 0.521, + "step": 96 + }, + { + "epoch": 0.02853719834613964, + "grad_norm": 2.941765785217285, + "learning_rate": 5.689655172413794e-06, + "loss": 0.4763, + "step": 99 + }, + { + "epoch": 0.028825452874888528, + "eval_loss": 0.4467589259147644, + "eval_mse": 0.4467589111328125, + "eval_runtime": 64.5355, + "eval_samples_per_second": 15.495, + "eval_steps_per_second": 3.874, + "step": 100 + }, + { + "epoch": 0.029401961932386297, + "grad_norm": 5.631622314453125, + "learning_rate": 5.862068965517242e-06, + "loss": 0.5672, + "step": 102 + }, + { + "epoch": 0.030266725518632954, + "grad_norm": 2.838885545730591, + "learning_rate": 6.03448275862069e-06, + "loss": 0.5572, + "step": 105 + }, + { + "epoch": 0.03113148910487961, + "grad_norm": 4.942019462585449, + "learning_rate": 6.206896551724138e-06, + "loss": 0.4501, + "step": 108 + }, + { + "epoch": 0.03199625269112626, + "grad_norm": 2.720856189727783, + "learning_rate": 6.379310344827587e-06, + "loss": 0.385, + "step": 111 + }, + { + "epoch": 0.03286101627737292, + "grad_norm": 14.2058744430542, + "learning_rate": 6.551724137931035e-06, + "loss": 0.4864, + "step": 114 + }, + { + "epoch": 0.033725779863619576, + "grad_norm": 3.2831430435180664, + "learning_rate": 6.724137931034484e-06, + "loss": 0.4978, + "step": 117 + }, + { + "epoch": 0.03459054344986623, + "grad_norm": 5.037907600402832, + "learning_rate": 6.896551724137932e-06, + "loss": 0.4359, + "step": 120 + }, + { + "epoch": 0.03545530703611289, + "grad_norm": 3.339016914367676, + "learning_rate": 7.0689655172413796e-06, + "loss": 0.4092, + "step": 123 + }, + { + "epoch": 0.03632007062235954, + "grad_norm": 1.9857500791549683, + "learning_rate": 7.241379310344828e-06, + "loss": 0.4031, + "step": 126 + }, + { + "epoch": 0.0371848342086062, + "grad_norm": 66.43729400634766, + "learning_rate": 7.413793103448277e-06, + "loss": 0.3927, + "step": 129 + }, + { + "epoch": 0.038049597794852855, + "grad_norm": 14.006072998046875, + "learning_rate": 7.586206896551724e-06, + "loss": 0.4676, + "step": 132 + }, + { + "epoch": 0.03891436138109951, + "grad_norm": 3.6296005249023438, + "learning_rate": 7.758620689655173e-06, + "loss": 0.3862, + "step": 135 + }, + { + "epoch": 0.03977912496734617, + "grad_norm": 9.856642723083496, + "learning_rate": 7.93103448275862e-06, + "loss": 0.3564, + "step": 138 + }, + { + "epoch": 0.04064388855359282, + "grad_norm": 21.199617385864258, + "learning_rate": 8.103448275862069e-06, + "loss": 0.3968, + "step": 141 + }, + { + "epoch": 0.04150865213983948, + "grad_norm": 3.6782989501953125, + "learning_rate": 8.275862068965518e-06, + "loss": 0.4045, + "step": 144 + }, + { + "epoch": 0.04237341572608613, + "grad_norm": 8.13273811340332, + "learning_rate": 8.448275862068966e-06, + "loss": 0.3674, + "step": 147 + }, + { + "epoch": 0.04323817931233279, + "grad_norm": 5.4698166847229, + "learning_rate": 8.620689655172414e-06, + "loss": 0.3479, + "step": 150 + }, + { + "epoch": 0.044102942898579446, + "grad_norm": 3.1589951515197754, + "learning_rate": 8.793103448275862e-06, + "loss": 0.3776, + "step": 153 + }, + { + "epoch": 0.0449677064848261, + "grad_norm": 2.283618211746216, + "learning_rate": 8.965517241379312e-06, + "loss": 0.3927, + "step": 156 + }, + { + "epoch": 0.04583247007107276, + "grad_norm": 8.170278549194336, + "learning_rate": 9.13793103448276e-06, + "loss": 0.3583, + "step": 159 + }, + { + "epoch": 0.04669723365731941, + "grad_norm": 3.574162483215332, + "learning_rate": 9.310344827586207e-06, + "loss": 0.3659, + "step": 162 + }, + { + "epoch": 0.04756199724356607, + "grad_norm": 12.687560081481934, + "learning_rate": 9.482758620689655e-06, + "loss": 0.3439, + "step": 165 + }, + { + "epoch": 0.048426760829812725, + "grad_norm": 3.187288761138916, + "learning_rate": 9.655172413793105e-06, + "loss": 0.3442, + "step": 168 + }, + { + "epoch": 0.04929152441605938, + "grad_norm": 9.385987281799316, + "learning_rate": 9.827586206896553e-06, + "loss": 0.3494, + "step": 171 + }, + { + "epoch": 0.05015628800230604, + "grad_norm": 2.9266576766967773, + "learning_rate": 1e-05, + "loss": 0.3061, + "step": 174 + }, + { + "epoch": 0.05102105158855269, + "grad_norm": 3.7128305435180664, + "learning_rate": 9.990895295902884e-06, + "loss": 0.3664, + "step": 177 + }, + { + "epoch": 0.05188581517479935, + "grad_norm": 5.970412731170654, + "learning_rate": 9.981790591805767e-06, + "loss": 0.4051, + "step": 180 + }, + { + "epoch": 0.052750578761046, + "grad_norm": 4.895346164703369, + "learning_rate": 9.972685887708651e-06, + "loss": 0.2907, + "step": 183 + }, + { + "epoch": 0.053615342347292656, + "grad_norm": 16.259010314941406, + "learning_rate": 9.963581183611534e-06, + "loss": 0.303, + "step": 186 + }, + { + "epoch": 0.054480105933539316, + "grad_norm": 4.193286418914795, + "learning_rate": 9.954476479514417e-06, + "loss": 0.3164, + "step": 189 + }, + { + "epoch": 0.05534486951978597, + "grad_norm": 8.485808372497559, + "learning_rate": 9.9453717754173e-06, + "loss": 0.3494, + "step": 192 + }, + { + "epoch": 0.05620963310603263, + "grad_norm": 5.708987712860107, + "learning_rate": 9.936267071320182e-06, + "loss": 0.3413, + "step": 195 + }, + { + "epoch": 0.05707439669227928, + "grad_norm": 5.146287441253662, + "learning_rate": 9.927162367223067e-06, + "loss": 0.3078, + "step": 198 + }, + { + "epoch": 0.057650905749777055, + "eval_loss": 0.31299448013305664, + "eval_mse": 0.31299447250366214, + "eval_runtime": 64.6715, + "eval_samples_per_second": 15.463, + "eval_steps_per_second": 3.866, + "step": 200 + }, + { + "epoch": 0.05793916027852594, + "grad_norm": 6.595048427581787, + "learning_rate": 9.91805766312595e-06, + "loss": 0.3819, + "step": 201 + }, + { + "epoch": 0.058803923864772595, + "grad_norm": 4.586561679840088, + "learning_rate": 9.908952959028833e-06, + "loss": 0.3159, + "step": 204 + }, + { + "epoch": 0.05966868745101925, + "grad_norm": 5.183395862579346, + "learning_rate": 9.899848254931715e-06, + "loss": 0.3129, + "step": 207 + }, + { + "epoch": 0.06053345103726591, + "grad_norm": 12.04688549041748, + "learning_rate": 9.890743550834598e-06, + "loss": 0.3276, + "step": 210 + }, + { + "epoch": 0.06139821462351256, + "grad_norm": 5.607716083526611, + "learning_rate": 9.881638846737481e-06, + "loss": 0.3125, + "step": 213 + }, + { + "epoch": 0.06226297820975922, + "grad_norm": 4.468930244445801, + "learning_rate": 9.872534142640366e-06, + "loss": 0.2952, + "step": 216 + }, + { + "epoch": 0.06312774179600587, + "grad_norm": 4.9333415031433105, + "learning_rate": 9.863429438543249e-06, + "loss": 0.3156, + "step": 219 + }, + { + "epoch": 0.06399250538225253, + "grad_norm": 3.3796207904815674, + "learning_rate": 9.854324734446131e-06, + "loss": 0.3484, + "step": 222 + }, + { + "epoch": 0.06485726896849918, + "grad_norm": 2.9608285427093506, + "learning_rate": 9.845220030349014e-06, + "loss": 0.3244, + "step": 225 + }, + { + "epoch": 0.06572203255474585, + "grad_norm": 2.9653444290161133, + "learning_rate": 9.836115326251897e-06, + "loss": 0.2834, + "step": 228 + }, + { + "epoch": 0.0665867961409925, + "grad_norm": 6.536613464355469, + "learning_rate": 9.827010622154782e-06, + "loss": 0.2884, + "step": 231 + }, + { + "epoch": 0.06745155972723915, + "grad_norm": 4.19090461730957, + "learning_rate": 9.817905918057664e-06, + "loss": 0.3355, + "step": 234 + }, + { + "epoch": 0.0683163233134858, + "grad_norm": 3.4930806159973145, + "learning_rate": 9.808801213960547e-06, + "loss": 0.3038, + "step": 237 + }, + { + "epoch": 0.06918108689973246, + "grad_norm": 13.04383659362793, + "learning_rate": 9.79969650986343e-06, + "loss": 0.2971, + "step": 240 + }, + { + "epoch": 0.07004585048597912, + "grad_norm": 6.37193489074707, + "learning_rate": 9.790591805766313e-06, + "loss": 0.2927, + "step": 243 + }, + { + "epoch": 0.07091061407222578, + "grad_norm": 3.0384926795959473, + "learning_rate": 9.781487101669198e-06, + "loss": 0.3084, + "step": 246 + }, + { + "epoch": 0.07177537765847243, + "grad_norm": 20.000625610351562, + "learning_rate": 9.77238239757208e-06, + "loss": 0.2508, + "step": 249 + }, + { + "epoch": 0.07264014124471908, + "grad_norm": 5.248627662658691, + "learning_rate": 9.763277693474963e-06, + "loss": 0.2514, + "step": 252 + }, + { + "epoch": 0.07350490483096574, + "grad_norm": 6.290755748748779, + "learning_rate": 9.754172989377846e-06, + "loss": 0.2514, + "step": 255 + }, + { + "epoch": 0.0743696684172124, + "grad_norm": 5.56666898727417, + "learning_rate": 9.745068285280729e-06, + "loss": 0.3, + "step": 258 + }, + { + "epoch": 0.07523443200345906, + "grad_norm": 3.098717212677002, + "learning_rate": 9.735963581183613e-06, + "loss": 0.2927, + "step": 261 + }, + { + "epoch": 0.07609919558970571, + "grad_norm": 3.6719772815704346, + "learning_rate": 9.726858877086496e-06, + "loss": 0.3462, + "step": 264 + }, + { + "epoch": 0.07696395917595236, + "grad_norm": 16.087696075439453, + "learning_rate": 9.717754172989379e-06, + "loss": 0.2776, + "step": 267 + }, + { + "epoch": 0.07782872276219902, + "grad_norm": 3.008305788040161, + "learning_rate": 9.708649468892262e-06, + "loss": 0.3359, + "step": 270 + }, + { + "epoch": 0.07869348634844568, + "grad_norm": 5.651501655578613, + "learning_rate": 9.699544764795145e-06, + "loss": 0.3544, + "step": 273 + }, + { + "epoch": 0.07955824993469233, + "grad_norm": 3.0839080810546875, + "learning_rate": 9.690440060698028e-06, + "loss": 0.3059, + "step": 276 + }, + { + "epoch": 0.08042301352093899, + "grad_norm": 6.679815292358398, + "learning_rate": 9.681335356600912e-06, + "loss": 0.2823, + "step": 279 + }, + { + "epoch": 0.08128777710718564, + "grad_norm": 3.161733627319336, + "learning_rate": 9.672230652503795e-06, + "loss": 0.2924, + "step": 282 + }, + { + "epoch": 0.0821525406934323, + "grad_norm": 7.885376930236816, + "learning_rate": 9.663125948406678e-06, + "loss": 0.3316, + "step": 285 + }, + { + "epoch": 0.08301730427967896, + "grad_norm": 4.813638687133789, + "learning_rate": 9.65402124430956e-06, + "loss": 0.2827, + "step": 288 + }, + { + "epoch": 0.08388206786592561, + "grad_norm": 3.7599964141845703, + "learning_rate": 9.644916540212444e-06, + "loss": 0.3653, + "step": 291 + }, + { + "epoch": 0.08474683145217227, + "grad_norm": 3.177720546722412, + "learning_rate": 9.635811836115328e-06, + "loss": 0.3014, + "step": 294 + }, + { + "epoch": 0.08561159503841892, + "grad_norm": 3.337841749191284, + "learning_rate": 9.626707132018211e-06, + "loss": 0.2627, + "step": 297 + }, + { + "epoch": 0.08647635862466559, + "grad_norm": 6.686293601989746, + "learning_rate": 9.617602427921094e-06, + "loss": 0.3088, + "step": 300 + }, + { + "epoch": 0.08647635862466559, + "eval_loss": 0.26950159668922424, + "eval_mse": 0.2695015795826912, + "eval_runtime": 64.3244, + "eval_samples_per_second": 15.546, + "eval_steps_per_second": 3.887, + "step": 300 + }, + { + "epoch": 0.08734112221091224, + "grad_norm": 4.119994163513184, + "learning_rate": 9.608497723823977e-06, + "loss": 0.2565, + "step": 303 + }, + { + "epoch": 0.08820588579715889, + "grad_norm": 2.523799419403076, + "learning_rate": 9.59939301972686e-06, + "loss": 0.2987, + "step": 306 + }, + { + "epoch": 0.08907064938340555, + "grad_norm": 4.713862419128418, + "learning_rate": 9.590288315629744e-06, + "loss": 0.2668, + "step": 309 + }, + { + "epoch": 0.0899354129696522, + "grad_norm": 9.041470527648926, + "learning_rate": 9.581183611532627e-06, + "loss": 0.2889, + "step": 312 + }, + { + "epoch": 0.09080017655589886, + "grad_norm": 2.4231677055358887, + "learning_rate": 9.57207890743551e-06, + "loss": 0.3115, + "step": 315 + }, + { + "epoch": 0.09166494014214552, + "grad_norm": 2.21040678024292, + "learning_rate": 9.562974203338393e-06, + "loss": 0.2281, + "step": 318 + }, + { + "epoch": 0.09252970372839217, + "grad_norm": 2.1169981956481934, + "learning_rate": 9.553869499241275e-06, + "loss": 0.2577, + "step": 321 + }, + { + "epoch": 0.09339446731463882, + "grad_norm": 2.232067823410034, + "learning_rate": 9.54476479514416e-06, + "loss": 0.2205, + "step": 324 + }, + { + "epoch": 0.09425923090088548, + "grad_norm": 6.556311130523682, + "learning_rate": 9.535660091047043e-06, + "loss": 0.2522, + "step": 327 + }, + { + "epoch": 0.09512399448713214, + "grad_norm": 5.824563980102539, + "learning_rate": 9.526555386949926e-06, + "loss": 0.298, + "step": 330 + }, + { + "epoch": 0.0959887580733788, + "grad_norm": 3.462920665740967, + "learning_rate": 9.517450682852808e-06, + "loss": 0.2836, + "step": 333 + }, + { + "epoch": 0.09685352165962545, + "grad_norm": 8.646162986755371, + "learning_rate": 9.508345978755691e-06, + "loss": 0.2844, + "step": 336 + }, + { + "epoch": 0.0977182852458721, + "grad_norm": 2.7092299461364746, + "learning_rate": 9.499241274658574e-06, + "loss": 0.2869, + "step": 339 + }, + { + "epoch": 0.09858304883211876, + "grad_norm": 2.3480119705200195, + "learning_rate": 9.490136570561459e-06, + "loss": 0.3123, + "step": 342 + }, + { + "epoch": 0.09944781241836542, + "grad_norm": 3.3885226249694824, + "learning_rate": 9.481031866464341e-06, + "loss": 0.2096, + "step": 345 + }, + { + "epoch": 0.10031257600461208, + "grad_norm": 4.4023590087890625, + "learning_rate": 9.471927162367224e-06, + "loss": 0.2715, + "step": 348 + }, + { + "epoch": 0.10117733959085873, + "grad_norm": 5.742528438568115, + "learning_rate": 9.462822458270107e-06, + "loss": 0.3155, + "step": 351 + }, + { + "epoch": 0.10204210317710538, + "grad_norm": 5.3731913566589355, + "learning_rate": 9.45371775417299e-06, + "loss": 0.2759, + "step": 354 + }, + { + "epoch": 0.10290686676335203, + "grad_norm": 2.147970199584961, + "learning_rate": 9.444613050075875e-06, + "loss": 0.2481, + "step": 357 + }, + { + "epoch": 0.1037716303495987, + "grad_norm": 2.099414110183716, + "learning_rate": 9.435508345978757e-06, + "loss": 0.3029, + "step": 360 + }, + { + "epoch": 0.10463639393584535, + "grad_norm": 2.6507697105407715, + "learning_rate": 9.42640364188164e-06, + "loss": 0.2731, + "step": 363 + }, + { + "epoch": 0.105501157522092, + "grad_norm": 2.6286544799804688, + "learning_rate": 9.417298937784523e-06, + "loss": 0.2532, + "step": 366 + }, + { + "epoch": 0.10636592110833866, + "grad_norm": 4.705443859100342, + "learning_rate": 9.408194233687406e-06, + "loss": 0.245, + "step": 369 + }, + { + "epoch": 0.10723068469458531, + "grad_norm": 2.616685628890991, + "learning_rate": 9.399089529590289e-06, + "loss": 0.2953, + "step": 372 + }, + { + "epoch": 0.10809544828083198, + "grad_norm": 2.7139439582824707, + "learning_rate": 9.389984825493173e-06, + "loss": 0.2691, + "step": 375 + }, + { + "epoch": 0.10896021186707863, + "grad_norm": 5.848327159881592, + "learning_rate": 9.380880121396056e-06, + "loss": 0.2735, + "step": 378 + }, + { + "epoch": 0.10982497545332529, + "grad_norm": 4.9311957359313965, + "learning_rate": 9.371775417298939e-06, + "loss": 0.2915, + "step": 381 + }, + { + "epoch": 0.11068973903957194, + "grad_norm": 2.206596612930298, + "learning_rate": 9.362670713201822e-06, + "loss": 0.2811, + "step": 384 + }, + { + "epoch": 0.11155450262581859, + "grad_norm": 4.341143608093262, + "learning_rate": 9.353566009104705e-06, + "loss": 0.2845, + "step": 387 + }, + { + "epoch": 0.11241926621206526, + "grad_norm": 144.06268310546875, + "learning_rate": 9.344461305007587e-06, + "loss": 0.3057, + "step": 390 + }, + { + "epoch": 0.11328402979831191, + "grad_norm": 4.772558689117432, + "learning_rate": 9.335356600910472e-06, + "loss": 0.2683, + "step": 393 + }, + { + "epoch": 0.11414879338455856, + "grad_norm": 4.352729320526123, + "learning_rate": 9.326251896813355e-06, + "loss": 0.2707, + "step": 396 + }, + { + "epoch": 0.11501355697080522, + "grad_norm": 2.260861396789551, + "learning_rate": 9.317147192716238e-06, + "loss": 0.2379, + "step": 399 + }, + { + "epoch": 0.11530181149955411, + "eval_loss": 0.26182544231414795, + "eval_mse": 0.26182544356794096, + "eval_runtime": 64.4569, + "eval_samples_per_second": 15.514, + "eval_steps_per_second": 3.879, + "step": 400 + }, + { + "epoch": 0.11587832055705188, + "grad_norm": 4.492468357086182, + "learning_rate": 9.30804248861912e-06, + "loss": 0.2976, + "step": 402 + }, + { + "epoch": 0.11674308414329854, + "grad_norm": 2.6328372955322266, + "learning_rate": 9.298937784522003e-06, + "loss": 0.2224, + "step": 405 + }, + { + "epoch": 0.11760784772954519, + "grad_norm": 4.318969249725342, + "learning_rate": 9.289833080424886e-06, + "loss": 0.2396, + "step": 408 + }, + { + "epoch": 0.11847261131579184, + "grad_norm": 3.1604599952697754, + "learning_rate": 9.28072837632777e-06, + "loss": 0.2492, + "step": 411 + }, + { + "epoch": 0.1193373749020385, + "grad_norm": 4.404653072357178, + "learning_rate": 9.271623672230654e-06, + "loss": 0.3145, + "step": 414 + }, + { + "epoch": 0.12020213848828516, + "grad_norm": 2.884675979614258, + "learning_rate": 9.262518968133536e-06, + "loss": 0.2851, + "step": 417 + }, + { + "epoch": 0.12106690207453182, + "grad_norm": 2.1803503036499023, + "learning_rate": 9.25341426403642e-06, + "loss": 0.2832, + "step": 420 + }, + { + "epoch": 0.12193166566077847, + "grad_norm": 4.99426794052124, + "learning_rate": 9.244309559939302e-06, + "loss": 0.3244, + "step": 423 + }, + { + "epoch": 0.12279642924702512, + "grad_norm": 2.84147310256958, + "learning_rate": 9.235204855842187e-06, + "loss": 0.2886, + "step": 426 + }, + { + "epoch": 0.12366119283327177, + "grad_norm": 7.547422885894775, + "learning_rate": 9.22610015174507e-06, + "loss": 0.3274, + "step": 429 + }, + { + "epoch": 0.12452595641951844, + "grad_norm": 3.3562684059143066, + "learning_rate": 9.216995447647952e-06, + "loss": 0.2726, + "step": 432 + }, + { + "epoch": 0.1253907200057651, + "grad_norm": 5.314599990844727, + "learning_rate": 9.207890743550835e-06, + "loss": 0.2757, + "step": 435 + }, + { + "epoch": 0.12625548359201175, + "grad_norm": 2.0773863792419434, + "learning_rate": 9.198786039453718e-06, + "loss": 0.2342, + "step": 438 + }, + { + "epoch": 0.1271202471782584, + "grad_norm": 4.290255546569824, + "learning_rate": 9.189681335356601e-06, + "loss": 0.3036, + "step": 441 + }, + { + "epoch": 0.12798501076450505, + "grad_norm": 5.141223907470703, + "learning_rate": 9.180576631259485e-06, + "loss": 0.2455, + "step": 444 + }, + { + "epoch": 0.1288497743507517, + "grad_norm": 2.8631341457366943, + "learning_rate": 9.171471927162368e-06, + "loss": 0.229, + "step": 447 + }, + { + "epoch": 0.12971453793699836, + "grad_norm": 6.279902935028076, + "learning_rate": 9.162367223065251e-06, + "loss": 0.2693, + "step": 450 + }, + { + "epoch": 0.130579301523245, + "grad_norm": 1.910251498222351, + "learning_rate": 9.153262518968134e-06, + "loss": 0.2384, + "step": 453 + }, + { + "epoch": 0.1314440651094917, + "grad_norm": 3.9534478187561035, + "learning_rate": 9.144157814871017e-06, + "loss": 0.3125, + "step": 456 + }, + { + "epoch": 0.13230882869573835, + "grad_norm": 4.347681522369385, + "learning_rate": 9.1350531107739e-06, + "loss": 0.3019, + "step": 459 + }, + { + "epoch": 0.133173592281985, + "grad_norm": 2.742349624633789, + "learning_rate": 9.125948406676784e-06, + "loss": 0.2647, + "step": 462 + }, + { + "epoch": 0.13403835586823165, + "grad_norm": 3.5790295600891113, + "learning_rate": 9.116843702579667e-06, + "loss": 0.3299, + "step": 465 + }, + { + "epoch": 0.1349031194544783, + "grad_norm": 2.90129017829895, + "learning_rate": 9.10773899848255e-06, + "loss": 0.2666, + "step": 468 + }, + { + "epoch": 0.13576788304072496, + "grad_norm": 1.934597134590149, + "learning_rate": 9.098634294385433e-06, + "loss": 0.2665, + "step": 471 + }, + { + "epoch": 0.1366326466269716, + "grad_norm": 2.528752088546753, + "learning_rate": 9.089529590288316e-06, + "loss": 0.2386, + "step": 474 + }, + { + "epoch": 0.13749741021321826, + "grad_norm": 3.6109018325805664, + "learning_rate": 9.080424886191198e-06, + "loss": 0.2733, + "step": 477 + }, + { + "epoch": 0.13836217379946492, + "grad_norm": 2.2715227603912354, + "learning_rate": 9.071320182094083e-06, + "loss": 0.2703, + "step": 480 + }, + { + "epoch": 0.1392269373857116, + "grad_norm": 6.626646995544434, + "learning_rate": 9.062215477996966e-06, + "loss": 0.2605, + "step": 483 + }, + { + "epoch": 0.14009170097195825, + "grad_norm": 2.91410756111145, + "learning_rate": 9.053110773899849e-06, + "loss": 0.2382, + "step": 486 + }, + { + "epoch": 0.1409564645582049, + "grad_norm": 4.639560699462891, + "learning_rate": 9.044006069802731e-06, + "loss": 0.2977, + "step": 489 + }, + { + "epoch": 0.14182122814445156, + "grad_norm": 3.643523693084717, + "learning_rate": 9.034901365705614e-06, + "loss": 0.2775, + "step": 492 + }, + { + "epoch": 0.1426859917306982, + "grad_norm": 4.396177291870117, + "learning_rate": 9.025796661608497e-06, + "loss": 0.23, + "step": 495 + }, + { + "epoch": 0.14355075531694486, + "grad_norm": 2.5602617263793945, + "learning_rate": 9.016691957511382e-06, + "loss": 0.289, + "step": 498 + }, + { + "epoch": 0.14412726437444262, + "eval_loss": 0.25831595063209534, + "eval_mse": 0.2583159562349319, + "eval_runtime": 64.3088, + "eval_samples_per_second": 15.55, + "eval_steps_per_second": 3.887, + "step": 500 + }, + { + "epoch": 0.14441551890319151, + "grad_norm": 3.0293095111846924, + "learning_rate": 9.007587253414265e-06, + "loss": 0.2419, + "step": 501 + }, + { + "epoch": 0.14528028248943817, + "grad_norm": 2.9694740772247314, + "learning_rate": 8.998482549317147e-06, + "loss": 0.2689, + "step": 504 + }, + { + "epoch": 0.14614504607568482, + "grad_norm": 4.87199068069458, + "learning_rate": 8.98937784522003e-06, + "loss": 0.2353, + "step": 507 + }, + { + "epoch": 0.14700980966193147, + "grad_norm": 2.5802090167999268, + "learning_rate": 8.980273141122913e-06, + "loss": 0.2865, + "step": 510 + }, + { + "epoch": 0.14787457324817815, + "grad_norm": 2.9250080585479736, + "learning_rate": 8.971168437025798e-06, + "loss": 0.24, + "step": 513 + }, + { + "epoch": 0.1487393368344248, + "grad_norm": 5.678205966949463, + "learning_rate": 8.96206373292868e-06, + "loss": 0.2357, + "step": 516 + }, + { + "epoch": 0.14960410042067146, + "grad_norm": 4.411201477050781, + "learning_rate": 8.952959028831563e-06, + "loss": 0.2605, + "step": 519 + }, + { + "epoch": 0.1504688640069181, + "grad_norm": 3.1570451259613037, + "learning_rate": 8.943854324734446e-06, + "loss": 0.2841, + "step": 522 + }, + { + "epoch": 0.15133362759316477, + "grad_norm": 4.235450744628906, + "learning_rate": 8.934749620637329e-06, + "loss": 0.2675, + "step": 525 + }, + { + "epoch": 0.15219839117941142, + "grad_norm": 5.139553070068359, + "learning_rate": 8.925644916540213e-06, + "loss": 0.2944, + "step": 528 + }, + { + "epoch": 0.15306315476565807, + "grad_norm": 2.4079232215881348, + "learning_rate": 8.916540212443096e-06, + "loss": 0.2169, + "step": 531 + }, + { + "epoch": 0.15392791835190472, + "grad_norm": 5.065981388092041, + "learning_rate": 8.90743550834598e-06, + "loss": 0.306, + "step": 534 + }, + { + "epoch": 0.15479268193815138, + "grad_norm": 2.997144937515259, + "learning_rate": 8.898330804248862e-06, + "loss": 0.2389, + "step": 537 + }, + { + "epoch": 0.15565744552439803, + "grad_norm": 6.104944705963135, + "learning_rate": 8.889226100151745e-06, + "loss": 0.258, + "step": 540 + }, + { + "epoch": 0.1565222091106447, + "grad_norm": 2.971264362335205, + "learning_rate": 8.880121396054628e-06, + "loss": 0.2914, + "step": 543 + }, + { + "epoch": 0.15738697269689136, + "grad_norm": 14.179126739501953, + "learning_rate": 8.871016691957512e-06, + "loss": 0.2632, + "step": 546 + }, + { + "epoch": 0.15825173628313802, + "grad_norm": 1.9280515909194946, + "learning_rate": 8.861911987860395e-06, + "loss": 0.2743, + "step": 549 + }, + { + "epoch": 0.15911649986938467, + "grad_norm": 9.689038276672363, + "learning_rate": 8.852807283763278e-06, + "loss": 0.2148, + "step": 552 + }, + { + "epoch": 0.15998126345563132, + "grad_norm": 2.9501137733459473, + "learning_rate": 8.84370257966616e-06, + "loss": 0.251, + "step": 555 + }, + { + "epoch": 0.16084602704187798, + "grad_norm": 6.09636116027832, + "learning_rate": 8.834597875569044e-06, + "loss": 0.3178, + "step": 558 + }, + { + "epoch": 0.16171079062812463, + "grad_norm": 3.230820894241333, + "learning_rate": 8.825493171471928e-06, + "loss": 0.2574, + "step": 561 + }, + { + "epoch": 0.16257555421437128, + "grad_norm": 8.147329330444336, + "learning_rate": 8.816388467374811e-06, + "loss": 0.3363, + "step": 564 + }, + { + "epoch": 0.16344031780061793, + "grad_norm": 6.0211100578308105, + "learning_rate": 8.807283763277694e-06, + "loss": 0.2568, + "step": 567 + }, + { + "epoch": 0.1643050813868646, + "grad_norm": 3.558443307876587, + "learning_rate": 8.798179059180577e-06, + "loss": 0.2346, + "step": 570 + }, + { + "epoch": 0.16516984497311127, + "grad_norm": 6.004063129425049, + "learning_rate": 8.78907435508346e-06, + "loss": 0.28, + "step": 573 + }, + { + "epoch": 0.16603460855935792, + "grad_norm": 3.2006189823150635, + "learning_rate": 8.779969650986344e-06, + "loss": 0.2356, + "step": 576 + }, + { + "epoch": 0.16689937214560457, + "grad_norm": 3.382519245147705, + "learning_rate": 8.770864946889227e-06, + "loss": 0.2565, + "step": 579 + }, + { + "epoch": 0.16776413573185123, + "grad_norm": 3.465315580368042, + "learning_rate": 8.76176024279211e-06, + "loss": 0.2757, + "step": 582 + }, + { + "epoch": 0.16862889931809788, + "grad_norm": 3.368386745452881, + "learning_rate": 8.752655538694993e-06, + "loss": 0.2663, + "step": 585 + }, + { + "epoch": 0.16949366290434453, + "grad_norm": 5.74733304977417, + "learning_rate": 8.743550834597875e-06, + "loss": 0.2539, + "step": 588 + }, + { + "epoch": 0.17035842649059119, + "grad_norm": 2.7959225177764893, + "learning_rate": 8.73444613050076e-06, + "loss": 0.232, + "step": 591 + }, + { + "epoch": 0.17122319007683784, + "grad_norm": 6.77617073059082, + "learning_rate": 8.725341426403643e-06, + "loss": 0.2858, + "step": 594 + }, + { + "epoch": 0.1720879536630845, + "grad_norm": 1.6699326038360596, + "learning_rate": 8.716236722306526e-06, + "loss": 0.2202, + "step": 597 + }, + { + "epoch": 0.17295271724933117, + "grad_norm": 7.734217166900635, + "learning_rate": 8.707132018209408e-06, + "loss": 0.3049, + "step": 600 + }, + { + "epoch": 0.17295271724933117, + "eval_loss": 0.27226680517196655, + "eval_mse": 0.27226682132482527, + "eval_runtime": 64.8839, + "eval_samples_per_second": 15.412, + "eval_steps_per_second": 3.853, + "step": 600 + }, + { + "epoch": 0.17381748083557783, + "grad_norm": 2.729203224182129, + "learning_rate": 8.698027314112291e-06, + "loss": 0.2705, + "step": 603 + }, + { + "epoch": 0.17468224442182448, + "grad_norm": 3.3243207931518555, + "learning_rate": 8.688922610015174e-06, + "loss": 0.2685, + "step": 606 + }, + { + "epoch": 0.17554700800807113, + "grad_norm": 3.1003451347351074, + "learning_rate": 8.679817905918059e-06, + "loss": 0.2483, + "step": 609 + }, + { + "epoch": 0.17641177159431778, + "grad_norm": 5.373556613922119, + "learning_rate": 8.670713201820942e-06, + "loss": 0.2593, + "step": 612 + }, + { + "epoch": 0.17727653518056444, + "grad_norm": 5.082779407501221, + "learning_rate": 8.661608497723824e-06, + "loss": 0.2755, + "step": 615 + }, + { + "epoch": 0.1781412987668111, + "grad_norm": 5.2915802001953125, + "learning_rate": 8.652503793626707e-06, + "loss": 0.2428, + "step": 618 + }, + { + "epoch": 0.17900606235305774, + "grad_norm": 4.1451568603515625, + "learning_rate": 8.64339908952959e-06, + "loss": 0.3072, + "step": 621 + }, + { + "epoch": 0.1798708259393044, + "grad_norm": 5.081033229827881, + "learning_rate": 8.634294385432475e-06, + "loss": 0.2515, + "step": 624 + }, + { + "epoch": 0.18073558952555105, + "grad_norm": 3.615787982940674, + "learning_rate": 8.625189681335357e-06, + "loss": 0.2421, + "step": 627 + }, + { + "epoch": 0.18160035311179773, + "grad_norm": 3.4306159019470215, + "learning_rate": 8.61608497723824e-06, + "loss": 0.2378, + "step": 630 + }, + { + "epoch": 0.18246511669804438, + "grad_norm": 6.089697360992432, + "learning_rate": 8.606980273141123e-06, + "loss": 0.2537, + "step": 633 + }, + { + "epoch": 0.18332988028429104, + "grad_norm": 3.5059478282928467, + "learning_rate": 8.597875569044006e-06, + "loss": 0.2637, + "step": 636 + }, + { + "epoch": 0.1841946438705377, + "grad_norm": 4.7960710525512695, + "learning_rate": 8.58877086494689e-06, + "loss": 0.2519, + "step": 639 + }, + { + "epoch": 0.18505940745678434, + "grad_norm": 2.7048394680023193, + "learning_rate": 8.579666160849773e-06, + "loss": 0.2757, + "step": 642 + }, + { + "epoch": 0.185924171043031, + "grad_norm": 4.179544925689697, + "learning_rate": 8.570561456752656e-06, + "loss": 0.2829, + "step": 645 + }, + { + "epoch": 0.18678893462927765, + "grad_norm": 2.122403383255005, + "learning_rate": 8.561456752655539e-06, + "loss": 0.2636, + "step": 648 + }, + { + "epoch": 0.1876536982155243, + "grad_norm": 7.5383830070495605, + "learning_rate": 8.552352048558422e-06, + "loss": 0.3059, + "step": 651 + }, + { + "epoch": 0.18851846180177095, + "grad_norm": 3.6806488037109375, + "learning_rate": 8.543247344461306e-06, + "loss": 0.2266, + "step": 654 + }, + { + "epoch": 0.1893832253880176, + "grad_norm": 4.183900833129883, + "learning_rate": 8.53414264036419e-06, + "loss": 0.2696, + "step": 657 + }, + { + "epoch": 0.1902479889742643, + "grad_norm": 7.027878761291504, + "learning_rate": 8.525037936267072e-06, + "loss": 0.2292, + "step": 660 + }, + { + "epoch": 0.19111275256051094, + "grad_norm": 6.008081436157227, + "learning_rate": 8.515933232169955e-06, + "loss": 0.2628, + "step": 663 + }, + { + "epoch": 0.1919775161467576, + "grad_norm": 1.9789228439331055, + "learning_rate": 8.506828528072838e-06, + "loss": 0.2328, + "step": 666 + }, + { + "epoch": 0.19284227973300425, + "grad_norm": 4.350831985473633, + "learning_rate": 8.49772382397572e-06, + "loss": 0.2743, + "step": 669 + }, + { + "epoch": 0.1937070433192509, + "grad_norm": 2.4189023971557617, + "learning_rate": 8.488619119878605e-06, + "loss": 0.2549, + "step": 672 + }, + { + "epoch": 0.19457180690549755, + "grad_norm": 2.49760103225708, + "learning_rate": 8.479514415781488e-06, + "loss": 0.2287, + "step": 675 + }, + { + "epoch": 0.1954365704917442, + "grad_norm": 4.735384941101074, + "learning_rate": 8.470409711684371e-06, + "loss": 0.2168, + "step": 678 + }, + { + "epoch": 0.19630133407799086, + "grad_norm": 1.813238263130188, + "learning_rate": 8.461305007587254e-06, + "loss": 0.2536, + "step": 681 + }, + { + "epoch": 0.1971660976642375, + "grad_norm": 2.416673421859741, + "learning_rate": 8.452200303490137e-06, + "loss": 0.2538, + "step": 684 + }, + { + "epoch": 0.19803086125048416, + "grad_norm": 2.3501391410827637, + "learning_rate": 8.443095599393021e-06, + "loss": 0.2682, + "step": 687 + }, + { + "epoch": 0.19889562483673084, + "grad_norm": 4.72876501083374, + "learning_rate": 8.433990895295904e-06, + "loss": 0.2192, + "step": 690 + }, + { + "epoch": 0.1997603884229775, + "grad_norm": 2.0732574462890625, + "learning_rate": 8.424886191198787e-06, + "loss": 0.2612, + "step": 693 + }, + { + "epoch": 0.20062515200922415, + "grad_norm": 3.1769752502441406, + "learning_rate": 8.41578148710167e-06, + "loss": 0.2434, + "step": 696 + }, + { + "epoch": 0.2014899155954708, + "grad_norm": 4.40995454788208, + "learning_rate": 8.406676783004552e-06, + "loss": 0.2292, + "step": 699 + }, + { + "epoch": 0.2017781701242197, + "eval_loss": 0.2476717233657837, + "eval_mse": 0.24767172618419864, + "eval_runtime": 64.3865, + "eval_samples_per_second": 15.531, + "eval_steps_per_second": 3.883, + "step": 700 + }, + { + "epoch": 0.20235467918171746, + "grad_norm": 1.8437422513961792, + "learning_rate": 8.397572078907437e-06, + "loss": 0.2751, + "step": 702 + }, + { + "epoch": 0.2032194427679641, + "grad_norm": 1.9612832069396973, + "learning_rate": 8.38846737481032e-06, + "loss": 0.2322, + "step": 705 + }, + { + "epoch": 0.20408420635421076, + "grad_norm": 1.9508056640625, + "learning_rate": 8.379362670713203e-06, + "loss": 0.2206, + "step": 708 + }, + { + "epoch": 0.20494896994045742, + "grad_norm": 12.081295013427734, + "learning_rate": 8.370257966616086e-06, + "loss": 0.2845, + "step": 711 + }, + { + "epoch": 0.20581373352670407, + "grad_norm": 7.8822479248046875, + "learning_rate": 8.361153262518968e-06, + "loss": 0.2439, + "step": 714 + }, + { + "epoch": 0.20667849711295075, + "grad_norm": 7.18080997467041, + "learning_rate": 8.352048558421853e-06, + "loss": 0.2813, + "step": 717 + }, + { + "epoch": 0.2075432606991974, + "grad_norm": 2.941767692565918, + "learning_rate": 8.342943854324736e-06, + "loss": 0.2384, + "step": 720 + }, + { + "epoch": 0.20840802428544405, + "grad_norm": 3.640935182571411, + "learning_rate": 8.333839150227619e-06, + "loss": 0.2535, + "step": 723 + }, + { + "epoch": 0.2092727878716907, + "grad_norm": 3.2023556232452393, + "learning_rate": 8.324734446130501e-06, + "loss": 0.2822, + "step": 726 + }, + { + "epoch": 0.21013755145793736, + "grad_norm": 4.521758556365967, + "learning_rate": 8.315629742033384e-06, + "loss": 0.2604, + "step": 729 + }, + { + "epoch": 0.211002315044184, + "grad_norm": 2.9931998252868652, + "learning_rate": 8.306525037936269e-06, + "loss": 0.239, + "step": 732 + }, + { + "epoch": 0.21186707863043067, + "grad_norm": 3.8456552028656006, + "learning_rate": 8.297420333839152e-06, + "loss": 0.2256, + "step": 735 + }, + { + "epoch": 0.21273184221667732, + "grad_norm": 3.0459201335906982, + "learning_rate": 8.288315629742034e-06, + "loss": 0.3055, + "step": 738 + }, + { + "epoch": 0.21359660580292397, + "grad_norm": 6.939425468444824, + "learning_rate": 8.279210925644917e-06, + "loss": 0.2945, + "step": 741 + }, + { + "epoch": 0.21446136938917063, + "grad_norm": 4.698008060455322, + "learning_rate": 8.2701062215478e-06, + "loss": 0.2558, + "step": 744 + }, + { + "epoch": 0.2153261329754173, + "grad_norm": 2.6725728511810303, + "learning_rate": 8.261001517450683e-06, + "loss": 0.2487, + "step": 747 + }, + { + "epoch": 0.21619089656166396, + "grad_norm": 3.1390607357025146, + "learning_rate": 8.251896813353568e-06, + "loss": 0.2371, + "step": 750 + }, + { + "epoch": 0.2170556601479106, + "grad_norm": 4.460456371307373, + "learning_rate": 8.24279210925645e-06, + "loss": 0.2595, + "step": 753 + }, + { + "epoch": 0.21792042373415726, + "grad_norm": 7.146320343017578, + "learning_rate": 8.233687405159333e-06, + "loss": 0.2687, + "step": 756 + }, + { + "epoch": 0.21878518732040392, + "grad_norm": 5.645358562469482, + "learning_rate": 8.224582701062216e-06, + "loss": 0.2402, + "step": 759 + }, + { + "epoch": 0.21964995090665057, + "grad_norm": 2.5356500148773193, + "learning_rate": 8.215477996965099e-06, + "loss": 0.2567, + "step": 762 + }, + { + "epoch": 0.22051471449289722, + "grad_norm": 5.814233303070068, + "learning_rate": 8.206373292867983e-06, + "loss": 0.2535, + "step": 765 + }, + { + "epoch": 0.22137947807914388, + "grad_norm": 2.0408260822296143, + "learning_rate": 8.197268588770866e-06, + "loss": 0.254, + "step": 768 + }, + { + "epoch": 0.22224424166539053, + "grad_norm": 2.689192056655884, + "learning_rate": 8.188163884673749e-06, + "loss": 0.2397, + "step": 771 + }, + { + "epoch": 0.22310900525163718, + "grad_norm": 4.501987934112549, + "learning_rate": 8.179059180576632e-06, + "loss": 0.2507, + "step": 774 + }, + { + "epoch": 0.22397376883788386, + "grad_norm": 3.8230247497558594, + "learning_rate": 8.169954476479515e-06, + "loss": 0.253, + "step": 777 + }, + { + "epoch": 0.22483853242413052, + "grad_norm": 4.210086345672607, + "learning_rate": 8.1608497723824e-06, + "loss": 0.231, + "step": 780 + }, + { + "epoch": 0.22570329601037717, + "grad_norm": 3.3682661056518555, + "learning_rate": 8.151745068285282e-06, + "loss": 0.2632, + "step": 783 + }, + { + "epoch": 0.22656805959662382, + "grad_norm": 3.226896286010742, + "learning_rate": 8.142640364188165e-06, + "loss": 0.2293, + "step": 786 + }, + { + "epoch": 0.22743282318287047, + "grad_norm": 1.880353569984436, + "learning_rate": 8.133535660091048e-06, + "loss": 0.2529, + "step": 789 + }, + { + "epoch": 0.22829758676911713, + "grad_norm": 5.078037738800049, + "learning_rate": 8.12443095599393e-06, + "loss": 0.2533, + "step": 792 + }, + { + "epoch": 0.22916235035536378, + "grad_norm": 5.995652198791504, + "learning_rate": 8.115326251896815e-06, + "loss": 0.2685, + "step": 795 + }, + { + "epoch": 0.23002711394161043, + "grad_norm": 4.134960174560547, + "learning_rate": 8.106221547799698e-06, + "loss": 0.2677, + "step": 798 + }, + { + "epoch": 0.23060362299910822, + "eval_loss": 0.23694634437561035, + "eval_mse": 0.23694635881483556, + "eval_runtime": 64.1432, + "eval_samples_per_second": 15.59, + "eval_steps_per_second": 3.898, + "step": 800 + }, + { + "epoch": 0.2308918775278571, + "grad_norm": 2.2709124088287354, + "learning_rate": 8.097116843702581e-06, + "loss": 0.2338, + "step": 801 + }, + { + "epoch": 0.23175664111410377, + "grad_norm": 3.1088995933532715, + "learning_rate": 8.088012139605464e-06, + "loss": 0.2342, + "step": 804 + }, + { + "epoch": 0.23262140470035042, + "grad_norm": 2.4137418270111084, + "learning_rate": 8.078907435508347e-06, + "loss": 0.2467, + "step": 807 + }, + { + "epoch": 0.23348616828659707, + "grad_norm": 14.154044151306152, + "learning_rate": 8.06980273141123e-06, + "loss": 0.2386, + "step": 810 + }, + { + "epoch": 0.23435093187284373, + "grad_norm": 6.077353477478027, + "learning_rate": 8.060698027314114e-06, + "loss": 0.2243, + "step": 813 + }, + { + "epoch": 0.23521569545909038, + "grad_norm": 2.9805984497070312, + "learning_rate": 8.051593323216997e-06, + "loss": 0.2451, + "step": 816 + }, + { + "epoch": 0.23608045904533703, + "grad_norm": 2.8187265396118164, + "learning_rate": 8.04248861911988e-06, + "loss": 0.2922, + "step": 819 + }, + { + "epoch": 0.23694522263158369, + "grad_norm": 2.3969638347625732, + "learning_rate": 8.033383915022763e-06, + "loss": 0.2579, + "step": 822 + }, + { + "epoch": 0.23780998621783034, + "grad_norm": 2.1792314052581787, + "learning_rate": 8.024279210925645e-06, + "loss": 0.2282, + "step": 825 + }, + { + "epoch": 0.238674749804077, + "grad_norm": 2.370007038116455, + "learning_rate": 8.01517450682853e-06, + "loss": 0.291, + "step": 828 + }, + { + "epoch": 0.23953951339032364, + "grad_norm": 1.9614750146865845, + "learning_rate": 8.006069802731413e-06, + "loss": 0.2391, + "step": 831 + }, + { + "epoch": 0.24040427697657032, + "grad_norm": 2.416917324066162, + "learning_rate": 7.996965098634296e-06, + "loss": 0.2546, + "step": 834 + }, + { + "epoch": 0.24126904056281698, + "grad_norm": 2.136915683746338, + "learning_rate": 7.987860394537178e-06, + "loss": 0.2664, + "step": 837 + }, + { + "epoch": 0.24213380414906363, + "grad_norm": 2.0941572189331055, + "learning_rate": 7.978755690440061e-06, + "loss": 0.2597, + "step": 840 + }, + { + "epoch": 0.24299856773531028, + "grad_norm": 3.3543283939361572, + "learning_rate": 7.969650986342944e-06, + "loss": 0.233, + "step": 843 + }, + { + "epoch": 0.24386333132155694, + "grad_norm": 2.2875921726226807, + "learning_rate": 7.960546282245829e-06, + "loss": 0.2302, + "step": 846 + }, + { + "epoch": 0.2447280949078036, + "grad_norm": 2.283102035522461, + "learning_rate": 7.951441578148712e-06, + "loss": 0.2825, + "step": 849 + }, + { + "epoch": 0.24559285849405024, + "grad_norm": 5.802175998687744, + "learning_rate": 7.942336874051594e-06, + "loss": 0.2515, + "step": 852 + }, + { + "epoch": 0.2464576220802969, + "grad_norm": 2.8898258209228516, + "learning_rate": 7.933232169954477e-06, + "loss": 0.2419, + "step": 855 + }, + { + "epoch": 0.24732238566654355, + "grad_norm": 1.9422801733016968, + "learning_rate": 7.92412746585736e-06, + "loss": 0.2361, + "step": 858 + }, + { + "epoch": 0.2481871492527902, + "grad_norm": 3.0359907150268555, + "learning_rate": 7.915022761760245e-06, + "loss": 0.1944, + "step": 861 + }, + { + "epoch": 0.24905191283903688, + "grad_norm": 3.544053554534912, + "learning_rate": 7.905918057663127e-06, + "loss": 0.2938, + "step": 864 + }, + { + "epoch": 0.24991667642528353, + "grad_norm": 2.4866981506347656, + "learning_rate": 7.89681335356601e-06, + "loss": 0.2339, + "step": 867 + }, + { + "epoch": 0.2507814400115302, + "grad_norm": 9.933099746704102, + "learning_rate": 7.887708649468893e-06, + "loss": 0.2824, + "step": 870 + }, + { + "epoch": 0.25164620359777684, + "grad_norm": 1.8425731658935547, + "learning_rate": 7.878603945371776e-06, + "loss": 0.2065, + "step": 873 + }, + { + "epoch": 0.2525109671840235, + "grad_norm": 14.634519577026367, + "learning_rate": 7.869499241274659e-06, + "loss": 0.2422, + "step": 876 + }, + { + "epoch": 0.25337573077027015, + "grad_norm": 3.1440680027008057, + "learning_rate": 7.860394537177543e-06, + "loss": 0.2418, + "step": 879 + }, + { + "epoch": 0.2542404943565168, + "grad_norm": 6.234373092651367, + "learning_rate": 7.851289833080426e-06, + "loss": 0.233, + "step": 882 + }, + { + "epoch": 0.25510525794276345, + "grad_norm": 2.140277147293091, + "learning_rate": 7.842185128983309e-06, + "loss": 0.2527, + "step": 885 + }, + { + "epoch": 0.2559700215290101, + "grad_norm": 3.195327043533325, + "learning_rate": 7.833080424886192e-06, + "loss": 0.2549, + "step": 888 + }, + { + "epoch": 0.25683478511525676, + "grad_norm": 2.8949568271636963, + "learning_rate": 7.823975720789075e-06, + "loss": 0.2318, + "step": 891 + }, + { + "epoch": 0.2576995487015034, + "grad_norm": 3.3351497650146484, + "learning_rate": 7.814871016691958e-06, + "loss": 0.2349, + "step": 894 + }, + { + "epoch": 0.25856431228775006, + "grad_norm": 7.218577861785889, + "learning_rate": 7.805766312594842e-06, + "loss": 0.2856, + "step": 897 + }, + { + "epoch": 0.2594290758739967, + "grad_norm": 3.1537606716156006, + "learning_rate": 7.796661608497725e-06, + "loss": 0.3181, + "step": 900 + }, + { + "epoch": 0.2594290758739967, + "eval_loss": 0.23068155348300934, + "eval_mse": 0.2306815572567284, + "eval_runtime": 64.5975, + "eval_samples_per_second": 15.48, + "eval_steps_per_second": 3.87, + "step": 900 + }, + { + "epoch": 0.26029383946024337, + "grad_norm": 4.1888933181762695, + "learning_rate": 7.787556904400608e-06, + "loss": 0.2682, + "step": 903 + }, + { + "epoch": 0.26115860304649, + "grad_norm": 2.9940059185028076, + "learning_rate": 7.77845220030349e-06, + "loss": 0.2381, + "step": 906 + }, + { + "epoch": 0.26202336663273673, + "grad_norm": 2.2147626876831055, + "learning_rate": 7.769347496206373e-06, + "loss": 0.2803, + "step": 909 + }, + { + "epoch": 0.2628881302189834, + "grad_norm": 2.8790485858917236, + "learning_rate": 7.760242792109256e-06, + "loss": 0.2372, + "step": 912 + }, + { + "epoch": 0.26375289380523004, + "grad_norm": 4.44177770614624, + "learning_rate": 7.75113808801214e-06, + "loss": 0.2345, + "step": 915 + }, + { + "epoch": 0.2646176573914767, + "grad_norm": 1.7946621179580688, + "learning_rate": 7.742033383915024e-06, + "loss": 0.2215, + "step": 918 + }, + { + "epoch": 0.26548242097772334, + "grad_norm": 2.3959579467773438, + "learning_rate": 7.732928679817907e-06, + "loss": 0.2286, + "step": 921 + }, + { + "epoch": 0.26634718456397, + "grad_norm": 6.54136848449707, + "learning_rate": 7.72382397572079e-06, + "loss": 0.2423, + "step": 924 + }, + { + "epoch": 0.26721194815021665, + "grad_norm": 7.17802095413208, + "learning_rate": 7.714719271623672e-06, + "loss": 0.2535, + "step": 927 + }, + { + "epoch": 0.2680767117364633, + "grad_norm": 2.7691524028778076, + "learning_rate": 7.705614567526557e-06, + "loss": 0.2141, + "step": 930 + }, + { + "epoch": 0.26894147532270996, + "grad_norm": 4.189211368560791, + "learning_rate": 7.69650986342944e-06, + "loss": 0.2577, + "step": 933 + }, + { + "epoch": 0.2698062389089566, + "grad_norm": 3.442680835723877, + "learning_rate": 7.687405159332322e-06, + "loss": 0.2438, + "step": 936 + }, + { + "epoch": 0.27067100249520326, + "grad_norm": 2.050264596939087, + "learning_rate": 7.678300455235205e-06, + "loss": 0.262, + "step": 939 + }, + { + "epoch": 0.2715357660814499, + "grad_norm": 6.750906944274902, + "learning_rate": 7.669195751138088e-06, + "loss": 0.2789, + "step": 942 + }, + { + "epoch": 0.27240052966769657, + "grad_norm": 14.716294288635254, + "learning_rate": 7.660091047040971e-06, + "loss": 0.294, + "step": 945 + }, + { + "epoch": 0.2732652932539432, + "grad_norm": 4.65310001373291, + "learning_rate": 7.650986342943855e-06, + "loss": 0.2938, + "step": 948 + }, + { + "epoch": 0.2741300568401899, + "grad_norm": 4.3455424308776855, + "learning_rate": 7.641881638846738e-06, + "loss": 0.2654, + "step": 951 + }, + { + "epoch": 0.2749948204264365, + "grad_norm": 18.063007354736328, + "learning_rate": 7.632776934749621e-06, + "loss": 0.2812, + "step": 954 + }, + { + "epoch": 0.2758595840126832, + "grad_norm": 2.212507963180542, + "learning_rate": 7.623672230652505e-06, + "loss": 0.2134, + "step": 957 + }, + { + "epoch": 0.27672434759892983, + "grad_norm": 2.462282180786133, + "learning_rate": 7.614567526555388e-06, + "loss": 0.235, + "step": 960 + }, + { + "epoch": 0.2775891111851765, + "grad_norm": 9.775514602661133, + "learning_rate": 7.6054628224582705e-06, + "loss": 0.258, + "step": 963 + }, + { + "epoch": 0.2784538747714232, + "grad_norm": 2.7479944229125977, + "learning_rate": 7.596358118361153e-06, + "loss": 0.2653, + "step": 966 + }, + { + "epoch": 0.27931863835766985, + "grad_norm": 3.1024341583251953, + "learning_rate": 7.587253414264037e-06, + "loss": 0.2611, + "step": 969 + }, + { + "epoch": 0.2801834019439165, + "grad_norm": 4.7744317054748535, + "learning_rate": 7.578148710166921e-06, + "loss": 0.2676, + "step": 972 + }, + { + "epoch": 0.28104816553016315, + "grad_norm": 2.3438374996185303, + "learning_rate": 7.569044006069804e-06, + "loss": 0.2521, + "step": 975 + }, + { + "epoch": 0.2819129291164098, + "grad_norm": 5.55867338180542, + "learning_rate": 7.5599393019726864e-06, + "loss": 0.2592, + "step": 978 + }, + { + "epoch": 0.28277769270265646, + "grad_norm": 3.6212918758392334, + "learning_rate": 7.550834597875569e-06, + "loss": 0.2717, + "step": 981 + }, + { + "epoch": 0.2836424562889031, + "grad_norm": 2.0010621547698975, + "learning_rate": 7.541729893778453e-06, + "loss": 0.2303, + "step": 984 + }, + { + "epoch": 0.28450721987514976, + "grad_norm": 2.170227289199829, + "learning_rate": 7.532625189681337e-06, + "loss": 0.2436, + "step": 987 + }, + { + "epoch": 0.2853719834613964, + "grad_norm": 3.4833590984344482, + "learning_rate": 7.5235204855842195e-06, + "loss": 0.2327, + "step": 990 + }, + { + "epoch": 0.28623674704764307, + "grad_norm": 2.1279664039611816, + "learning_rate": 7.514415781487102e-06, + "loss": 0.2706, + "step": 993 + }, + { + "epoch": 0.2871015106338897, + "grad_norm": 3.14772891998291, + "learning_rate": 7.505311077389985e-06, + "loss": 0.2325, + "step": 996 + }, + { + "epoch": 0.2879662742201364, + "grad_norm": 3.4754092693328857, + "learning_rate": 7.496206373292868e-06, + "loss": 0.2551, + "step": 999 + }, + { + "epoch": 0.28825452874888524, + "eval_loss": 0.24107317626476288, + "eval_mse": 0.24107318028528243, + "eval_runtime": 64.3256, + "eval_samples_per_second": 15.546, + "eval_steps_per_second": 3.886, + "step": 1000 + }, + { + "epoch": 0.28883103780638303, + "grad_norm": 2.1394267082214355, + "learning_rate": 7.487101669195752e-06, + "loss": 0.2094, + "step": 1002 + }, + { + "epoch": 0.2896958013926297, + "grad_norm": 1.6514432430267334, + "learning_rate": 7.477996965098635e-06, + "loss": 0.2475, + "step": 1005 + }, + { + "epoch": 0.29056056497887633, + "grad_norm": 5.082777500152588, + "learning_rate": 7.468892261001518e-06, + "loss": 0.2829, + "step": 1008 + }, + { + "epoch": 0.291425328565123, + "grad_norm": 3.7276644706726074, + "learning_rate": 7.459787556904401e-06, + "loss": 0.2955, + "step": 1011 + }, + { + "epoch": 0.29229009215136964, + "grad_norm": 3.5418541431427, + "learning_rate": 7.450682852807284e-06, + "loss": 0.2359, + "step": 1014 + }, + { + "epoch": 0.2931548557376163, + "grad_norm": 1.7081390619277954, + "learning_rate": 7.441578148710168e-06, + "loss": 0.2377, + "step": 1017 + }, + { + "epoch": 0.29401961932386295, + "grad_norm": 3.4884068965911865, + "learning_rate": 7.4324734446130505e-06, + "loss": 0.1981, + "step": 1020 + }, + { + "epoch": 0.2948843829101096, + "grad_norm": 2.5671768188476562, + "learning_rate": 7.423368740515934e-06, + "loss": 0.2657, + "step": 1023 + }, + { + "epoch": 0.2957491464963563, + "grad_norm": 3.0763845443725586, + "learning_rate": 7.414264036418817e-06, + "loss": 0.2391, + "step": 1026 + }, + { + "epoch": 0.29661391008260296, + "grad_norm": 2.09977388381958, + "learning_rate": 7.4051593323217e-06, + "loss": 0.2426, + "step": 1029 + }, + { + "epoch": 0.2974786736688496, + "grad_norm": 1.816253900527954, + "learning_rate": 7.3960546282245835e-06, + "loss": 0.2309, + "step": 1032 + }, + { + "epoch": 0.29834343725509627, + "grad_norm": 2.3119256496429443, + "learning_rate": 7.386949924127466e-06, + "loss": 0.2379, + "step": 1035 + }, + { + "epoch": 0.2992082008413429, + "grad_norm": 4.264431953430176, + "learning_rate": 7.377845220030349e-06, + "loss": 0.2603, + "step": 1038 + }, + { + "epoch": 0.3000729644275896, + "grad_norm": 2.759507656097412, + "learning_rate": 7.368740515933233e-06, + "loss": 0.2634, + "step": 1041 + }, + { + "epoch": 0.3009377280138362, + "grad_norm": 2.629415988922119, + "learning_rate": 7.359635811836116e-06, + "loss": 0.2091, + "step": 1044 + }, + { + "epoch": 0.3018024916000829, + "grad_norm": 1.7376865148544312, + "learning_rate": 7.3505311077389994e-06, + "loss": 0.227, + "step": 1047 + }, + { + "epoch": 0.30266725518632953, + "grad_norm": 2.700201988220215, + "learning_rate": 7.341426403641882e-06, + "loss": 0.2336, + "step": 1050 + }, + { + "epoch": 0.3035320187725762, + "grad_norm": 17.49102020263672, + "learning_rate": 7.332321699544765e-06, + "loss": 0.2714, + "step": 1053 + }, + { + "epoch": 0.30439678235882284, + "grad_norm": 3.3329529762268066, + "learning_rate": 7.323216995447649e-06, + "loss": 0.2199, + "step": 1056 + }, + { + "epoch": 0.3052615459450695, + "grad_norm": 2.0356404781341553, + "learning_rate": 7.314112291350532e-06, + "loss": 0.2329, + "step": 1059 + }, + { + "epoch": 0.30612630953131614, + "grad_norm": 1.992146611213684, + "learning_rate": 7.305007587253415e-06, + "loss": 0.2305, + "step": 1062 + }, + { + "epoch": 0.3069910731175628, + "grad_norm": 3.694671630859375, + "learning_rate": 7.295902883156298e-06, + "loss": 0.2149, + "step": 1065 + }, + { + "epoch": 0.30785583670380945, + "grad_norm": 2.258823871612549, + "learning_rate": 7.286798179059181e-06, + "loss": 0.2347, + "step": 1068 + }, + { + "epoch": 0.3087206002900561, + "grad_norm": 2.762303590774536, + "learning_rate": 7.277693474962064e-06, + "loss": 0.2355, + "step": 1071 + }, + { + "epoch": 0.30958536387630275, + "grad_norm": 2.68644118309021, + "learning_rate": 7.2685887708649476e-06, + "loss": 0.2308, + "step": 1074 + }, + { + "epoch": 0.3104501274625494, + "grad_norm": 5.270909786224365, + "learning_rate": 7.25948406676783e-06, + "loss": 0.2781, + "step": 1077 + }, + { + "epoch": 0.31131489104879606, + "grad_norm": 1.9733515977859497, + "learning_rate": 7.250379362670714e-06, + "loss": 0.2465, + "step": 1080 + }, + { + "epoch": 0.31217965463504277, + "grad_norm": 4.668141841888428, + "learning_rate": 7.241274658573597e-06, + "loss": 0.2743, + "step": 1083 + }, + { + "epoch": 0.3130444182212894, + "grad_norm": 5.158870220184326, + "learning_rate": 7.23216995447648e-06, + "loss": 0.2702, + "step": 1086 + }, + { + "epoch": 0.3139091818075361, + "grad_norm": 20.275535583496094, + "learning_rate": 7.223065250379363e-06, + "loss": 0.2609, + "step": 1089 + }, + { + "epoch": 0.31477394539378273, + "grad_norm": 3.8496415615081787, + "learning_rate": 7.213960546282246e-06, + "loss": 0.2641, + "step": 1092 + }, + { + "epoch": 0.3156387089800294, + "grad_norm": 4.2407073974609375, + "learning_rate": 7.20485584218513e-06, + "loss": 0.2255, + "step": 1095 + }, + { + "epoch": 0.31650347256627603, + "grad_norm": 6.420517444610596, + "learning_rate": 7.195751138088013e-06, + "loss": 0.2743, + "step": 1098 + }, + { + "epoch": 0.31707998162377377, + "eval_loss": 0.2349877655506134, + "eval_mse": 0.2349877648057918, + "eval_runtime": 64.4388, + "eval_samples_per_second": 15.519, + "eval_steps_per_second": 3.88, + "step": 1100 + }, + { + "epoch": 0.3173682361525227, + "grad_norm": 3.0954813957214355, + "learning_rate": 7.186646433990896e-06, + "loss": 0.2681, + "step": 1101 + }, + { + "epoch": 0.31823299973876934, + "grad_norm": 3.8299410343170166, + "learning_rate": 7.1775417298937785e-06, + "loss": 0.2378, + "step": 1104 + }, + { + "epoch": 0.319097763325016, + "grad_norm": 2.5736284255981445, + "learning_rate": 7.168437025796661e-06, + "loss": 0.2441, + "step": 1107 + }, + { + "epoch": 0.31996252691126265, + "grad_norm": 2.7379114627838135, + "learning_rate": 7.159332321699546e-06, + "loss": 0.2216, + "step": 1110 + }, + { + "epoch": 0.3208272904975093, + "grad_norm": 11.14561653137207, + "learning_rate": 7.150227617602429e-06, + "loss": 0.2574, + "step": 1113 + }, + { + "epoch": 0.32169205408375595, + "grad_norm": 3.286914348602295, + "learning_rate": 7.141122913505312e-06, + "loss": 0.2575, + "step": 1116 + }, + { + "epoch": 0.3225568176700026, + "grad_norm": 5.4453020095825195, + "learning_rate": 7.1320182094081944e-06, + "loss": 0.2307, + "step": 1119 + }, + { + "epoch": 0.32342158125624926, + "grad_norm": 3.9706311225891113, + "learning_rate": 7.122913505311077e-06, + "loss": 0.2622, + "step": 1122 + }, + { + "epoch": 0.3242863448424959, + "grad_norm": 5.497633934020996, + "learning_rate": 7.113808801213962e-06, + "loss": 0.2433, + "step": 1125 + }, + { + "epoch": 0.32515110842874256, + "grad_norm": 5.734874725341797, + "learning_rate": 7.104704097116845e-06, + "loss": 0.265, + "step": 1128 + }, + { + "epoch": 0.3260158720149892, + "grad_norm": 1.9510211944580078, + "learning_rate": 7.0955993930197275e-06, + "loss": 0.26, + "step": 1131 + }, + { + "epoch": 0.32688063560123587, + "grad_norm": 11.36854076385498, + "learning_rate": 7.08649468892261e-06, + "loss": 0.2378, + "step": 1134 + }, + { + "epoch": 0.3277453991874825, + "grad_norm": 1.991607666015625, + "learning_rate": 7.077389984825493e-06, + "loss": 0.2399, + "step": 1137 + }, + { + "epoch": 0.3286101627737292, + "grad_norm": 4.413910388946533, + "learning_rate": 7.068285280728376e-06, + "loss": 0.2411, + "step": 1140 + }, + { + "epoch": 0.3294749263599759, + "grad_norm": 4.106217861175537, + "learning_rate": 7.0591805766312606e-06, + "loss": 0.2713, + "step": 1143 + }, + { + "epoch": 0.33033968994622254, + "grad_norm": 2.9739654064178467, + "learning_rate": 7.050075872534143e-06, + "loss": 0.2463, + "step": 1146 + }, + { + "epoch": 0.3312044535324692, + "grad_norm": 6.642212390899658, + "learning_rate": 7.040971168437026e-06, + "loss": 0.2773, + "step": 1149 + }, + { + "epoch": 0.33206921711871584, + "grad_norm": 4.020822048187256, + "learning_rate": 7.031866464339909e-06, + "loss": 0.2493, + "step": 1152 + }, + { + "epoch": 0.3329339807049625, + "grad_norm": 4.196906566619873, + "learning_rate": 7.022761760242792e-06, + "loss": 0.227, + "step": 1155 + }, + { + "epoch": 0.33379874429120915, + "grad_norm": 3.0572118759155273, + "learning_rate": 7.0136570561456765e-06, + "loss": 0.2594, + "step": 1158 + }, + { + "epoch": 0.3346635078774558, + "grad_norm": 2.7601091861724854, + "learning_rate": 7.004552352048559e-06, + "loss": 0.2299, + "step": 1161 + }, + { + "epoch": 0.33552827146370245, + "grad_norm": 2.025371551513672, + "learning_rate": 6.995447647951442e-06, + "loss": 0.2484, + "step": 1164 + }, + { + "epoch": 0.3363930350499491, + "grad_norm": 4.210418701171875, + "learning_rate": 6.986342943854325e-06, + "loss": 0.2247, + "step": 1167 + }, + { + "epoch": 0.33725779863619576, + "grad_norm": 5.4730544090271, + "learning_rate": 6.977238239757208e-06, + "loss": 0.2616, + "step": 1170 + }, + { + "epoch": 0.3381225622224424, + "grad_norm": 3.7742884159088135, + "learning_rate": 6.968133535660092e-06, + "loss": 0.2428, + "step": 1173 + }, + { + "epoch": 0.33898732580868907, + "grad_norm": 3.491050958633423, + "learning_rate": 6.959028831562975e-06, + "loss": 0.2296, + "step": 1176 + }, + { + "epoch": 0.3398520893949357, + "grad_norm": 2.1583101749420166, + "learning_rate": 6.949924127465858e-06, + "loss": 0.2649, + "step": 1179 + }, + { + "epoch": 0.34071685298118237, + "grad_norm": 2.2460827827453613, + "learning_rate": 6.940819423368741e-06, + "loss": 0.2308, + "step": 1182 + }, + { + "epoch": 0.341581616567429, + "grad_norm": 4.751939296722412, + "learning_rate": 6.931714719271624e-06, + "loss": 0.2491, + "step": 1185 + }, + { + "epoch": 0.3424463801536757, + "grad_norm": 2.2470085620880127, + "learning_rate": 6.922610015174508e-06, + "loss": 0.2135, + "step": 1188 + }, + { + "epoch": 0.34331114373992233, + "grad_norm": 3.2550578117370605, + "learning_rate": 6.913505311077391e-06, + "loss": 0.2345, + "step": 1191 + }, + { + "epoch": 0.344175907326169, + "grad_norm": 2.159435749053955, + "learning_rate": 6.904400606980274e-06, + "loss": 0.2413, + "step": 1194 + }, + { + "epoch": 0.34504067091241564, + "grad_norm": 23.409839630126953, + "learning_rate": 6.895295902883157e-06, + "loss": 0.2101, + "step": 1197 + }, + { + "epoch": 0.34590543449866235, + "grad_norm": 3.5459580421447754, + "learning_rate": 6.88619119878604e-06, + "loss": 0.2383, + "step": 1200 + }, + { + "epoch": 0.34590543449866235, + "eval_loss": 0.2423837035894394, + "eval_mse": 0.2423837145415746, + "eval_runtime": 64.1907, + "eval_samples_per_second": 15.579, + "eval_steps_per_second": 3.895, + "step": 1200 + }, + { + "epoch": 0.346770198084909, + "grad_norm": 4.81601095199585, + "learning_rate": 6.8770864946889225e-06, + "loss": 0.2548, + "step": 1203 + }, + { + "epoch": 0.34763496167115565, + "grad_norm": 1.821042776107788, + "learning_rate": 6.867981790591807e-06, + "loss": 0.2054, + "step": 1206 + }, + { + "epoch": 0.3484997252574023, + "grad_norm": 2.1432528495788574, + "learning_rate": 6.85887708649469e-06, + "loss": 0.2324, + "step": 1209 + }, + { + "epoch": 0.34936448884364896, + "grad_norm": 2.9332826137542725, + "learning_rate": 6.849772382397573e-06, + "loss": 0.2336, + "step": 1212 + }, + { + "epoch": 0.3502292524298956, + "grad_norm": 6.8339619636535645, + "learning_rate": 6.8406676783004556e-06, + "loss": 0.2132, + "step": 1215 + }, + { + "epoch": 0.35109401601614226, + "grad_norm": 2.1698455810546875, + "learning_rate": 6.831562974203338e-06, + "loss": 0.235, + "step": 1218 + }, + { + "epoch": 0.3519587796023889, + "grad_norm": 4.076608657836914, + "learning_rate": 6.822458270106223e-06, + "loss": 0.2444, + "step": 1221 + }, + { + "epoch": 0.35282354318863557, + "grad_norm": 2.6955060958862305, + "learning_rate": 6.813353566009106e-06, + "loss": 0.228, + "step": 1224 + }, + { + "epoch": 0.3536883067748822, + "grad_norm": 2.3251149654388428, + "learning_rate": 6.804248861911989e-06, + "loss": 0.2501, + "step": 1227 + }, + { + "epoch": 0.3545530703611289, + "grad_norm": 3.7349984645843506, + "learning_rate": 6.7951441578148715e-06, + "loss": 0.2332, + "step": 1230 + }, + { + "epoch": 0.3554178339473755, + "grad_norm": 2.2955520153045654, + "learning_rate": 6.786039453717754e-06, + "loss": 0.2196, + "step": 1233 + }, + { + "epoch": 0.3562825975336222, + "grad_norm": 2.053947687149048, + "learning_rate": 6.776934749620638e-06, + "loss": 0.2281, + "step": 1236 + }, + { + "epoch": 0.35714736111986883, + "grad_norm": 1.7476141452789307, + "learning_rate": 6.767830045523522e-06, + "loss": 0.2403, + "step": 1239 + }, + { + "epoch": 0.3580121247061155, + "grad_norm": 3.03908109664917, + "learning_rate": 6.7587253414264045e-06, + "loss": 0.266, + "step": 1242 + }, + { + "epoch": 0.35887688829236214, + "grad_norm": 3.2814714908599854, + "learning_rate": 6.749620637329287e-06, + "loss": 0.2137, + "step": 1245 + }, + { + "epoch": 0.3597416518786088, + "grad_norm": 2.523711681365967, + "learning_rate": 6.74051593323217e-06, + "loss": 0.2396, + "step": 1248 + }, + { + "epoch": 0.36060641546485545, + "grad_norm": 2.1850292682647705, + "learning_rate": 6.731411229135054e-06, + "loss": 0.2297, + "step": 1251 + }, + { + "epoch": 0.3614711790511021, + "grad_norm": 2.3782458305358887, + "learning_rate": 6.722306525037937e-06, + "loss": 0.2159, + "step": 1254 + }, + { + "epoch": 0.36233594263734875, + "grad_norm": 4.745922088623047, + "learning_rate": 6.7132018209408204e-06, + "loss": 0.2111, + "step": 1257 + }, + { + "epoch": 0.36320070622359546, + "grad_norm": 2.0191493034362793, + "learning_rate": 6.704097116843703e-06, + "loss": 0.2648, + "step": 1260 + }, + { + "epoch": 0.3640654698098421, + "grad_norm": 61.380306243896484, + "learning_rate": 6.694992412746586e-06, + "loss": 0.2484, + "step": 1263 + }, + { + "epoch": 0.36493023339608877, + "grad_norm": 3.0049686431884766, + "learning_rate": 6.685887708649469e-06, + "loss": 0.2456, + "step": 1266 + }, + { + "epoch": 0.3657949969823354, + "grad_norm": 3.6466712951660156, + "learning_rate": 6.676783004552353e-06, + "loss": 0.3093, + "step": 1269 + }, + { + "epoch": 0.36665976056858207, + "grad_norm": 2.6709084510803223, + "learning_rate": 6.6676783004552355e-06, + "loss": 0.2129, + "step": 1272 + }, + { + "epoch": 0.3675245241548287, + "grad_norm": 3.721860408782959, + "learning_rate": 6.658573596358119e-06, + "loss": 0.2476, + "step": 1275 + }, + { + "epoch": 0.3683892877410754, + "grad_norm": 3.2325708866119385, + "learning_rate": 6.649468892261002e-06, + "loss": 0.2281, + "step": 1278 + }, + { + "epoch": 0.36925405132732203, + "grad_norm": 3.1280910968780518, + "learning_rate": 6.640364188163885e-06, + "loss": 0.2791, + "step": 1281 + }, + { + "epoch": 0.3701188149135687, + "grad_norm": 2.1942126750946045, + "learning_rate": 6.6312594840667686e-06, + "loss": 0.2323, + "step": 1284 + }, + { + "epoch": 0.37098357849981534, + "grad_norm": 4.829765796661377, + "learning_rate": 6.622154779969651e-06, + "loss": 0.204, + "step": 1287 + }, + { + "epoch": 0.371848342086062, + "grad_norm": 2.7358474731445312, + "learning_rate": 6.613050075872534e-06, + "loss": 0.2484, + "step": 1290 + }, + { + "epoch": 0.37271310567230864, + "grad_norm": 3.0888988971710205, + "learning_rate": 6.603945371775418e-06, + "loss": 0.251, + "step": 1293 + }, + { + "epoch": 0.3735778692585553, + "grad_norm": 2.374920129776001, + "learning_rate": 6.594840667678301e-06, + "loss": 0.2513, + "step": 1296 + }, + { + "epoch": 0.37444263284480195, + "grad_norm": 4.422918319702148, + "learning_rate": 6.5857359635811845e-06, + "loss": 0.2191, + "step": 1299 + }, + { + "epoch": 0.37473088737355087, + "eval_loss": 0.22791720926761627, + "eval_mse": 0.2279172136860434, + "eval_runtime": 64.4369, + "eval_samples_per_second": 15.519, + "eval_steps_per_second": 3.88, + "step": 1300 + }, + { + "epoch": 0.3753073964310486, + "grad_norm": 3.1236884593963623, + "learning_rate": 6.576631259484067e-06, + "loss": 0.2676, + "step": 1302 + }, + { + "epoch": 0.37617216001729525, + "grad_norm": 2.3095321655273438, + "learning_rate": 6.56752655538695e-06, + "loss": 0.2373, + "step": 1305 + }, + { + "epoch": 0.3770369236035419, + "grad_norm": 2.6782171726226807, + "learning_rate": 6.558421851289834e-06, + "loss": 0.2147, + "step": 1308 + }, + { + "epoch": 0.37790168718978856, + "grad_norm": 2.2795357704162598, + "learning_rate": 6.549317147192717e-06, + "loss": 0.2458, + "step": 1311 + }, + { + "epoch": 0.3787664507760352, + "grad_norm": 2.5536789894104004, + "learning_rate": 6.5402124430956e-06, + "loss": 0.2643, + "step": 1314 + }, + { + "epoch": 0.3796312143622819, + "grad_norm": 2.8398616313934326, + "learning_rate": 6.531107738998483e-06, + "loss": 0.2647, + "step": 1317 + }, + { + "epoch": 0.3804959779485286, + "grad_norm": 4.880849838256836, + "learning_rate": 6.522003034901366e-06, + "loss": 0.2151, + "step": 1320 + }, + { + "epoch": 0.3813607415347752, + "grad_norm": 3.158966302871704, + "learning_rate": 6.512898330804249e-06, + "loss": 0.2884, + "step": 1323 + }, + { + "epoch": 0.3822255051210219, + "grad_norm": 1.8538583517074585, + "learning_rate": 6.503793626707133e-06, + "loss": 0.2238, + "step": 1326 + }, + { + "epoch": 0.38309026870726853, + "grad_norm": 3.2521581649780273, + "learning_rate": 6.4946889226100154e-06, + "loss": 0.2328, + "step": 1329 + }, + { + "epoch": 0.3839550322935152, + "grad_norm": 2.3196299076080322, + "learning_rate": 6.485584218512899e-06, + "loss": 0.2885, + "step": 1332 + }, + { + "epoch": 0.38481979587976184, + "grad_norm": 3.506424903869629, + "learning_rate": 6.476479514415782e-06, + "loss": 0.2208, + "step": 1335 + }, + { + "epoch": 0.3856845594660085, + "grad_norm": 2.378120183944702, + "learning_rate": 6.467374810318665e-06, + "loss": 0.2187, + "step": 1338 + }, + { + "epoch": 0.38654932305225514, + "grad_norm": 2.1336095333099365, + "learning_rate": 6.458270106221548e-06, + "loss": 0.2626, + "step": 1341 + }, + { + "epoch": 0.3874140866385018, + "grad_norm": 1.8846355676651, + "learning_rate": 6.449165402124431e-06, + "loss": 0.2468, + "step": 1344 + }, + { + "epoch": 0.38827885022474845, + "grad_norm": 3.833599805831909, + "learning_rate": 6.440060698027315e-06, + "loss": 0.2432, + "step": 1347 + }, + { + "epoch": 0.3891436138109951, + "grad_norm": 2.8712267875671387, + "learning_rate": 6.430955993930198e-06, + "loss": 0.2383, + "step": 1350 + }, + { + "epoch": 0.39000837739724176, + "grad_norm": 2.344010829925537, + "learning_rate": 6.421851289833081e-06, + "loss": 0.233, + "step": 1353 + }, + { + "epoch": 0.3908731409834884, + "grad_norm": 4.56198263168335, + "learning_rate": 6.4127465857359636e-06, + "loss": 0.2583, + "step": 1356 + }, + { + "epoch": 0.39173790456973506, + "grad_norm": 3.3095390796661377, + "learning_rate": 6.403641881638846e-06, + "loss": 0.2228, + "step": 1359 + }, + { + "epoch": 0.3926026681559817, + "grad_norm": 5.623388290405273, + "learning_rate": 6.394537177541731e-06, + "loss": 0.2425, + "step": 1362 + }, + { + "epoch": 0.39346743174222837, + "grad_norm": 4.270349502563477, + "learning_rate": 6.385432473444614e-06, + "loss": 0.2283, + "step": 1365 + }, + { + "epoch": 0.394332195328475, + "grad_norm": 4.91237211227417, + "learning_rate": 6.376327769347497e-06, + "loss": 0.2472, + "step": 1368 + }, + { + "epoch": 0.3951969589147217, + "grad_norm": 3.120549440383911, + "learning_rate": 6.3672230652503795e-06, + "loss": 0.2724, + "step": 1371 + }, + { + "epoch": 0.3960617225009683, + "grad_norm": 2.615607500076294, + "learning_rate": 6.358118361153262e-06, + "loss": 0.2779, + "step": 1374 + }, + { + "epoch": 0.39692648608721504, + "grad_norm": 3.6636528968811035, + "learning_rate": 6.349013657056147e-06, + "loss": 0.2561, + "step": 1377 + }, + { + "epoch": 0.3977912496734617, + "grad_norm": 3.095202684402466, + "learning_rate": 6.33990895295903e-06, + "loss": 0.2556, + "step": 1380 + }, + { + "epoch": 0.39865601325970834, + "grad_norm": 1.8077003955841064, + "learning_rate": 6.3308042488619125e-06, + "loss": 0.2337, + "step": 1383 + }, + { + "epoch": 0.399520776845955, + "grad_norm": 1.9146672487258911, + "learning_rate": 6.321699544764795e-06, + "loss": 0.206, + "step": 1386 + }, + { + "epoch": 0.40038554043220165, + "grad_norm": 1.8629751205444336, + "learning_rate": 6.312594840667678e-06, + "loss": 0.2303, + "step": 1389 + }, + { + "epoch": 0.4012503040184483, + "grad_norm": 3.0480451583862305, + "learning_rate": 6.303490136570563e-06, + "loss": 0.2745, + "step": 1392 + }, + { + "epoch": 0.40211506760469495, + "grad_norm": 2.0094082355499268, + "learning_rate": 6.294385432473446e-06, + "loss": 0.2122, + "step": 1395 + }, + { + "epoch": 0.4029798311909416, + "grad_norm": 3.787971019744873, + "learning_rate": 6.2852807283763284e-06, + "loss": 0.2431, + "step": 1398 + }, + { + "epoch": 0.4035563402484394, + "eval_loss": 0.22321856021881104, + "eval_mse": 0.2232185562737286, + "eval_runtime": 64.7184, + "eval_samples_per_second": 15.452, + "eval_steps_per_second": 3.863, + "step": 1400 + }, + { + "epoch": 0.40384459477718826, + "grad_norm": 2.886836290359497, + "learning_rate": 6.276176024279211e-06, + "loss": 0.1921, + "step": 1401 + }, + { + "epoch": 0.4047093583634349, + "grad_norm": 3.5917375087738037, + "learning_rate": 6.267071320182094e-06, + "loss": 0.2708, + "step": 1404 + }, + { + "epoch": 0.40557412194968157, + "grad_norm": 1.947647213935852, + "learning_rate": 6.257966616084977e-06, + "loss": 0.2386, + "step": 1407 + }, + { + "epoch": 0.4064388855359282, + "grad_norm": 21.491071701049805, + "learning_rate": 6.2488619119878615e-06, + "loss": 0.2368, + "step": 1410 + }, + { + "epoch": 0.40730364912217487, + "grad_norm": 2.2164909839630127, + "learning_rate": 6.239757207890744e-06, + "loss": 0.2206, + "step": 1413 + }, + { + "epoch": 0.4081684127084215, + "grad_norm": 4.416579246520996, + "learning_rate": 6.230652503793627e-06, + "loss": 0.2655, + "step": 1416 + }, + { + "epoch": 0.4090331762946682, + "grad_norm": 2.469964027404785, + "learning_rate": 6.22154779969651e-06, + "loss": 0.2633, + "step": 1419 + }, + { + "epoch": 0.40989793988091483, + "grad_norm": 2.6230666637420654, + "learning_rate": 6.212443095599393e-06, + "loss": 0.2043, + "step": 1422 + }, + { + "epoch": 0.4107627034671615, + "grad_norm": 2.4938716888427734, + "learning_rate": 6.203338391502277e-06, + "loss": 0.2508, + "step": 1425 + }, + { + "epoch": 0.41162746705340814, + "grad_norm": 2.508373975753784, + "learning_rate": 6.19423368740516e-06, + "loss": 0.2663, + "step": 1428 + }, + { + "epoch": 0.4124922306396548, + "grad_norm": 2.955127716064453, + "learning_rate": 6.185128983308043e-06, + "loss": 0.2933, + "step": 1431 + }, + { + "epoch": 0.4133569942259015, + "grad_norm": 2.3575446605682373, + "learning_rate": 6.176024279210926e-06, + "loss": 0.2369, + "step": 1434 + }, + { + "epoch": 0.41422175781214815, + "grad_norm": 2.6992990970611572, + "learning_rate": 6.166919575113809e-06, + "loss": 0.2584, + "step": 1437 + }, + { + "epoch": 0.4150865213983948, + "grad_norm": 3.9453766345977783, + "learning_rate": 6.157814871016693e-06, + "loss": 0.27, + "step": 1440 + }, + { + "epoch": 0.41595128498464146, + "grad_norm": 2.111201286315918, + "learning_rate": 6.148710166919576e-06, + "loss": 0.2312, + "step": 1443 + }, + { + "epoch": 0.4168160485708881, + "grad_norm": 2.37361741065979, + "learning_rate": 6.139605462822459e-06, + "loss": 0.2638, + "step": 1446 + }, + { + "epoch": 0.41768081215713476, + "grad_norm": 5.3990631103515625, + "learning_rate": 6.130500758725342e-06, + "loss": 0.242, + "step": 1449 + }, + { + "epoch": 0.4185455757433814, + "grad_norm": 3.4745547771453857, + "learning_rate": 6.121396054628225e-06, + "loss": 0.2404, + "step": 1452 + }, + { + "epoch": 0.41941033932962807, + "grad_norm": 2.619647979736328, + "learning_rate": 6.112291350531108e-06, + "loss": 0.211, + "step": 1455 + }, + { + "epoch": 0.4202751029158747, + "grad_norm": 5.034475326538086, + "learning_rate": 6.103186646433992e-06, + "loss": 0.2398, + "step": 1458 + }, + { + "epoch": 0.4211398665021214, + "grad_norm": 3.0307469367980957, + "learning_rate": 6.094081942336875e-06, + "loss": 0.2708, + "step": 1461 + }, + { + "epoch": 0.422004630088368, + "grad_norm": 7.210414409637451, + "learning_rate": 6.084977238239758e-06, + "loss": 0.2348, + "step": 1464 + }, + { + "epoch": 0.4228693936746147, + "grad_norm": 2.5068700313568115, + "learning_rate": 6.075872534142641e-06, + "loss": 0.2168, + "step": 1467 + }, + { + "epoch": 0.42373415726086133, + "grad_norm": 1.664629578590393, + "learning_rate": 6.0667678300455234e-06, + "loss": 0.2166, + "step": 1470 + }, + { + "epoch": 0.424598920847108, + "grad_norm": 2.4851441383361816, + "learning_rate": 6.057663125948408e-06, + "loss": 0.2202, + "step": 1473 + }, + { + "epoch": 0.42546368443335464, + "grad_norm": 3.4534077644348145, + "learning_rate": 6.048558421851291e-06, + "loss": 0.2236, + "step": 1476 + }, + { + "epoch": 0.4263284480196013, + "grad_norm": 1.7571723461151123, + "learning_rate": 6.039453717754174e-06, + "loss": 0.2499, + "step": 1479 + }, + { + "epoch": 0.42719321160584794, + "grad_norm": 6.117722034454346, + "learning_rate": 6.0303490136570565e-06, + "loss": 0.2408, + "step": 1482 + }, + { + "epoch": 0.4280579751920946, + "grad_norm": 6.158169746398926, + "learning_rate": 6.021244309559939e-06, + "loss": 0.2673, + "step": 1485 + }, + { + "epoch": 0.42892273877834125, + "grad_norm": 2.713707208633423, + "learning_rate": 6.012139605462823e-06, + "loss": 0.2544, + "step": 1488 + }, + { + "epoch": 0.42978750236458796, + "grad_norm": 2.2053639888763428, + "learning_rate": 6.003034901365707e-06, + "loss": 0.2218, + "step": 1491 + }, + { + "epoch": 0.4306522659508346, + "grad_norm": 2.291422128677368, + "learning_rate": 5.9939301972685896e-06, + "loss": 0.1924, + "step": 1494 + }, + { + "epoch": 0.43151702953708126, + "grad_norm": 4.670581817626953, + "learning_rate": 5.984825493171472e-06, + "loss": 0.209, + "step": 1497 + }, + { + "epoch": 0.4323817931233279, + "grad_norm": 2.0881216526031494, + "learning_rate": 5.975720789074355e-06, + "loss": 0.2161, + "step": 1500 + }, + { + "epoch": 0.4323817931233279, + "eval_loss": 0.23065043985843658, + "eval_mse": 0.23065044428501277, + "eval_runtime": 64.6792, + "eval_samples_per_second": 15.461, + "eval_steps_per_second": 3.865, + "step": 1500 + }, + { + "epoch": 0.43324655670957457, + "grad_norm": 4.005910396575928, + "learning_rate": 5.966616084977239e-06, + "loss": 0.2585, + "step": 1503 + }, + { + "epoch": 0.4341113202958212, + "grad_norm": 3.212456703186035, + "learning_rate": 5.957511380880122e-06, + "loss": 0.2328, + "step": 1506 + }, + { + "epoch": 0.4349760838820679, + "grad_norm": 17.725706100463867, + "learning_rate": 5.9484066767830055e-06, + "loss": 0.2658, + "step": 1509 + }, + { + "epoch": 0.43584084746831453, + "grad_norm": 5.476046562194824, + "learning_rate": 5.939301972685888e-06, + "loss": 0.2512, + "step": 1512 + }, + { + "epoch": 0.4367056110545612, + "grad_norm": 6.102067470550537, + "learning_rate": 5.930197268588771e-06, + "loss": 0.2248, + "step": 1515 + }, + { + "epoch": 0.43757037464080784, + "grad_norm": 1.9437217712402344, + "learning_rate": 5.921092564491655e-06, + "loss": 0.2424, + "step": 1518 + }, + { + "epoch": 0.4384351382270545, + "grad_norm": 4.161287307739258, + "learning_rate": 5.911987860394538e-06, + "loss": 0.238, + "step": 1521 + }, + { + "epoch": 0.43929990181330114, + "grad_norm": 4.9400458335876465, + "learning_rate": 5.9028831562974205e-06, + "loss": 0.2587, + "step": 1524 + }, + { + "epoch": 0.4401646653995478, + "grad_norm": 3.344257354736328, + "learning_rate": 5.893778452200304e-06, + "loss": 0.259, + "step": 1527 + }, + { + "epoch": 0.44102942898579445, + "grad_norm": 3.5931825637817383, + "learning_rate": 5.884673748103187e-06, + "loss": 0.2444, + "step": 1530 + }, + { + "epoch": 0.4418941925720411, + "grad_norm": 2.649014949798584, + "learning_rate": 5.87556904400607e-06, + "loss": 0.2303, + "step": 1533 + }, + { + "epoch": 0.44275895615828775, + "grad_norm": 2.2219338417053223, + "learning_rate": 5.866464339908954e-06, + "loss": 0.2177, + "step": 1536 + }, + { + "epoch": 0.4436237197445344, + "grad_norm": 1.748891830444336, + "learning_rate": 5.8573596358118364e-06, + "loss": 0.2048, + "step": 1539 + }, + { + "epoch": 0.44448848333078106, + "grad_norm": 1.8763848543167114, + "learning_rate": 5.848254931714719e-06, + "loss": 0.2451, + "step": 1542 + }, + { + "epoch": 0.4453532469170277, + "grad_norm": 3.2289862632751465, + "learning_rate": 5.839150227617603e-06, + "loss": 0.2773, + "step": 1545 + }, + { + "epoch": 0.44621801050327436, + "grad_norm": 2.1212282180786133, + "learning_rate": 5.830045523520486e-06, + "loss": 0.2212, + "step": 1548 + }, + { + "epoch": 0.4470827740895211, + "grad_norm": 3.2190654277801514, + "learning_rate": 5.8209408194233695e-06, + "loss": 0.2574, + "step": 1551 + }, + { + "epoch": 0.4479475376757677, + "grad_norm": 2.9537158012390137, + "learning_rate": 5.811836115326252e-06, + "loss": 0.2216, + "step": 1554 + }, + { + "epoch": 0.4488123012620144, + "grad_norm": 4.764918327331543, + "learning_rate": 5.802731411229135e-06, + "loss": 0.2479, + "step": 1557 + }, + { + "epoch": 0.44967706484826103, + "grad_norm": 2.5907225608825684, + "learning_rate": 5.793626707132019e-06, + "loss": 0.2266, + "step": 1560 + }, + { + "epoch": 0.4505418284345077, + "grad_norm": 6.063370704650879, + "learning_rate": 5.784522003034902e-06, + "loss": 0.2378, + "step": 1563 + }, + { + "epoch": 0.45140659202075434, + "grad_norm": 2.465559720993042, + "learning_rate": 5.775417298937785e-06, + "loss": 0.2762, + "step": 1566 + }, + { + "epoch": 0.452271355607001, + "grad_norm": 2.4620933532714844, + "learning_rate": 5.766312594840668e-06, + "loss": 0.2412, + "step": 1569 + }, + { + "epoch": 0.45313611919324764, + "grad_norm": 1.7580320835113525, + "learning_rate": 5.757207890743551e-06, + "loss": 0.2045, + "step": 1572 + }, + { + "epoch": 0.4540008827794943, + "grad_norm": 3.2482638359069824, + "learning_rate": 5.748103186646434e-06, + "loss": 0.2528, + "step": 1575 + }, + { + "epoch": 0.45486564636574095, + "grad_norm": 1.7835358381271362, + "learning_rate": 5.738998482549318e-06, + "loss": 0.2245, + "step": 1578 + }, + { + "epoch": 0.4557304099519876, + "grad_norm": 2.004756450653076, + "learning_rate": 5.729893778452201e-06, + "loss": 0.246, + "step": 1581 + }, + { + "epoch": 0.45659517353823426, + "grad_norm": 2.1401569843292236, + "learning_rate": 5.720789074355084e-06, + "loss": 0.221, + "step": 1584 + }, + { + "epoch": 0.4574599371244809, + "grad_norm": 2.868241786956787, + "learning_rate": 5.711684370257967e-06, + "loss": 0.2211, + "step": 1587 + }, + { + "epoch": 0.45832470071072756, + "grad_norm": 2.1547744274139404, + "learning_rate": 5.70257966616085e-06, + "loss": 0.2259, + "step": 1590 + }, + { + "epoch": 0.4591894642969742, + "grad_norm": 3.7882297039031982, + "learning_rate": 5.693474962063733e-06, + "loss": 0.2486, + "step": 1593 + }, + { + "epoch": 0.46005422788322087, + "grad_norm": 2.8157854080200195, + "learning_rate": 5.684370257966616e-06, + "loss": 0.2299, + "step": 1596 + }, + { + "epoch": 0.4609189914694675, + "grad_norm": 2.409140110015869, + "learning_rate": 5.6752655538695e-06, + "loss": 0.2459, + "step": 1599 + }, + { + "epoch": 0.46120724599821644, + "eval_loss": 0.2246265709400177, + "eval_mse": 0.22462656778097154, + "eval_runtime": 64.6238, + "eval_samples_per_second": 15.474, + "eval_steps_per_second": 3.869, + "step": 1600 + }, + { + "epoch": 0.4617837550557142, + "grad_norm": 4.253114223480225, + "learning_rate": 5.666160849772383e-06, + "loss": 0.2543, + "step": 1602 + }, + { + "epoch": 0.4626485186419608, + "grad_norm": 1.6027945280075073, + "learning_rate": 5.657056145675266e-06, + "loss": 0.224, + "step": 1605 + }, + { + "epoch": 0.46351328222820753, + "grad_norm": 4.508523941040039, + "learning_rate": 5.647951441578149e-06, + "loss": 0.2278, + "step": 1608 + }, + { + "epoch": 0.4643780458144542, + "grad_norm": 2.6588051319122314, + "learning_rate": 5.6388467374810314e-06, + "loss": 0.2368, + "step": 1611 + }, + { + "epoch": 0.46524280940070084, + "grad_norm": 2.123340129852295, + "learning_rate": 5.629742033383916e-06, + "loss": 0.2308, + "step": 1614 + }, + { + "epoch": 0.4661075729869475, + "grad_norm": 4.028223991394043, + "learning_rate": 5.620637329286799e-06, + "loss": 0.2405, + "step": 1617 + }, + { + "epoch": 0.46697233657319415, + "grad_norm": 1.7611653804779053, + "learning_rate": 5.611532625189682e-06, + "loss": 0.2131, + "step": 1620 + }, + { + "epoch": 0.4678371001594408, + "grad_norm": 2.331615924835205, + "learning_rate": 5.6024279210925645e-06, + "loss": 0.2435, + "step": 1623 + }, + { + "epoch": 0.46870186374568745, + "grad_norm": 1.896458625793457, + "learning_rate": 5.593323216995447e-06, + "loss": 0.2447, + "step": 1626 + }, + { + "epoch": 0.4695666273319341, + "grad_norm": 3.185960292816162, + "learning_rate": 5.584218512898332e-06, + "loss": 0.2412, + "step": 1629 + }, + { + "epoch": 0.47043139091818076, + "grad_norm": 2.1739768981933594, + "learning_rate": 5.575113808801215e-06, + "loss": 0.2396, + "step": 1632 + }, + { + "epoch": 0.4712961545044274, + "grad_norm": 1.8949024677276611, + "learning_rate": 5.5660091047040976e-06, + "loss": 0.2152, + "step": 1635 + }, + { + "epoch": 0.47216091809067406, + "grad_norm": 2.536170482635498, + "learning_rate": 5.55690440060698e-06, + "loss": 0.2447, + "step": 1638 + }, + { + "epoch": 0.4730256816769207, + "grad_norm": 2.3221828937530518, + "learning_rate": 5.547799696509863e-06, + "loss": 0.2497, + "step": 1641 + }, + { + "epoch": 0.47389044526316737, + "grad_norm": 2.3624842166900635, + "learning_rate": 5.538694992412748e-06, + "loss": 0.2058, + "step": 1644 + }, + { + "epoch": 0.474755208849414, + "grad_norm": 2.1354258060455322, + "learning_rate": 5.529590288315631e-06, + "loss": 0.2133, + "step": 1647 + }, + { + "epoch": 0.4756199724356607, + "grad_norm": 2.2622838020324707, + "learning_rate": 5.5204855842185135e-06, + "loss": 0.247, + "step": 1650 + }, + { + "epoch": 0.47648473602190733, + "grad_norm": 3.781003952026367, + "learning_rate": 5.511380880121396e-06, + "loss": 0.2231, + "step": 1653 + }, + { + "epoch": 0.477349499608154, + "grad_norm": 4.45161247253418, + "learning_rate": 5.502276176024279e-06, + "loss": 0.26, + "step": 1656 + }, + { + "epoch": 0.47821426319440064, + "grad_norm": 2.7861571311950684, + "learning_rate": 5.493171471927162e-06, + "loss": 0.21, + "step": 1659 + }, + { + "epoch": 0.4790790267806473, + "grad_norm": 2.1872150897979736, + "learning_rate": 5.4840667678300465e-06, + "loss": 0.2048, + "step": 1662 + }, + { + "epoch": 0.47994379036689394, + "grad_norm": 5.789909839630127, + "learning_rate": 5.474962063732929e-06, + "loss": 0.2213, + "step": 1665 + }, + { + "epoch": 0.48080855395314065, + "grad_norm": 2.418182849884033, + "learning_rate": 5.465857359635812e-06, + "loss": 0.27, + "step": 1668 + }, + { + "epoch": 0.4816733175393873, + "grad_norm": 2.442600727081299, + "learning_rate": 5.456752655538695e-06, + "loss": 0.2234, + "step": 1671 + }, + { + "epoch": 0.48253808112563396, + "grad_norm": 5.272119522094727, + "learning_rate": 5.447647951441578e-06, + "loss": 0.2374, + "step": 1674 + }, + { + "epoch": 0.4834028447118806, + "grad_norm": 2.1721715927124023, + "learning_rate": 5.4385432473444624e-06, + "loss": 0.2081, + "step": 1677 + }, + { + "epoch": 0.48426760829812726, + "grad_norm": 2.432861328125, + "learning_rate": 5.429438543247345e-06, + "loss": 0.2488, + "step": 1680 + }, + { + "epoch": 0.4851323718843739, + "grad_norm": 2.196579933166504, + "learning_rate": 5.420333839150228e-06, + "loss": 0.231, + "step": 1683 + }, + { + "epoch": 0.48599713547062057, + "grad_norm": 2.4169063568115234, + "learning_rate": 5.411229135053111e-06, + "loss": 0.2145, + "step": 1686 + }, + { + "epoch": 0.4868618990568672, + "grad_norm": 3.8229868412017822, + "learning_rate": 5.402124430955994e-06, + "loss": 0.2518, + "step": 1689 + }, + { + "epoch": 0.4877266626431139, + "grad_norm": 2.484478235244751, + "learning_rate": 5.393019726858878e-06, + "loss": 0.2145, + "step": 1692 + }, + { + "epoch": 0.4885914262293605, + "grad_norm": 2.7220754623413086, + "learning_rate": 5.383915022761761e-06, + "loss": 0.2584, + "step": 1695 + }, + { + "epoch": 0.4894561898156072, + "grad_norm": 2.954333543777466, + "learning_rate": 5.374810318664644e-06, + "loss": 0.2403, + "step": 1698 + }, + { + "epoch": 0.49003269887310497, + "eval_loss": 0.22324734926223755, + "eval_mse": 0.22324735438660717, + "eval_runtime": 64.5172, + "eval_samples_per_second": 15.5, + "eval_steps_per_second": 3.875, + "step": 1700 + }, + { + "epoch": 0.49032095340185383, + "grad_norm": 3.1045334339141846, + "learning_rate": 5.365705614567527e-06, + "loss": 0.2655, + "step": 1701 + }, + { + "epoch": 0.4911857169881005, + "grad_norm": 2.5739636421203613, + "learning_rate": 5.35660091047041e-06, + "loss": 0.245, + "step": 1704 + }, + { + "epoch": 0.49205048057434714, + "grad_norm": 2.3004565238952637, + "learning_rate": 5.347496206373293e-06, + "loss": 0.2261, + "step": 1707 + }, + { + "epoch": 0.4929152441605938, + "grad_norm": 2.6713101863861084, + "learning_rate": 5.338391502276177e-06, + "loss": 0.2152, + "step": 1710 + }, + { + "epoch": 0.49378000774684044, + "grad_norm": 2.055208683013916, + "learning_rate": 5.32928679817906e-06, + "loss": 0.2247, + "step": 1713 + }, + { + "epoch": 0.4946447713330871, + "grad_norm": 2.4551613330841064, + "learning_rate": 5.320182094081943e-06, + "loss": 0.2494, + "step": 1716 + }, + { + "epoch": 0.49550953491933375, + "grad_norm": 2.6361169815063477, + "learning_rate": 5.311077389984826e-06, + "loss": 0.261, + "step": 1719 + }, + { + "epoch": 0.4963742985055804, + "grad_norm": 2.158512830734253, + "learning_rate": 5.301972685887709e-06, + "loss": 0.2106, + "step": 1722 + }, + { + "epoch": 0.4972390620918271, + "grad_norm": 2.550480365753174, + "learning_rate": 5.292867981790593e-06, + "loss": 0.1984, + "step": 1725 + }, + { + "epoch": 0.49810382567807376, + "grad_norm": 5.54836893081665, + "learning_rate": 5.283763277693476e-06, + "loss": 0.2339, + "step": 1728 + }, + { + "epoch": 0.4989685892643204, + "grad_norm": 1.3806259632110596, + "learning_rate": 5.274658573596359e-06, + "loss": 0.2622, + "step": 1731 + }, + { + "epoch": 0.49983335285056707, + "grad_norm": 2.9417316913604736, + "learning_rate": 5.2655538694992415e-06, + "loss": 0.2401, + "step": 1734 + }, + { + "epoch": 0.5006981164368137, + "grad_norm": 3.094886541366577, + "learning_rate": 5.256449165402124e-06, + "loss": 0.2189, + "step": 1737 + }, + { + "epoch": 0.5015628800230604, + "grad_norm": 2.4346110820770264, + "learning_rate": 5.247344461305008e-06, + "loss": 0.2029, + "step": 1740 + }, + { + "epoch": 0.502427643609307, + "grad_norm": 2.5410044193267822, + "learning_rate": 5.238239757207892e-06, + "loss": 0.2417, + "step": 1743 + }, + { + "epoch": 0.5032924071955537, + "grad_norm": 2.1265604496002197, + "learning_rate": 5.229135053110775e-06, + "loss": 0.2476, + "step": 1746 + }, + { + "epoch": 0.5041571707818003, + "grad_norm": 2.449608087539673, + "learning_rate": 5.2200303490136574e-06, + "loss": 0.2413, + "step": 1749 + }, + { + "epoch": 0.505021934368047, + "grad_norm": 5.359190464019775, + "learning_rate": 5.21092564491654e-06, + "loss": 0.2459, + "step": 1752 + }, + { + "epoch": 0.5058866979542936, + "grad_norm": 6.479160785675049, + "learning_rate": 5.201820940819424e-06, + "loss": 0.2257, + "step": 1755 + }, + { + "epoch": 0.5067514615405403, + "grad_norm": 3.1292057037353516, + "learning_rate": 5.192716236722307e-06, + "loss": 0.2508, + "step": 1758 + }, + { + "epoch": 0.507616225126787, + "grad_norm": 5.284337997436523, + "learning_rate": 5.1836115326251905e-06, + "loss": 0.2817, + "step": 1761 + }, + { + "epoch": 0.5084809887130336, + "grad_norm": 2.8322136402130127, + "learning_rate": 5.174506828528073e-06, + "loss": 0.2358, + "step": 1764 + }, + { + "epoch": 0.5093457522992803, + "grad_norm": 2.5609347820281982, + "learning_rate": 5.165402124430956e-06, + "loss": 0.2321, + "step": 1767 + }, + { + "epoch": 0.5102105158855269, + "grad_norm": 3.2140045166015625, + "learning_rate": 5.15629742033384e-06, + "loss": 0.2335, + "step": 1770 + }, + { + "epoch": 0.5110752794717736, + "grad_norm": 2.619800567626953, + "learning_rate": 5.147192716236723e-06, + "loss": 0.2544, + "step": 1773 + }, + { + "epoch": 0.5119400430580202, + "grad_norm": 3.5388903617858887, + "learning_rate": 5.1380880121396055e-06, + "loss": 0.2797, + "step": 1776 + }, + { + "epoch": 0.5128048066442669, + "grad_norm": 4.132437705993652, + "learning_rate": 5.128983308042489e-06, + "loss": 0.2722, + "step": 1779 + }, + { + "epoch": 0.5136695702305135, + "grad_norm": 1.7023959159851074, + "learning_rate": 5.119878603945372e-06, + "loss": 0.2563, + "step": 1782 + }, + { + "epoch": 0.5145343338167602, + "grad_norm": 3.2654476165771484, + "learning_rate": 5.110773899848256e-06, + "loss": 0.2547, + "step": 1785 + }, + { + "epoch": 0.5153990974030068, + "grad_norm": 2.9560678005218506, + "learning_rate": 5.101669195751139e-06, + "loss": 0.2019, + "step": 1788 + }, + { + "epoch": 0.5162638609892535, + "grad_norm": 3.0013792514801025, + "learning_rate": 5.0925644916540215e-06, + "loss": 0.2237, + "step": 1791 + }, + { + "epoch": 0.5171286245755001, + "grad_norm": 5.049393653869629, + "learning_rate": 5.083459787556905e-06, + "loss": 0.2326, + "step": 1794 + }, + { + "epoch": 0.5179933881617468, + "grad_norm": 2.8161003589630127, + "learning_rate": 5.074355083459788e-06, + "loss": 0.255, + "step": 1797 + }, + { + "epoch": 0.5188581517479934, + "grad_norm": 3.8299570083618164, + "learning_rate": 5.065250379362671e-06, + "loss": 0.251, + "step": 1800 + }, + { + "epoch": 0.5188581517479934, + "eval_loss": 0.24214114248752594, + "eval_mse": 0.2421411427827552, + "eval_runtime": 64.4904, + "eval_samples_per_second": 15.506, + "eval_steps_per_second": 3.877, + "step": 1800 + }, + { + "epoch": 0.5197229153342401, + "grad_norm": 6.20158576965332, + "learning_rate": 5.0561456752655545e-06, + "loss": 0.2574, + "step": 1803 + }, + { + "epoch": 0.5205876789204867, + "grad_norm": 3.9144914150238037, + "learning_rate": 5.047040971168437e-06, + "loss": 0.2249, + "step": 1806 + }, + { + "epoch": 0.5214524425067334, + "grad_norm": 4.534792423248291, + "learning_rate": 5.03793626707132e-06, + "loss": 0.2357, + "step": 1809 + }, + { + "epoch": 0.52231720609298, + "grad_norm": 2.006342649459839, + "learning_rate": 5.028831562974204e-06, + "loss": 0.2434, + "step": 1812 + }, + { + "epoch": 0.5231819696792268, + "grad_norm": 2.1051185131073, + "learning_rate": 5.019726858877087e-06, + "loss": 0.2159, + "step": 1815 + }, + { + "epoch": 0.5240467332654735, + "grad_norm": 2.2918484210968018, + "learning_rate": 5.0106221547799704e-06, + "loss": 0.2254, + "step": 1818 + }, + { + "epoch": 0.5249114968517201, + "grad_norm": 2.324146270751953, + "learning_rate": 5.001517450682853e-06, + "loss": 0.2593, + "step": 1821 + }, + { + "epoch": 0.5257762604379668, + "grad_norm": 2.0753884315490723, + "learning_rate": 4.992412746585736e-06, + "loss": 0.2344, + "step": 1824 + }, + { + "epoch": 0.5266410240242134, + "grad_norm": 2.431459426879883, + "learning_rate": 4.983308042488619e-06, + "loss": 0.2544, + "step": 1827 + }, + { + "epoch": 0.5275057876104601, + "grad_norm": 3.231693983078003, + "learning_rate": 4.974203338391503e-06, + "loss": 0.2232, + "step": 1830 + }, + { + "epoch": 0.5283705511967067, + "grad_norm": 4.636984825134277, + "learning_rate": 4.9650986342943855e-06, + "loss": 0.2505, + "step": 1833 + }, + { + "epoch": 0.5292353147829534, + "grad_norm": 1.6613258123397827, + "learning_rate": 4.955993930197269e-06, + "loss": 0.2259, + "step": 1836 + }, + { + "epoch": 0.5301000783692, + "grad_norm": 2.0384411811828613, + "learning_rate": 4.946889226100152e-06, + "loss": 0.2459, + "step": 1839 + }, + { + "epoch": 0.5309648419554467, + "grad_norm": 3.353332281112671, + "learning_rate": 4.937784522003035e-06, + "loss": 0.2132, + "step": 1842 + }, + { + "epoch": 0.5318296055416933, + "grad_norm": 2.3782637119293213, + "learning_rate": 4.9286798179059185e-06, + "loss": 0.2531, + "step": 1845 + }, + { + "epoch": 0.53269436912794, + "grad_norm": 2.1125853061676025, + "learning_rate": 4.919575113808801e-06, + "loss": 0.247, + "step": 1848 + }, + { + "epoch": 0.5335591327141866, + "grad_norm": 2.698594808578491, + "learning_rate": 4.910470409711684e-06, + "loss": 0.2599, + "step": 1851 + }, + { + "epoch": 0.5344238963004333, + "grad_norm": 3.285856008529663, + "learning_rate": 4.901365705614568e-06, + "loss": 0.2242, + "step": 1854 + }, + { + "epoch": 0.53528865988668, + "grad_norm": 2.558708429336548, + "learning_rate": 4.892261001517451e-06, + "loss": 0.2629, + "step": 1857 + }, + { + "epoch": 0.5361534234729266, + "grad_norm": 1.9437116384506226, + "learning_rate": 4.8831562974203345e-06, + "loss": 0.2363, + "step": 1860 + }, + { + "epoch": 0.5370181870591733, + "grad_norm": 1.8736016750335693, + "learning_rate": 4.874051593323217e-06, + "loss": 0.231, + "step": 1863 + }, + { + "epoch": 0.5378829506454199, + "grad_norm": 3.5459728240966797, + "learning_rate": 4.8649468892261e-06, + "loss": 0.2617, + "step": 1866 + }, + { + "epoch": 0.5387477142316666, + "grad_norm": 2.2355449199676514, + "learning_rate": 4.855842185128984e-06, + "loss": 0.221, + "step": 1869 + }, + { + "epoch": 0.5396124778179132, + "grad_norm": 3.19442081451416, + "learning_rate": 4.846737481031867e-06, + "loss": 0.2517, + "step": 1872 + }, + { + "epoch": 0.5404772414041599, + "grad_norm": 3.5516788959503174, + "learning_rate": 4.8376327769347495e-06, + "loss": 0.228, + "step": 1875 + }, + { + "epoch": 0.5413420049904065, + "grad_norm": 3.464395523071289, + "learning_rate": 4.828528072837633e-06, + "loss": 0.2469, + "step": 1878 + }, + { + "epoch": 0.5422067685766532, + "grad_norm": 2.241856813430786, + "learning_rate": 4.819423368740516e-06, + "loss": 0.239, + "step": 1881 + }, + { + "epoch": 0.5430715321628998, + "grad_norm": 2.296264886856079, + "learning_rate": 4.8103186646434e-06, + "loss": 0.204, + "step": 1884 + }, + { + "epoch": 0.5439362957491465, + "grad_norm": 1.9935134649276733, + "learning_rate": 4.801213960546283e-06, + "loss": 0.237, + "step": 1887 + }, + { + "epoch": 0.5448010593353931, + "grad_norm": 4.519099235534668, + "learning_rate": 4.792109256449165e-06, + "loss": 0.2981, + "step": 1890 + }, + { + "epoch": 0.5456658229216398, + "grad_norm": 2.157167673110962, + "learning_rate": 4.783004552352049e-06, + "loss": 0.221, + "step": 1893 + }, + { + "epoch": 0.5465305865078864, + "grad_norm": 1.8143630027770996, + "learning_rate": 4.773899848254932e-06, + "loss": 0.2221, + "step": 1896 + }, + { + "epoch": 0.5473953500941331, + "grad_norm": 3.750279188156128, + "learning_rate": 4.764795144157816e-06, + "loss": 0.2565, + "step": 1899 + }, + { + "epoch": 0.547683604622882, + "eval_loss": 0.2207367867231369, + "eval_mse": 0.2207367917916272, + "eval_runtime": 64.5517, + "eval_samples_per_second": 15.491, + "eval_steps_per_second": 3.873, + "step": 1900 + }, + { + "epoch": 0.5482601136803797, + "grad_norm": 2.6302292346954346, + "learning_rate": 4.7556904400606985e-06, + "loss": 0.2202, + "step": 1902 + }, + { + "epoch": 0.5491248772666264, + "grad_norm": 13.511824607849121, + "learning_rate": 4.746585735963581e-06, + "loss": 0.2335, + "step": 1905 + }, + { + "epoch": 0.549989640852873, + "grad_norm": 2.213571548461914, + "learning_rate": 4.737481031866465e-06, + "loss": 0.2219, + "step": 1908 + }, + { + "epoch": 0.5508544044391197, + "grad_norm": 2.6418211460113525, + "learning_rate": 4.728376327769348e-06, + "loss": 0.2317, + "step": 1911 + }, + { + "epoch": 0.5517191680253664, + "grad_norm": 2.2119081020355225, + "learning_rate": 4.719271623672231e-06, + "loss": 0.2468, + "step": 1914 + }, + { + "epoch": 0.552583931611613, + "grad_norm": 2.070915460586548, + "learning_rate": 4.710166919575114e-06, + "loss": 0.2503, + "step": 1917 + }, + { + "epoch": 0.5534486951978597, + "grad_norm": 3.188049077987671, + "learning_rate": 4.701062215477997e-06, + "loss": 0.2212, + "step": 1920 + }, + { + "epoch": 0.5543134587841063, + "grad_norm": 2.545793294906616, + "learning_rate": 4.691957511380881e-06, + "loss": 0.2107, + "step": 1923 + }, + { + "epoch": 0.555178222370353, + "grad_norm": 1.9591479301452637, + "learning_rate": 4.682852807283764e-06, + "loss": 0.2208, + "step": 1926 + }, + { + "epoch": 0.5560429859565996, + "grad_norm": 4.691300392150879, + "learning_rate": 4.673748103186647e-06, + "loss": 0.2373, + "step": 1929 + }, + { + "epoch": 0.5569077495428464, + "grad_norm": 2.1649136543273926, + "learning_rate": 4.66464339908953e-06, + "loss": 0.2126, + "step": 1932 + }, + { + "epoch": 0.557772513129093, + "grad_norm": 1.9450371265411377, + "learning_rate": 4.655538694992413e-06, + "loss": 0.2137, + "step": 1935 + }, + { + "epoch": 0.5586372767153397, + "grad_norm": 2.2597336769104004, + "learning_rate": 4.646433990895296e-06, + "loss": 0.2073, + "step": 1938 + }, + { + "epoch": 0.5595020403015863, + "grad_norm": 1.523226022720337, + "learning_rate": 4.63732928679818e-06, + "loss": 0.2259, + "step": 1941 + }, + { + "epoch": 0.560366803887833, + "grad_norm": 2.4895904064178467, + "learning_rate": 4.6282245827010625e-06, + "loss": 0.2412, + "step": 1944 + }, + { + "epoch": 0.5612315674740797, + "grad_norm": 1.6229183673858643, + "learning_rate": 4.619119878603946e-06, + "loss": 0.213, + "step": 1947 + }, + { + "epoch": 0.5620963310603263, + "grad_norm": 3.478339672088623, + "learning_rate": 4.610015174506829e-06, + "loss": 0.2821, + "step": 1950 + }, + { + "epoch": 0.562961094646573, + "grad_norm": 1.5027942657470703, + "learning_rate": 4.600910470409712e-06, + "loss": 0.2346, + "step": 1953 + }, + { + "epoch": 0.5638258582328196, + "grad_norm": 38.24848556518555, + "learning_rate": 4.591805766312596e-06, + "loss": 0.2452, + "step": 1956 + }, + { + "epoch": 0.5646906218190663, + "grad_norm": 2.314491033554077, + "learning_rate": 4.582701062215478e-06, + "loss": 0.2634, + "step": 1959 + }, + { + "epoch": 0.5655553854053129, + "grad_norm": 2.920741081237793, + "learning_rate": 4.573596358118362e-06, + "loss": 0.209, + "step": 1962 + }, + { + "epoch": 0.5664201489915596, + "grad_norm": 3.8211729526519775, + "learning_rate": 4.564491654021245e-06, + "loss": 0.2145, + "step": 1965 + }, + { + "epoch": 0.5672849125778062, + "grad_norm": 3.976433753967285, + "learning_rate": 4.555386949924128e-06, + "loss": 0.2308, + "step": 1968 + }, + { + "epoch": 0.5681496761640529, + "grad_norm": 2.2235753536224365, + "learning_rate": 4.5462822458270115e-06, + "loss": 0.1959, + "step": 1971 + }, + { + "epoch": 0.5690144397502995, + "grad_norm": 2.7206735610961914, + "learning_rate": 4.537177541729894e-06, + "loss": 0.221, + "step": 1974 + }, + { + "epoch": 0.5698792033365462, + "grad_norm": 1.9556628465652466, + "learning_rate": 4.528072837632777e-06, + "loss": 0.212, + "step": 1977 + }, + { + "epoch": 0.5707439669227928, + "grad_norm": 2.0342814922332764, + "learning_rate": 4.518968133535661e-06, + "loss": 0.2388, + "step": 1980 + }, + { + "epoch": 0.5716087305090395, + "grad_norm": 1.7735626697540283, + "learning_rate": 4.509863429438544e-06, + "loss": 0.2227, + "step": 1983 + }, + { + "epoch": 0.5724734940952861, + "grad_norm": 1.7684601545333862, + "learning_rate": 4.500758725341427e-06, + "loss": 0.2195, + "step": 1986 + }, + { + "epoch": 0.5733382576815328, + "grad_norm": 1.8815613985061646, + "learning_rate": 4.49165402124431e-06, + "loss": 0.262, + "step": 1989 + }, + { + "epoch": 0.5742030212677794, + "grad_norm": 1.985158085823059, + "learning_rate": 4.482549317147193e-06, + "loss": 0.2109, + "step": 1992 + }, + { + "epoch": 0.5750677848540261, + "grad_norm": 3.0801327228546143, + "learning_rate": 4.473444613050077e-06, + "loss": 0.2017, + "step": 1995 + }, + { + "epoch": 0.5759325484402728, + "grad_norm": 1.847817301750183, + "learning_rate": 4.46433990895296e-06, + "loss": 0.2274, + "step": 1998 + }, + { + "epoch": 0.5765090574977705, + "eval_loss": 0.22936804592609406, + "eval_mse": 0.22936805277447275, + "eval_runtime": 64.5577, + "eval_samples_per_second": 15.49, + "eval_steps_per_second": 3.873, + "step": 2000 + }, + { + "epoch": 0.5767973120265194, + "grad_norm": 2.9898228645324707, + "learning_rate": 4.4552352048558425e-06, + "loss": 0.2044, + "step": 2001 + }, + { + "epoch": 0.5776620756127661, + "grad_norm": 3.691619873046875, + "learning_rate": 4.446130500758726e-06, + "loss": 0.2331, + "step": 2004 + }, + { + "epoch": 0.5785268391990127, + "grad_norm": 1.9646469354629517, + "learning_rate": 4.437025796661609e-06, + "loss": 0.2331, + "step": 2007 + }, + { + "epoch": 0.5793916027852594, + "grad_norm": 1.9815963506698608, + "learning_rate": 4.427921092564492e-06, + "loss": 0.2068, + "step": 2010 + }, + { + "epoch": 0.580256366371506, + "grad_norm": 4.30540657043457, + "learning_rate": 4.4188163884673755e-06, + "loss": 0.2588, + "step": 2013 + }, + { + "epoch": 0.5811211299577527, + "grad_norm": 2.5641026496887207, + "learning_rate": 4.409711684370258e-06, + "loss": 0.2704, + "step": 2016 + }, + { + "epoch": 0.5819858935439993, + "grad_norm": 2.0352160930633545, + "learning_rate": 4.400606980273141e-06, + "loss": 0.2249, + "step": 2019 + }, + { + "epoch": 0.582850657130246, + "grad_norm": 2.3772523403167725, + "learning_rate": 4.391502276176025e-06, + "loss": 0.2381, + "step": 2022 + }, + { + "epoch": 0.5837154207164926, + "grad_norm": 2.5510172843933105, + "learning_rate": 4.382397572078908e-06, + "loss": 0.2464, + "step": 2025 + }, + { + "epoch": 0.5845801843027393, + "grad_norm": 2.5375447273254395, + "learning_rate": 4.3732928679817906e-06, + "loss": 0.2299, + "step": 2028 + }, + { + "epoch": 0.5854449478889859, + "grad_norm": 4.41203498840332, + "learning_rate": 4.364188163884674e-06, + "loss": 0.2281, + "step": 2031 + }, + { + "epoch": 0.5863097114752326, + "grad_norm": 1.7787522077560425, + "learning_rate": 4.355083459787557e-06, + "loss": 0.2518, + "step": 2034 + }, + { + "epoch": 0.5871744750614792, + "grad_norm": 2.4308512210845947, + "learning_rate": 4.34597875569044e-06, + "loss": 0.2621, + "step": 2037 + }, + { + "epoch": 0.5880392386477259, + "grad_norm": 3.5610814094543457, + "learning_rate": 4.336874051593324e-06, + "loss": 0.2537, + "step": 2040 + }, + { + "epoch": 0.5889040022339725, + "grad_norm": 3.633169174194336, + "learning_rate": 4.3277693474962065e-06, + "loss": 0.2533, + "step": 2043 + }, + { + "epoch": 0.5897687658202192, + "grad_norm": 2.8986027240753174, + "learning_rate": 4.31866464339909e-06, + "loss": 0.2156, + "step": 2046 + }, + { + "epoch": 0.590633529406466, + "grad_norm": 1.8997621536254883, + "learning_rate": 4.309559939301973e-06, + "loss": 0.191, + "step": 2049 + }, + { + "epoch": 0.5914982929927126, + "grad_norm": 3.2994544506073, + "learning_rate": 4.300455235204856e-06, + "loss": 0.2083, + "step": 2052 + }, + { + "epoch": 0.5923630565789593, + "grad_norm": 2.4655439853668213, + "learning_rate": 4.2913505311077395e-06, + "loss": 0.2369, + "step": 2055 + }, + { + "epoch": 0.5932278201652059, + "grad_norm": 1.669376254081726, + "learning_rate": 4.282245827010622e-06, + "loss": 0.226, + "step": 2058 + }, + { + "epoch": 0.5940925837514526, + "grad_norm": 2.234907388687134, + "learning_rate": 4.273141122913505e-06, + "loss": 0.207, + "step": 2061 + }, + { + "epoch": 0.5949573473376992, + "grad_norm": 5.61646032333374, + "learning_rate": 4.264036418816389e-06, + "loss": 0.2293, + "step": 2064 + }, + { + "epoch": 0.5958221109239459, + "grad_norm": 6.289272785186768, + "learning_rate": 4.254931714719272e-06, + "loss": 0.1854, + "step": 2067 + }, + { + "epoch": 0.5966868745101925, + "grad_norm": 3.474231481552124, + "learning_rate": 4.245827010622155e-06, + "loss": 0.2221, + "step": 2070 + }, + { + "epoch": 0.5975516380964392, + "grad_norm": 3.175489902496338, + "learning_rate": 4.236722306525038e-06, + "loss": 0.2118, + "step": 2073 + }, + { + "epoch": 0.5984164016826858, + "grad_norm": 3.48205304145813, + "learning_rate": 4.227617602427921e-06, + "loss": 0.2107, + "step": 2076 + }, + { + "epoch": 0.5992811652689325, + "grad_norm": 3.100991725921631, + "learning_rate": 4.218512898330804e-06, + "loss": 0.2238, + "step": 2079 + }, + { + "epoch": 0.6001459288551791, + "grad_norm": 2.0803310871124268, + "learning_rate": 4.209408194233688e-06, + "loss": 0.2469, + "step": 2082 + }, + { + "epoch": 0.6010106924414258, + "grad_norm": 1.8892669677734375, + "learning_rate": 4.2003034901365705e-06, + "loss": 0.1876, + "step": 2085 + }, + { + "epoch": 0.6018754560276725, + "grad_norm": 2.093517541885376, + "learning_rate": 4.191198786039454e-06, + "loss": 0.2136, + "step": 2088 + }, + { + "epoch": 0.6027402196139191, + "grad_norm": 2.2591047286987305, + "learning_rate": 4.182094081942337e-06, + "loss": 0.2309, + "step": 2091 + }, + { + "epoch": 0.6036049832001658, + "grad_norm": 1.6561003923416138, + "learning_rate": 4.17298937784522e-06, + "loss": 0.1958, + "step": 2094 + }, + { + "epoch": 0.6044697467864124, + "grad_norm": 4.529551029205322, + "learning_rate": 4.1638846737481036e-06, + "loss": 0.2551, + "step": 2097 + }, + { + "epoch": 0.6053345103726591, + "grad_norm": 3.610400915145874, + "learning_rate": 4.154779969650986e-06, + "loss": 0.2272, + "step": 2100 + }, + { + "epoch": 0.6053345103726591, + "eval_loss": 0.21923716366291046, + "eval_mse": 0.21923717018315803, + "eval_runtime": 64.6466, + "eval_samples_per_second": 15.469, + "eval_steps_per_second": 3.867, + "step": 2100 + }, + { + "epoch": 0.6061992739589057, + "grad_norm": 5.296643257141113, + "learning_rate": 4.145675265553869e-06, + "loss": 0.2147, + "step": 2103 + }, + { + "epoch": 0.6070640375451524, + "grad_norm": 1.9191397428512573, + "learning_rate": 4.136570561456753e-06, + "loss": 0.2303, + "step": 2106 + }, + { + "epoch": 0.607928801131399, + "grad_norm": 5.336986541748047, + "learning_rate": 4.127465857359636e-06, + "loss": 0.2544, + "step": 2109 + }, + { + "epoch": 0.6087935647176457, + "grad_norm": 2.223107099533081, + "learning_rate": 4.1183611532625195e-06, + "loss": 0.2246, + "step": 2112 + }, + { + "epoch": 0.6096583283038923, + "grad_norm": 3.1443583965301514, + "learning_rate": 4.109256449165402e-06, + "loss": 0.2594, + "step": 2115 + }, + { + "epoch": 0.610523091890139, + "grad_norm": 2.582916498184204, + "learning_rate": 4.100151745068285e-06, + "loss": 0.2307, + "step": 2118 + }, + { + "epoch": 0.6113878554763856, + "grad_norm": 3.238401412963867, + "learning_rate": 4.091047040971169e-06, + "loss": 0.2595, + "step": 2121 + }, + { + "epoch": 0.6122526190626323, + "grad_norm": 4.401957988739014, + "learning_rate": 4.081942336874052e-06, + "loss": 0.2156, + "step": 2124 + }, + { + "epoch": 0.6131173826488789, + "grad_norm": 1.7487517595291138, + "learning_rate": 4.072837632776935e-06, + "loss": 0.2376, + "step": 2127 + }, + { + "epoch": 0.6139821462351256, + "grad_norm": 2.8765621185302734, + "learning_rate": 4.063732928679818e-06, + "loss": 0.22, + "step": 2130 + }, + { + "epoch": 0.6148469098213722, + "grad_norm": 2.272740125656128, + "learning_rate": 4.054628224582701e-06, + "loss": 0.2528, + "step": 2133 + }, + { + "epoch": 0.6157116734076189, + "grad_norm": 2.2663707733154297, + "learning_rate": 4.045523520485585e-06, + "loss": 0.185, + "step": 2136 + }, + { + "epoch": 0.6165764369938656, + "grad_norm": 1.8243865966796875, + "learning_rate": 4.036418816388468e-06, + "loss": 0.2154, + "step": 2139 + }, + { + "epoch": 0.6174412005801122, + "grad_norm": 3.036996364593506, + "learning_rate": 4.0273141122913504e-06, + "loss": 0.2388, + "step": 2142 + }, + { + "epoch": 0.6183059641663589, + "grad_norm": 2.0164997577667236, + "learning_rate": 4.018209408194234e-06, + "loss": 0.2371, + "step": 2145 + }, + { + "epoch": 0.6191707277526055, + "grad_norm": 3.748368978500366, + "learning_rate": 4.009104704097117e-06, + "loss": 0.2377, + "step": 2148 + }, + { + "epoch": 0.6200354913388522, + "grad_norm": 3.3998799324035645, + "learning_rate": 4.000000000000001e-06, + "loss": 0.2359, + "step": 2151 + }, + { + "epoch": 0.6209002549250988, + "grad_norm": 1.6474953889846802, + "learning_rate": 3.9908952959028835e-06, + "loss": 0.2049, + "step": 2154 + }, + { + "epoch": 0.6217650185113455, + "grad_norm": 2.3346197605133057, + "learning_rate": 3.981790591805766e-06, + "loss": 0.2263, + "step": 2157 + }, + { + "epoch": 0.6226297820975921, + "grad_norm": 2.6755003929138184, + "learning_rate": 3.97268588770865e-06, + "loss": 0.248, + "step": 2160 + }, + { + "epoch": 0.6234945456838388, + "grad_norm": 3.026387929916382, + "learning_rate": 3.963581183611533e-06, + "loss": 0.2784, + "step": 2163 + }, + { + "epoch": 0.6243593092700855, + "grad_norm": 3.390216112136841, + "learning_rate": 3.9544764795144166e-06, + "loss": 0.241, + "step": 2166 + }, + { + "epoch": 0.6252240728563322, + "grad_norm": 5.739169120788574, + "learning_rate": 3.945371775417299e-06, + "loss": 0.226, + "step": 2169 + }, + { + "epoch": 0.6260888364425788, + "grad_norm": 1.7892546653747559, + "learning_rate": 3.936267071320182e-06, + "loss": 0.2337, + "step": 2172 + }, + { + "epoch": 0.6269536000288255, + "grad_norm": 4.317811965942383, + "learning_rate": 3.927162367223066e-06, + "loss": 0.2528, + "step": 2175 + }, + { + "epoch": 0.6278183636150722, + "grad_norm": 2.44039249420166, + "learning_rate": 3.918057663125949e-06, + "loss": 0.2604, + "step": 2178 + }, + { + "epoch": 0.6286831272013188, + "grad_norm": 3.1203863620758057, + "learning_rate": 3.908952959028832e-06, + "loss": 0.2116, + "step": 2181 + }, + { + "epoch": 0.6295478907875655, + "grad_norm": 2.5261828899383545, + "learning_rate": 3.899848254931715e-06, + "loss": 0.2279, + "step": 2184 + }, + { + "epoch": 0.6304126543738121, + "grad_norm": 2.393629550933838, + "learning_rate": 3.890743550834598e-06, + "loss": 0.2338, + "step": 2187 + }, + { + "epoch": 0.6312774179600588, + "grad_norm": 2.138955593109131, + "learning_rate": 3.881638846737482e-06, + "loss": 0.2942, + "step": 2190 + }, + { + "epoch": 0.6321421815463054, + "grad_norm": 2.024564266204834, + "learning_rate": 3.872534142640365e-06, + "loss": 0.2666, + "step": 2193 + }, + { + "epoch": 0.6330069451325521, + "grad_norm": 5.208636283874512, + "learning_rate": 3.8634294385432475e-06, + "loss": 0.2594, + "step": 2196 + }, + { + "epoch": 0.6338717087187987, + "grad_norm": 2.6842122077941895, + "learning_rate": 3.854324734446131e-06, + "loss": 0.2668, + "step": 2199 + }, + { + "epoch": 0.6341599632475475, + "eval_loss": 0.22037872672080994, + "eval_mse": 0.22037871759198607, + "eval_runtime": 64.5846, + "eval_samples_per_second": 15.484, + "eval_steps_per_second": 3.871, + "step": 2200 + }, + { + "epoch": 0.6347364723050454, + "grad_norm": 3.914788246154785, + "learning_rate": 3.845220030349014e-06, + "loss": 0.238, + "step": 2202 + }, + { + "epoch": 0.635601235891292, + "grad_norm": 2.634946823120117, + "learning_rate": 3.836115326251897e-06, + "loss": 0.2302, + "step": 2205 + }, + { + "epoch": 0.6364659994775387, + "grad_norm": 2.131955862045288, + "learning_rate": 3.827010622154781e-06, + "loss": 0.2251, + "step": 2208 + }, + { + "epoch": 0.6373307630637853, + "grad_norm": 2.292926788330078, + "learning_rate": 3.8179059180576634e-06, + "loss": 0.2182, + "step": 2211 + }, + { + "epoch": 0.638195526650032, + "grad_norm": 2.0773963928222656, + "learning_rate": 3.8088012139605467e-06, + "loss": 0.2191, + "step": 2214 + }, + { + "epoch": 0.6390602902362786, + "grad_norm": 7.258689880371094, + "learning_rate": 3.7996965098634296e-06, + "loss": 0.2496, + "step": 2217 + }, + { + "epoch": 0.6399250538225253, + "grad_norm": 2.341298818588257, + "learning_rate": 3.790591805766313e-06, + "loss": 0.2353, + "step": 2220 + }, + { + "epoch": 0.640789817408772, + "grad_norm": 2.6624574661254883, + "learning_rate": 3.781487101669196e-06, + "loss": 0.2242, + "step": 2223 + }, + { + "epoch": 0.6416545809950186, + "grad_norm": 6.106657981872559, + "learning_rate": 3.772382397572079e-06, + "loss": 0.2428, + "step": 2226 + }, + { + "epoch": 0.6425193445812653, + "grad_norm": 2.1354269981384277, + "learning_rate": 3.7632776934749626e-06, + "loss": 0.2239, + "step": 2229 + }, + { + "epoch": 0.6433841081675119, + "grad_norm": 3.34881329536438, + "learning_rate": 3.7541729893778455e-06, + "loss": 0.2513, + "step": 2232 + }, + { + "epoch": 0.6442488717537586, + "grad_norm": 3.310595750808716, + "learning_rate": 3.7450682852807287e-06, + "loss": 0.1986, + "step": 2235 + }, + { + "epoch": 0.6451136353400052, + "grad_norm": 2.3163015842437744, + "learning_rate": 3.735963581183612e-06, + "loss": 0.2228, + "step": 2238 + }, + { + "epoch": 0.6459783989262519, + "grad_norm": 2.8986473083496094, + "learning_rate": 3.726858877086495e-06, + "loss": 0.2542, + "step": 2241 + }, + { + "epoch": 0.6468431625124985, + "grad_norm": 1.9085699319839478, + "learning_rate": 3.717754172989378e-06, + "loss": 0.2215, + "step": 2244 + }, + { + "epoch": 0.6477079260987452, + "grad_norm": 1.8858418464660645, + "learning_rate": 3.7086494688922614e-06, + "loss": 0.2081, + "step": 2247 + }, + { + "epoch": 0.6485726896849918, + "grad_norm": 2.7756083011627197, + "learning_rate": 3.699544764795144e-06, + "loss": 0.2483, + "step": 2250 + }, + { + "epoch": 0.6494374532712385, + "grad_norm": 2.3329155445098877, + "learning_rate": 3.690440060698028e-06, + "loss": 0.2106, + "step": 2253 + }, + { + "epoch": 0.6503022168574851, + "grad_norm": 2.95639967918396, + "learning_rate": 3.6813353566009107e-06, + "loss": 0.2115, + "step": 2256 + }, + { + "epoch": 0.6511669804437318, + "grad_norm": 2.3648903369903564, + "learning_rate": 3.6722306525037936e-06, + "loss": 0.2156, + "step": 2259 + }, + { + "epoch": 0.6520317440299784, + "grad_norm": 2.9157233238220215, + "learning_rate": 3.6631259484066773e-06, + "loss": 0.2542, + "step": 2262 + }, + { + "epoch": 0.6528965076162251, + "grad_norm": 2.67684268951416, + "learning_rate": 3.65402124430956e-06, + "loss": 0.2434, + "step": 2265 + }, + { + "epoch": 0.6537612712024717, + "grad_norm": 1.8347716331481934, + "learning_rate": 3.644916540212443e-06, + "loss": 0.2236, + "step": 2268 + }, + { + "epoch": 0.6546260347887184, + "grad_norm": 2.5923428535461426, + "learning_rate": 3.6358118361153266e-06, + "loss": 0.2232, + "step": 2271 + }, + { + "epoch": 0.655490798374965, + "grad_norm": 1.9584388732910156, + "learning_rate": 3.6267071320182095e-06, + "loss": 0.1951, + "step": 2274 + }, + { + "epoch": 0.6563555619612117, + "grad_norm": 1.8902173042297363, + "learning_rate": 3.617602427921093e-06, + "loss": 0.2051, + "step": 2277 + }, + { + "epoch": 0.6572203255474584, + "grad_norm": 1.735851764678955, + "learning_rate": 3.608497723823976e-06, + "loss": 0.22, + "step": 2280 + }, + { + "epoch": 0.6580850891337051, + "grad_norm": 3.9721834659576416, + "learning_rate": 3.599393019726859e-06, + "loss": 0.2518, + "step": 2283 + }, + { + "epoch": 0.6589498527199518, + "grad_norm": 3.3028452396392822, + "learning_rate": 3.5902883156297426e-06, + "loss": 0.2379, + "step": 2286 + }, + { + "epoch": 0.6598146163061984, + "grad_norm": 3.636674404144287, + "learning_rate": 3.5811836115326254e-06, + "loss": 0.2736, + "step": 2289 + }, + { + "epoch": 0.6606793798924451, + "grad_norm": 3.664700984954834, + "learning_rate": 3.572078907435509e-06, + "loss": 0.2409, + "step": 2292 + }, + { + "epoch": 0.6615441434786917, + "grad_norm": 1.9176503419876099, + "learning_rate": 3.562974203338392e-06, + "loss": 0.1983, + "step": 2295 + }, + { + "epoch": 0.6624089070649384, + "grad_norm": 1.9897313117980957, + "learning_rate": 3.5538694992412748e-06, + "loss": 0.2434, + "step": 2298 + }, + { + "epoch": 0.6629854161224361, + "eval_loss": 0.2195964902639389, + "eval_mse": 0.21959649334591813, + "eval_runtime": 64.3698, + "eval_samples_per_second": 15.535, + "eval_steps_per_second": 3.884, + "step": 2300 + }, + { + "epoch": 0.663273670651185, + "grad_norm": 2.568549871444702, + "learning_rate": 3.5447647951441585e-06, + "loss": 0.2006, + "step": 2301 + }, + { + "epoch": 0.6641384342374317, + "grad_norm": 2.855433940887451, + "learning_rate": 3.5356600910470413e-06, + "loss": 0.2064, + "step": 2304 + }, + { + "epoch": 0.6650031978236783, + "grad_norm": 3.653754949569702, + "learning_rate": 3.526555386949924e-06, + "loss": 0.2349, + "step": 2307 + }, + { + "epoch": 0.665867961409925, + "grad_norm": 2.384378671646118, + "learning_rate": 3.517450682852808e-06, + "loss": 0.2334, + "step": 2310 + }, + { + "epoch": 0.6667327249961716, + "grad_norm": 1.966973900794983, + "learning_rate": 3.5083459787556907e-06, + "loss": 0.2148, + "step": 2313 + }, + { + "epoch": 0.6675974885824183, + "grad_norm": 2.5276942253112793, + "learning_rate": 3.499241274658574e-06, + "loss": 0.2302, + "step": 2316 + }, + { + "epoch": 0.668462252168665, + "grad_norm": 3.5509471893310547, + "learning_rate": 3.490136570561457e-06, + "loss": 0.2672, + "step": 2319 + }, + { + "epoch": 0.6693270157549116, + "grad_norm": 6.874312877655029, + "learning_rate": 3.48103186646434e-06, + "loss": 0.2538, + "step": 2322 + }, + { + "epoch": 0.6701917793411583, + "grad_norm": 2.7544713020324707, + "learning_rate": 3.4719271623672233e-06, + "loss": 0.2146, + "step": 2325 + }, + { + "epoch": 0.6710565429274049, + "grad_norm": 2.2370007038116455, + "learning_rate": 3.4628224582701066e-06, + "loss": 0.2373, + "step": 2328 + }, + { + "epoch": 0.6719213065136516, + "grad_norm": 2.3013980388641357, + "learning_rate": 3.45371775417299e-06, + "loss": 0.2437, + "step": 2331 + }, + { + "epoch": 0.6727860700998982, + "grad_norm": 2.2957603931427, + "learning_rate": 3.4446130500758727e-06, + "loss": 0.1974, + "step": 2334 + }, + { + "epoch": 0.6736508336861449, + "grad_norm": 2.216355562210083, + "learning_rate": 3.435508345978756e-06, + "loss": 0.2274, + "step": 2337 + }, + { + "epoch": 0.6745155972723915, + "grad_norm": 2.5096681118011475, + "learning_rate": 3.4264036418816392e-06, + "loss": 0.2423, + "step": 2340 + }, + { + "epoch": 0.6753803608586382, + "grad_norm": 2.094209671020508, + "learning_rate": 3.417298937784522e-06, + "loss": 0.2296, + "step": 2343 + }, + { + "epoch": 0.6762451244448848, + "grad_norm": 2.772343635559082, + "learning_rate": 3.4081942336874053e-06, + "loss": 0.241, + "step": 2346 + }, + { + "epoch": 0.6771098880311315, + "grad_norm": 3.8428046703338623, + "learning_rate": 3.3990895295902886e-06, + "loss": 0.2055, + "step": 2349 + }, + { + "epoch": 0.6779746516173781, + "grad_norm": 1.7676074504852295, + "learning_rate": 3.3899848254931714e-06, + "loss": 0.1845, + "step": 2352 + }, + { + "epoch": 0.6788394152036248, + "grad_norm": 3.0652945041656494, + "learning_rate": 3.380880121396055e-06, + "loss": 0.2376, + "step": 2355 + }, + { + "epoch": 0.6797041787898714, + "grad_norm": 3.206815242767334, + "learning_rate": 3.371775417298938e-06, + "loss": 0.2543, + "step": 2358 + }, + { + "epoch": 0.6805689423761181, + "grad_norm": 3.2649717330932617, + "learning_rate": 3.3626707132018212e-06, + "loss": 0.2255, + "step": 2361 + }, + { + "epoch": 0.6814337059623647, + "grad_norm": 2.2816710472106934, + "learning_rate": 3.3535660091047045e-06, + "loss": 0.2318, + "step": 2364 + }, + { + "epoch": 0.6822984695486114, + "grad_norm": 2.291609287261963, + "learning_rate": 3.3444613050075873e-06, + "loss": 0.235, + "step": 2367 + }, + { + "epoch": 0.683163233134858, + "grad_norm": 2.037982940673828, + "learning_rate": 3.3353566009104706e-06, + "loss": 0.2437, + "step": 2370 + }, + { + "epoch": 0.6840279967211047, + "grad_norm": 2.614445209503174, + "learning_rate": 3.326251896813354e-06, + "loss": 0.2452, + "step": 2373 + }, + { + "epoch": 0.6848927603073514, + "grad_norm": 1.8799285888671875, + "learning_rate": 3.3171471927162367e-06, + "loss": 0.2567, + "step": 2376 + }, + { + "epoch": 0.685757523893598, + "grad_norm": 3.457829236984253, + "learning_rate": 3.3080424886191204e-06, + "loss": 0.2307, + "step": 2379 + }, + { + "epoch": 0.6866222874798447, + "grad_norm": 3.6573526859283447, + "learning_rate": 3.2989377845220033e-06, + "loss": 0.2321, + "step": 2382 + }, + { + "epoch": 0.6874870510660913, + "grad_norm": 2.0287060737609863, + "learning_rate": 3.289833080424886e-06, + "loss": 0.246, + "step": 2385 + }, + { + "epoch": 0.688351814652338, + "grad_norm": 2.5227603912353516, + "learning_rate": 3.2807283763277698e-06, + "loss": 0.1936, + "step": 2388 + }, + { + "epoch": 0.6892165782385846, + "grad_norm": 1.6194448471069336, + "learning_rate": 3.2716236722306526e-06, + "loss": 0.2108, + "step": 2391 + }, + { + "epoch": 0.6900813418248313, + "grad_norm": 3.9881162643432617, + "learning_rate": 3.2625189681335363e-06, + "loss": 0.2384, + "step": 2394 + }, + { + "epoch": 0.6909461054110779, + "grad_norm": 2.262324094772339, + "learning_rate": 3.253414264036419e-06, + "loss": 0.2791, + "step": 2397 + }, + { + "epoch": 0.6918108689973247, + "grad_norm": 3.344790458679199, + "learning_rate": 3.244309559939302e-06, + "loss": 0.2464, + "step": 2400 + }, + { + "epoch": 0.6918108689973247, + "eval_loss": 0.21846170723438263, + "eval_mse": 0.21846168629289606, + "eval_runtime": 64.3607, + "eval_samples_per_second": 15.537, + "eval_steps_per_second": 3.884, + "step": 2400 + }, + { + "epoch": 0.6926756325835713, + "grad_norm": 3.4839906692504883, + "learning_rate": 3.2352048558421857e-06, + "loss": 0.2318, + "step": 2403 + }, + { + "epoch": 0.693540396169818, + "grad_norm": 1.9182045459747314, + "learning_rate": 3.2261001517450685e-06, + "loss": 0.2366, + "step": 2406 + }, + { + "epoch": 0.6944051597560646, + "grad_norm": 3.907069683074951, + "learning_rate": 3.2169954476479514e-06, + "loss": 0.2068, + "step": 2409 + }, + { + "epoch": 0.6952699233423113, + "grad_norm": 2.516395330429077, + "learning_rate": 3.207890743550835e-06, + "loss": 0.2467, + "step": 2412 + }, + { + "epoch": 0.696134686928558, + "grad_norm": 8.096402168273926, + "learning_rate": 3.198786039453718e-06, + "loss": 0.205, + "step": 2415 + }, + { + "epoch": 0.6969994505148046, + "grad_norm": 2.871696949005127, + "learning_rate": 3.1896813353566016e-06, + "loss": 0.2259, + "step": 2418 + }, + { + "epoch": 0.6978642141010513, + "grad_norm": 2.6868882179260254, + "learning_rate": 3.1805766312594844e-06, + "loss": 0.2389, + "step": 2421 + }, + { + "epoch": 0.6987289776872979, + "grad_norm": 2.687101364135742, + "learning_rate": 3.1714719271623673e-06, + "loss": 0.2148, + "step": 2424 + }, + { + "epoch": 0.6995937412735446, + "grad_norm": 10.40469741821289, + "learning_rate": 3.162367223065251e-06, + "loss": 0.2412, + "step": 2427 + }, + { + "epoch": 0.7004585048597912, + "grad_norm": 2.589078903198242, + "learning_rate": 3.153262518968134e-06, + "loss": 0.2411, + "step": 2430 + }, + { + "epoch": 0.7013232684460379, + "grad_norm": 2.443969488143921, + "learning_rate": 3.1441578148710167e-06, + "loss": 0.2378, + "step": 2433 + }, + { + "epoch": 0.7021880320322845, + "grad_norm": 3.190467596054077, + "learning_rate": 3.1350531107739003e-06, + "loss": 0.2148, + "step": 2436 + }, + { + "epoch": 0.7030527956185312, + "grad_norm": 6.994554042816162, + "learning_rate": 3.125948406676783e-06, + "loss": 0.2349, + "step": 2439 + }, + { + "epoch": 0.7039175592047778, + "grad_norm": 4.769686698913574, + "learning_rate": 3.1168437025796665e-06, + "loss": 0.225, + "step": 2442 + }, + { + "epoch": 0.7047823227910245, + "grad_norm": 1.8445836305618286, + "learning_rate": 3.1077389984825497e-06, + "loss": 0.2498, + "step": 2445 + }, + { + "epoch": 0.7056470863772711, + "grad_norm": 2.2382473945617676, + "learning_rate": 3.0986342943854326e-06, + "loss": 0.21, + "step": 2448 + }, + { + "epoch": 0.7065118499635178, + "grad_norm": 2.9342122077941895, + "learning_rate": 3.089529590288316e-06, + "loss": 0.2316, + "step": 2451 + }, + { + "epoch": 0.7073766135497644, + "grad_norm": 2.4339959621429443, + "learning_rate": 3.080424886191199e-06, + "loss": 0.2446, + "step": 2454 + }, + { + "epoch": 0.7082413771360111, + "grad_norm": 3.4156157970428467, + "learning_rate": 3.0713201820940824e-06, + "loss": 0.2465, + "step": 2457 + }, + { + "epoch": 0.7091061407222577, + "grad_norm": 2.0606541633605957, + "learning_rate": 3.062215477996965e-06, + "loss": 0.2538, + "step": 2460 + }, + { + "epoch": 0.7099709043085044, + "grad_norm": 2.633622407913208, + "learning_rate": 3.0531107738998485e-06, + "loss": 0.2176, + "step": 2463 + }, + { + "epoch": 0.710835667894751, + "grad_norm": 5.920553207397461, + "learning_rate": 3.0440060698027317e-06, + "loss": 0.2238, + "step": 2466 + }, + { + "epoch": 0.7117004314809977, + "grad_norm": 1.6151419878005981, + "learning_rate": 3.0349013657056146e-06, + "loss": 0.1892, + "step": 2469 + }, + { + "epoch": 0.7125651950672444, + "grad_norm": 2.7643418312072754, + "learning_rate": 3.025796661608498e-06, + "loss": 0.2222, + "step": 2472 + }, + { + "epoch": 0.713429958653491, + "grad_norm": 2.580146551132202, + "learning_rate": 3.016691957511381e-06, + "loss": 0.261, + "step": 2475 + }, + { + "epoch": 0.7142947222397377, + "grad_norm": 1.7853143215179443, + "learning_rate": 3.0075872534142644e-06, + "loss": 0.2295, + "step": 2478 + }, + { + "epoch": 0.7151594858259843, + "grad_norm": 1.8063633441925049, + "learning_rate": 2.9984825493171476e-06, + "loss": 0.2186, + "step": 2481 + }, + { + "epoch": 0.716024249412231, + "grad_norm": 5.130537986755371, + "learning_rate": 2.9893778452200305e-06, + "loss": 0.242, + "step": 2484 + }, + { + "epoch": 0.7168890129984776, + "grad_norm": 2.46856951713562, + "learning_rate": 2.9802731411229137e-06, + "loss": 0.2491, + "step": 2487 + }, + { + "epoch": 0.7177537765847243, + "grad_norm": 3.249279260635376, + "learning_rate": 2.971168437025797e-06, + "loss": 0.2291, + "step": 2490 + }, + { + "epoch": 0.7186185401709709, + "grad_norm": 2.0126419067382812, + "learning_rate": 2.96206373292868e-06, + "loss": 0.2038, + "step": 2493 + }, + { + "epoch": 0.7194833037572176, + "grad_norm": 11.745406150817871, + "learning_rate": 2.9529590288315635e-06, + "loss": 0.2555, + "step": 2496 + }, + { + "epoch": 0.7203480673434642, + "grad_norm": 1.5893986225128174, + "learning_rate": 2.9438543247344464e-06, + "loss": 0.2338, + "step": 2499 + }, + { + "epoch": 0.7206363218722132, + "eval_loss": 0.21656151115894318, + "eval_mse": 0.21656151949986815, + "eval_runtime": 64.4174, + "eval_samples_per_second": 15.524, + "eval_steps_per_second": 3.881, + "step": 2500 + }, + { + "epoch": 0.7212128309297109, + "grad_norm": 2.0566763877868652, + "learning_rate": 2.9347496206373292e-06, + "loss": 0.2395, + "step": 2502 + }, + { + "epoch": 0.7220775945159575, + "grad_norm": 2.7235090732574463, + "learning_rate": 2.925644916540213e-06, + "loss": 0.2057, + "step": 2505 + }, + { + "epoch": 0.7229423581022042, + "grad_norm": 4.992210388183594, + "learning_rate": 2.9165402124430958e-06, + "loss": 0.2557, + "step": 2508 + }, + { + "epoch": 0.7238071216884508, + "grad_norm": 2.544344186782837, + "learning_rate": 2.9074355083459786e-06, + "loss": 0.2299, + "step": 2511 + }, + { + "epoch": 0.7246718852746975, + "grad_norm": 3.472134590148926, + "learning_rate": 2.8983308042488623e-06, + "loss": 0.2813, + "step": 2514 + }, + { + "epoch": 0.7255366488609443, + "grad_norm": 1.9130823612213135, + "learning_rate": 2.889226100151745e-06, + "loss": 0.2167, + "step": 2517 + }, + { + "epoch": 0.7264014124471909, + "grad_norm": 1.9955520629882812, + "learning_rate": 2.880121396054629e-06, + "loss": 0.2372, + "step": 2520 + }, + { + "epoch": 0.7272661760334376, + "grad_norm": 2.574761152267456, + "learning_rate": 2.8710166919575117e-06, + "loss": 0.2256, + "step": 2523 + }, + { + "epoch": 0.7281309396196842, + "grad_norm": 1.6945056915283203, + "learning_rate": 2.8619119878603945e-06, + "loss": 0.2177, + "step": 2526 + }, + { + "epoch": 0.7289957032059309, + "grad_norm": 2.1304931640625, + "learning_rate": 2.852807283763278e-06, + "loss": 0.2438, + "step": 2529 + }, + { + "epoch": 0.7298604667921775, + "grad_norm": 1.626629114151001, + "learning_rate": 2.843702579666161e-06, + "loss": 0.1954, + "step": 2532 + }, + { + "epoch": 0.7307252303784242, + "grad_norm": 2.735403537750244, + "learning_rate": 2.834597875569044e-06, + "loss": 0.2274, + "step": 2535 + }, + { + "epoch": 0.7315899939646708, + "grad_norm": 2.8743979930877686, + "learning_rate": 2.8254931714719276e-06, + "loss": 0.235, + "step": 2538 + }, + { + "epoch": 0.7324547575509175, + "grad_norm": 1.9181838035583496, + "learning_rate": 2.8163884673748104e-06, + "loss": 0.217, + "step": 2541 + }, + { + "epoch": 0.7333195211371641, + "grad_norm": 1.9460699558258057, + "learning_rate": 2.807283763277694e-06, + "loss": 0.1983, + "step": 2544 + }, + { + "epoch": 0.7341842847234108, + "grad_norm": 2.0447700023651123, + "learning_rate": 2.798179059180577e-06, + "loss": 0.2168, + "step": 2547 + }, + { + "epoch": 0.7350490483096574, + "grad_norm": 2.380274534225464, + "learning_rate": 2.78907435508346e-06, + "loss": 0.2275, + "step": 2550 + }, + { + "epoch": 0.7359138118959041, + "grad_norm": 1.5381513833999634, + "learning_rate": 2.7799696509863435e-06, + "loss": 0.1808, + "step": 2553 + }, + { + "epoch": 0.7367785754821508, + "grad_norm": 1.8658198118209839, + "learning_rate": 2.7708649468892263e-06, + "loss": 0.2094, + "step": 2556 + }, + { + "epoch": 0.7376433390683974, + "grad_norm": 1.8906834125518799, + "learning_rate": 2.7617602427921096e-06, + "loss": 0.2315, + "step": 2559 + }, + { + "epoch": 0.7385081026546441, + "grad_norm": 2.179494619369507, + "learning_rate": 2.752655538694993e-06, + "loss": 0.2317, + "step": 2562 + }, + { + "epoch": 0.7393728662408907, + "grad_norm": 1.9197441339492798, + "learning_rate": 2.7435508345978757e-06, + "loss": 0.2023, + "step": 2565 + }, + { + "epoch": 0.7402376298271374, + "grad_norm": 3.4614813327789307, + "learning_rate": 2.734446130500759e-06, + "loss": 0.2112, + "step": 2568 + }, + { + "epoch": 0.741102393413384, + "grad_norm": 4.101759433746338, + "learning_rate": 2.7253414264036422e-06, + "loss": 0.2087, + "step": 2571 + }, + { + "epoch": 0.7419671569996307, + "grad_norm": 2.005354166030884, + "learning_rate": 2.716236722306525e-06, + "loss": 0.2689, + "step": 2574 + }, + { + "epoch": 0.7428319205858773, + "grad_norm": 1.8958799839019775, + "learning_rate": 2.7071320182094083e-06, + "loss": 0.2639, + "step": 2577 + }, + { + "epoch": 0.743696684172124, + "grad_norm": 2.3008432388305664, + "learning_rate": 2.6980273141122916e-06, + "loss": 0.2412, + "step": 2580 + }, + { + "epoch": 0.7445614477583706, + "grad_norm": 2.5842695236206055, + "learning_rate": 2.688922610015175e-06, + "loss": 0.2339, + "step": 2583 + }, + { + "epoch": 0.7454262113446173, + "grad_norm": 2.5189950466156006, + "learning_rate": 2.6798179059180577e-06, + "loss": 0.2413, + "step": 2586 + }, + { + "epoch": 0.7462909749308639, + "grad_norm": 1.8702943325042725, + "learning_rate": 2.670713201820941e-06, + "loss": 0.2242, + "step": 2589 + }, + { + "epoch": 0.7471557385171106, + "grad_norm": 1.5371664762496948, + "learning_rate": 2.6616084977238242e-06, + "loss": 0.2089, + "step": 2592 + }, + { + "epoch": 0.7480205021033572, + "grad_norm": 2.6837081909179688, + "learning_rate": 2.652503793626707e-06, + "loss": 0.214, + "step": 2595 + }, + { + "epoch": 0.7488852656896039, + "grad_norm": 2.3358068466186523, + "learning_rate": 2.6433990895295904e-06, + "loss": 0.243, + "step": 2598 + }, + { + "epoch": 0.7494617747471017, + "eval_loss": 0.2165103405714035, + "eval_mse": 0.2165103500261903, + "eval_runtime": 64.7115, + "eval_samples_per_second": 15.453, + "eval_steps_per_second": 3.863, + "step": 2600 + }, + { + "epoch": 0.7497500292758505, + "grad_norm": 4.396569728851318, + "learning_rate": 2.6342943854324736e-06, + "loss": 0.2704, + "step": 2601 + }, + { + "epoch": 0.7506147928620972, + "grad_norm": 3.1913275718688965, + "learning_rate": 2.625189681335357e-06, + "loss": 0.2682, + "step": 2604 + }, + { + "epoch": 0.7514795564483439, + "grad_norm": 2.679997444152832, + "learning_rate": 2.61608497723824e-06, + "loss": 0.1992, + "step": 2607 + }, + { + "epoch": 0.7523443200345905, + "grad_norm": 3.6164557933807373, + "learning_rate": 2.606980273141123e-06, + "loss": 0.2249, + "step": 2610 + }, + { + "epoch": 0.7532090836208372, + "grad_norm": 2.7497098445892334, + "learning_rate": 2.5978755690440063e-06, + "loss": 0.245, + "step": 2613 + }, + { + "epoch": 0.7540738472070838, + "grad_norm": 2.927504777908325, + "learning_rate": 2.5887708649468895e-06, + "loss": 0.2114, + "step": 2616 + }, + { + "epoch": 0.7549386107933305, + "grad_norm": 1.5433193445205688, + "learning_rate": 2.5796661608497724e-06, + "loss": 0.2014, + "step": 2619 + }, + { + "epoch": 0.7558033743795771, + "grad_norm": 2.3192660808563232, + "learning_rate": 2.570561456752656e-06, + "loss": 0.2901, + "step": 2622 + }, + { + "epoch": 0.7566681379658238, + "grad_norm": 1.924660086631775, + "learning_rate": 2.561456752655539e-06, + "loss": 0.2085, + "step": 2625 + }, + { + "epoch": 0.7575329015520704, + "grad_norm": 3.05586576461792, + "learning_rate": 2.5523520485584217e-06, + "loss": 0.2056, + "step": 2628 + }, + { + "epoch": 0.7583976651383171, + "grad_norm": 3.1275217533111572, + "learning_rate": 2.5432473444613054e-06, + "loss": 0.2282, + "step": 2631 + }, + { + "epoch": 0.7592624287245638, + "grad_norm": 2.8783745765686035, + "learning_rate": 2.5341426403641883e-06, + "loss": 0.2216, + "step": 2634 + }, + { + "epoch": 0.7601271923108105, + "grad_norm": 2.206467390060425, + "learning_rate": 2.525037936267071e-06, + "loss": 0.2383, + "step": 2637 + }, + { + "epoch": 0.7609919558970571, + "grad_norm": 1.7947461605072021, + "learning_rate": 2.515933232169955e-06, + "loss": 0.2242, + "step": 2640 + }, + { + "epoch": 0.7618567194833038, + "grad_norm": 2.2603344917297363, + "learning_rate": 2.5068285280728376e-06, + "loss": 0.2196, + "step": 2643 + }, + { + "epoch": 0.7627214830695505, + "grad_norm": 4.005883693695068, + "learning_rate": 2.497723823975721e-06, + "loss": 0.2683, + "step": 2646 + }, + { + "epoch": 0.7635862466557971, + "grad_norm": 1.8957757949829102, + "learning_rate": 2.488619119878604e-06, + "loss": 0.2437, + "step": 2649 + }, + { + "epoch": 0.7644510102420438, + "grad_norm": 3.7723991870880127, + "learning_rate": 2.4795144157814874e-06, + "loss": 0.2453, + "step": 2652 + }, + { + "epoch": 0.7653157738282904, + "grad_norm": 2.494797945022583, + "learning_rate": 2.4704097116843703e-06, + "loss": 0.2285, + "step": 2655 + }, + { + "epoch": 0.7661805374145371, + "grad_norm": 2.8127262592315674, + "learning_rate": 2.4613050075872536e-06, + "loss": 0.1973, + "step": 2658 + }, + { + "epoch": 0.7670453010007837, + "grad_norm": 1.8228410482406616, + "learning_rate": 2.452200303490137e-06, + "loss": 0.2155, + "step": 2661 + }, + { + "epoch": 0.7679100645870304, + "grad_norm": 5.944820404052734, + "learning_rate": 2.44309559939302e-06, + "loss": 0.2137, + "step": 2664 + }, + { + "epoch": 0.768774828173277, + "grad_norm": 1.749942421913147, + "learning_rate": 2.4339908952959034e-06, + "loss": 0.2134, + "step": 2667 + }, + { + "epoch": 0.7696395917595237, + "grad_norm": 2.2554376125335693, + "learning_rate": 2.424886191198786e-06, + "loss": 0.2397, + "step": 2670 + }, + { + "epoch": 0.7705043553457703, + "grad_norm": 2.416917085647583, + "learning_rate": 2.4157814871016695e-06, + "loss": 0.2201, + "step": 2673 + }, + { + "epoch": 0.771369118932017, + "grad_norm": 2.578611373901367, + "learning_rate": 2.4066767830045527e-06, + "loss": 0.2197, + "step": 2676 + }, + { + "epoch": 0.7722338825182636, + "grad_norm": 2.857250690460205, + "learning_rate": 2.397572078907436e-06, + "loss": 0.2395, + "step": 2679 + }, + { + "epoch": 0.7730986461045103, + "grad_norm": 2.608745574951172, + "learning_rate": 2.388467374810319e-06, + "loss": 0.2421, + "step": 2682 + }, + { + "epoch": 0.7739634096907569, + "grad_norm": 2.9792795181274414, + "learning_rate": 2.379362670713202e-06, + "loss": 0.2319, + "step": 2685 + }, + { + "epoch": 0.7748281732770036, + "grad_norm": 3.69161319732666, + "learning_rate": 2.3702579666160854e-06, + "loss": 0.2322, + "step": 2688 + }, + { + "epoch": 0.7756929368632502, + "grad_norm": 2.1806881427764893, + "learning_rate": 2.3611532625189686e-06, + "loss": 0.2392, + "step": 2691 + }, + { + "epoch": 0.7765577004494969, + "grad_norm": 4.177742004394531, + "learning_rate": 2.3520485584218515e-06, + "loss": 0.2221, + "step": 2694 + }, + { + "epoch": 0.7774224640357436, + "grad_norm": 3.6789214611053467, + "learning_rate": 2.3429438543247347e-06, + "loss": 0.2021, + "step": 2697 + }, + { + "epoch": 0.7782872276219902, + "grad_norm": 2.2298083305358887, + "learning_rate": 2.333839150227618e-06, + "loss": 0.1891, + "step": 2700 + }, + { + "epoch": 0.7782872276219902, + "eval_loss": 0.22008302807807922, + "eval_mse": 0.22008301681582815, + "eval_runtime": 64.7274, + "eval_samples_per_second": 15.449, + "eval_steps_per_second": 3.862, + "step": 2700 + }, + { + "epoch": 0.7791519912082369, + "grad_norm": 1.7022091150283813, + "learning_rate": 2.324734446130501e-06, + "loss": 0.2267, + "step": 2703 + }, + { + "epoch": 0.7800167547944835, + "grad_norm": 2.402782440185547, + "learning_rate": 2.315629742033384e-06, + "loss": 0.2112, + "step": 2706 + }, + { + "epoch": 0.7808815183807302, + "grad_norm": 12.773355484008789, + "learning_rate": 2.3065250379362674e-06, + "loss": 0.2078, + "step": 2709 + }, + { + "epoch": 0.7817462819669768, + "grad_norm": 2.3418192863464355, + "learning_rate": 2.2974203338391502e-06, + "loss": 0.2358, + "step": 2712 + }, + { + "epoch": 0.7826110455532235, + "grad_norm": 3.0130553245544434, + "learning_rate": 2.2883156297420335e-06, + "loss": 0.2576, + "step": 2715 + }, + { + "epoch": 0.7834758091394701, + "grad_norm": 4.195178031921387, + "learning_rate": 2.2792109256449168e-06, + "loss": 0.2335, + "step": 2718 + }, + { + "epoch": 0.7843405727257168, + "grad_norm": 1.9151793718338013, + "learning_rate": 2.2701062215477996e-06, + "loss": 0.2328, + "step": 2721 + }, + { + "epoch": 0.7852053363119634, + "grad_norm": 2.500737190246582, + "learning_rate": 2.261001517450683e-06, + "loss": 0.2166, + "step": 2724 + }, + { + "epoch": 0.7860700998982101, + "grad_norm": 2.213944911956787, + "learning_rate": 2.251896813353566e-06, + "loss": 0.2296, + "step": 2727 + }, + { + "epoch": 0.7869348634844567, + "grad_norm": 4.156373023986816, + "learning_rate": 2.2427921092564494e-06, + "loss": 0.213, + "step": 2730 + }, + { + "epoch": 0.7877996270707034, + "grad_norm": 1.7524887323379517, + "learning_rate": 2.2336874051593322e-06, + "loss": 0.1983, + "step": 2733 + }, + { + "epoch": 0.78866439065695, + "grad_norm": 2.1907401084899902, + "learning_rate": 2.2245827010622155e-06, + "loss": 0.2227, + "step": 2736 + }, + { + "epoch": 0.7895291542431967, + "grad_norm": 6.940857410430908, + "learning_rate": 2.2154779969650988e-06, + "loss": 0.2385, + "step": 2739 + }, + { + "epoch": 0.7903939178294433, + "grad_norm": 8.225829124450684, + "learning_rate": 2.206373292867982e-06, + "loss": 0.2084, + "step": 2742 + }, + { + "epoch": 0.79125868141569, + "grad_norm": 1.9167605638504028, + "learning_rate": 2.197268588770865e-06, + "loss": 0.2065, + "step": 2745 + }, + { + "epoch": 0.7921234450019367, + "grad_norm": 2.273833751678467, + "learning_rate": 2.188163884673748e-06, + "loss": 0.2379, + "step": 2748 + }, + { + "epoch": 0.7929882085881834, + "grad_norm": 2.6898539066314697, + "learning_rate": 2.1790591805766314e-06, + "loss": 0.2173, + "step": 2751 + }, + { + "epoch": 0.7938529721744301, + "grad_norm": 3.421860694885254, + "learning_rate": 2.1699544764795147e-06, + "loss": 0.2463, + "step": 2754 + }, + { + "epoch": 0.7947177357606767, + "grad_norm": 5.2192487716674805, + "learning_rate": 2.1608497723823975e-06, + "loss": 0.238, + "step": 2757 + }, + { + "epoch": 0.7955824993469234, + "grad_norm": 2.7427492141723633, + "learning_rate": 2.1517450682852808e-06, + "loss": 0.221, + "step": 2760 + }, + { + "epoch": 0.79644726293317, + "grad_norm": 2.2411558628082275, + "learning_rate": 2.142640364188164e-06, + "loss": 0.2266, + "step": 2763 + }, + { + "epoch": 0.7973120265194167, + "grad_norm": 3.7260963916778564, + "learning_rate": 2.1335356600910473e-06, + "loss": 0.2538, + "step": 2766 + }, + { + "epoch": 0.7981767901056633, + "grad_norm": 2.7217321395874023, + "learning_rate": 2.12443095599393e-06, + "loss": 0.2246, + "step": 2769 + }, + { + "epoch": 0.79904155369191, + "grad_norm": 2.651681900024414, + "learning_rate": 2.1153262518968134e-06, + "loss": 0.2334, + "step": 2772 + }, + { + "epoch": 0.7999063172781566, + "grad_norm": 1.9291447401046753, + "learning_rate": 2.1062215477996967e-06, + "loss": 0.2122, + "step": 2775 + }, + { + "epoch": 0.8007710808644033, + "grad_norm": 2.2406091690063477, + "learning_rate": 2.09711684370258e-06, + "loss": 0.219, + "step": 2778 + }, + { + "epoch": 0.80163584445065, + "grad_norm": 2.099543809890747, + "learning_rate": 2.0880121396054632e-06, + "loss": 0.2376, + "step": 2781 + }, + { + "epoch": 0.8025006080368966, + "grad_norm": 2.7685112953186035, + "learning_rate": 2.078907435508346e-06, + "loss": 0.2352, + "step": 2784 + }, + { + "epoch": 0.8033653716231433, + "grad_norm": 2.3922276496887207, + "learning_rate": 2.0698027314112293e-06, + "loss": 0.2482, + "step": 2787 + }, + { + "epoch": 0.8042301352093899, + "grad_norm": 2.0313024520874023, + "learning_rate": 2.0606980273141126e-06, + "loss": 0.2328, + "step": 2790 + }, + { + "epoch": 0.8050948987956366, + "grad_norm": 2.234933853149414, + "learning_rate": 2.051593323216996e-06, + "loss": 0.2315, + "step": 2793 + }, + { + "epoch": 0.8059596623818832, + "grad_norm": 2.251796245574951, + "learning_rate": 2.0424886191198787e-06, + "loss": 0.2669, + "step": 2796 + }, + { + "epoch": 0.8068244259681299, + "grad_norm": 2.3901267051696777, + "learning_rate": 2.033383915022762e-06, + "loss": 0.2355, + "step": 2799 + }, + { + "epoch": 0.8071126804968788, + "eval_loss": 0.2166597843170166, + "eval_mse": 0.2166597851237748, + "eval_runtime": 64.6977, + "eval_samples_per_second": 15.457, + "eval_steps_per_second": 3.864, + "step": 2800 + }, + { + "epoch": 0.8076891895543765, + "grad_norm": 2.007464647293091, + "learning_rate": 2.0242792109256452e-06, + "loss": 0.2071, + "step": 2802 + }, + { + "epoch": 0.8085539531406232, + "grad_norm": 3.2795073986053467, + "learning_rate": 2.0151745068285285e-06, + "loss": 0.2191, + "step": 2805 + }, + { + "epoch": 0.8094187167268698, + "grad_norm": 4.439976692199707, + "learning_rate": 2.0060698027314113e-06, + "loss": 0.2236, + "step": 2808 + }, + { + "epoch": 0.8102834803131165, + "grad_norm": 2.0333492755889893, + "learning_rate": 1.9969650986342946e-06, + "loss": 0.2395, + "step": 2811 + }, + { + "epoch": 0.8111482438993631, + "grad_norm": 3.058572292327881, + "learning_rate": 1.987860394537178e-06, + "loss": 0.2226, + "step": 2814 + }, + { + "epoch": 0.8120130074856098, + "grad_norm": 1.869295358657837, + "learning_rate": 1.978755690440061e-06, + "loss": 0.1924, + "step": 2817 + }, + { + "epoch": 0.8128777710718564, + "grad_norm": 2.6618704795837402, + "learning_rate": 1.969650986342944e-06, + "loss": 0.2401, + "step": 2820 + }, + { + "epoch": 0.8137425346581031, + "grad_norm": 2.647307872772217, + "learning_rate": 1.9605462822458273e-06, + "loss": 0.2416, + "step": 2823 + }, + { + "epoch": 0.8146072982443497, + "grad_norm": 2.7761855125427246, + "learning_rate": 1.9514415781487105e-06, + "loss": 0.2248, + "step": 2826 + }, + { + "epoch": 0.8154720618305964, + "grad_norm": 2.126979112625122, + "learning_rate": 1.9423368740515934e-06, + "loss": 0.2115, + "step": 2829 + }, + { + "epoch": 0.816336825416843, + "grad_norm": 1.8128594160079956, + "learning_rate": 1.9332321699544766e-06, + "loss": 0.2208, + "step": 2832 + }, + { + "epoch": 0.8172015890030897, + "grad_norm": 2.527730703353882, + "learning_rate": 1.92412746585736e-06, + "loss": 0.2531, + "step": 2835 + }, + { + "epoch": 0.8180663525893364, + "grad_norm": 2.3727030754089355, + "learning_rate": 1.9150227617602427e-06, + "loss": 0.205, + "step": 2838 + }, + { + "epoch": 0.818931116175583, + "grad_norm": 1.696937918663025, + "learning_rate": 1.9059180576631262e-06, + "loss": 0.2091, + "step": 2841 + }, + { + "epoch": 0.8197958797618297, + "grad_norm": 1.8432416915893555, + "learning_rate": 1.8968133535660093e-06, + "loss": 0.2378, + "step": 2844 + }, + { + "epoch": 0.8206606433480763, + "grad_norm": 3.7460954189300537, + "learning_rate": 1.8877086494688923e-06, + "loss": 0.2068, + "step": 2847 + }, + { + "epoch": 0.821525406934323, + "grad_norm": 3.467381000518799, + "learning_rate": 1.8786039453717756e-06, + "loss": 0.2033, + "step": 2850 + }, + { + "epoch": 0.8223901705205696, + "grad_norm": 1.7389558553695679, + "learning_rate": 1.8694992412746589e-06, + "loss": 0.2268, + "step": 2853 + }, + { + "epoch": 0.8232549341068163, + "grad_norm": 2.4290714263916016, + "learning_rate": 1.860394537177542e-06, + "loss": 0.2535, + "step": 2856 + }, + { + "epoch": 0.8241196976930629, + "grad_norm": 1.9990437030792236, + "learning_rate": 1.851289833080425e-06, + "loss": 0.2351, + "step": 2859 + }, + { + "epoch": 0.8249844612793096, + "grad_norm": 2.0908126831054688, + "learning_rate": 1.8421851289833082e-06, + "loss": 0.2604, + "step": 2862 + }, + { + "epoch": 0.8258492248655563, + "grad_norm": 1.664847731590271, + "learning_rate": 1.8330804248861913e-06, + "loss": 0.2202, + "step": 2865 + }, + { + "epoch": 0.826713988451803, + "grad_norm": 2.0932564735412598, + "learning_rate": 1.8239757207890745e-06, + "loss": 0.2398, + "step": 2868 + }, + { + "epoch": 0.8275787520380496, + "grad_norm": 1.6450400352478027, + "learning_rate": 1.8148710166919576e-06, + "loss": 0.2132, + "step": 2871 + }, + { + "epoch": 0.8284435156242963, + "grad_norm": 3.0486483573913574, + "learning_rate": 1.8057663125948407e-06, + "loss": 0.2488, + "step": 2874 + }, + { + "epoch": 0.829308279210543, + "grad_norm": 4.125293254852295, + "learning_rate": 1.796661608497724e-06, + "loss": 0.229, + "step": 2877 + }, + { + "epoch": 0.8301730427967896, + "grad_norm": 2.4615678787231445, + "learning_rate": 1.7875569044006072e-06, + "loss": 0.2302, + "step": 2880 + }, + { + "epoch": 0.8310378063830363, + "grad_norm": 2.3922412395477295, + "learning_rate": 1.7784522003034905e-06, + "loss": 0.2023, + "step": 2883 + }, + { + "epoch": 0.8319025699692829, + "grad_norm": 2.2707550525665283, + "learning_rate": 1.7693474962063733e-06, + "loss": 0.2125, + "step": 2886 + }, + { + "epoch": 0.8327673335555296, + "grad_norm": 2.5821027755737305, + "learning_rate": 1.7602427921092566e-06, + "loss": 0.2022, + "step": 2889 + }, + { + "epoch": 0.8336320971417762, + "grad_norm": 1.5880707502365112, + "learning_rate": 1.7511380880121398e-06, + "loss": 0.2494, + "step": 2892 + }, + { + "epoch": 0.8344968607280229, + "grad_norm": 2.1693811416625977, + "learning_rate": 1.742033383915023e-06, + "loss": 0.2364, + "step": 2895 + }, + { + "epoch": 0.8353616243142695, + "grad_norm": 2.134542465209961, + "learning_rate": 1.732928679817906e-06, + "loss": 0.2231, + "step": 2898 + }, + { + "epoch": 0.8359381333717673, + "eval_loss": 0.2168218344449997, + "eval_mse": 0.21682184532732934, + "eval_runtime": 64.4646, + "eval_samples_per_second": 15.512, + "eval_steps_per_second": 3.878, + "step": 2900 + }, + { + "epoch": 0.8362263879005162, + "grad_norm": 2.96658992767334, + "learning_rate": 1.7238239757207892e-06, + "loss": 0.2326, + "step": 2901 + }, + { + "epoch": 0.8370911514867628, + "grad_norm": 2.925720453262329, + "learning_rate": 1.7147192716236725e-06, + "loss": 0.2265, + "step": 2904 + }, + { + "epoch": 0.8379559150730095, + "grad_norm": 2.380636692047119, + "learning_rate": 1.7056145675265557e-06, + "loss": 0.2272, + "step": 2907 + }, + { + "epoch": 0.8388206786592561, + "grad_norm": 1.5978946685791016, + "learning_rate": 1.6965098634294386e-06, + "loss": 0.2238, + "step": 2910 + }, + { + "epoch": 0.8396854422455028, + "grad_norm": 2.8584859371185303, + "learning_rate": 1.6874051593323218e-06, + "loss": 0.1995, + "step": 2913 + }, + { + "epoch": 0.8405502058317494, + "grad_norm": 2.5383777618408203, + "learning_rate": 1.6783004552352051e-06, + "loss": 0.247, + "step": 2916 + }, + { + "epoch": 0.8414149694179961, + "grad_norm": 3.073789119720459, + "learning_rate": 1.6691957511380882e-06, + "loss": 0.2323, + "step": 2919 + }, + { + "epoch": 0.8422797330042427, + "grad_norm": 3.0712456703186035, + "learning_rate": 1.6600910470409712e-06, + "loss": 0.2123, + "step": 2922 + }, + { + "epoch": 0.8431444965904894, + "grad_norm": 2.1077163219451904, + "learning_rate": 1.6509863429438545e-06, + "loss": 0.1908, + "step": 2925 + }, + { + "epoch": 0.844009260176736, + "grad_norm": 3.4034552574157715, + "learning_rate": 1.6418816388467375e-06, + "loss": 0.2366, + "step": 2928 + }, + { + "epoch": 0.8448740237629827, + "grad_norm": 2.9055449962615967, + "learning_rate": 1.6327769347496208e-06, + "loss": 0.2156, + "step": 2931 + }, + { + "epoch": 0.8457387873492294, + "grad_norm": 2.108920097351074, + "learning_rate": 1.6236722306525039e-06, + "loss": 0.2424, + "step": 2934 + }, + { + "epoch": 0.846603550935476, + "grad_norm": 2.038133144378662, + "learning_rate": 1.614567526555387e-06, + "loss": 0.2302, + "step": 2937 + }, + { + "epoch": 0.8474683145217227, + "grad_norm": 4.533400058746338, + "learning_rate": 1.6054628224582702e-06, + "loss": 0.1981, + "step": 2940 + }, + { + "epoch": 0.8483330781079693, + "grad_norm": 3.4339189529418945, + "learning_rate": 1.5963581183611534e-06, + "loss": 0.2072, + "step": 2943 + }, + { + "epoch": 0.849197841694216, + "grad_norm": 2.856205463409424, + "learning_rate": 1.5872534142640367e-06, + "loss": 0.2157, + "step": 2946 + }, + { + "epoch": 0.8500626052804626, + "grad_norm": 3.1973226070404053, + "learning_rate": 1.5781487101669196e-06, + "loss": 0.2361, + "step": 2949 + }, + { + "epoch": 0.8509273688667093, + "grad_norm": 2.0900986194610596, + "learning_rate": 1.5690440060698028e-06, + "loss": 0.1873, + "step": 2952 + }, + { + "epoch": 0.8517921324529559, + "grad_norm": 2.7864749431610107, + "learning_rate": 1.559939301972686e-06, + "loss": 0.2268, + "step": 2955 + }, + { + "epoch": 0.8526568960392026, + "grad_norm": 1.8518025875091553, + "learning_rate": 1.5508345978755694e-06, + "loss": 0.2399, + "step": 2958 + }, + { + "epoch": 0.8535216596254492, + "grad_norm": 1.9978440999984741, + "learning_rate": 1.5417298937784522e-06, + "loss": 0.2445, + "step": 2961 + }, + { + "epoch": 0.8543864232116959, + "grad_norm": 3.5526225566864014, + "learning_rate": 1.5326251896813355e-06, + "loss": 0.2132, + "step": 2964 + }, + { + "epoch": 0.8552511867979425, + "grad_norm": 2.19614315032959, + "learning_rate": 1.5235204855842187e-06, + "loss": 0.2329, + "step": 2967 + }, + { + "epoch": 0.8561159503841892, + "grad_norm": 2.024761438369751, + "learning_rate": 1.514415781487102e-06, + "loss": 0.2351, + "step": 2970 + }, + { + "epoch": 0.8569807139704358, + "grad_norm": 2.374737024307251, + "learning_rate": 1.5053110773899848e-06, + "loss": 0.2085, + "step": 2973 + }, + { + "epoch": 0.8578454775566825, + "grad_norm": 2.209737539291382, + "learning_rate": 1.496206373292868e-06, + "loss": 0.1754, + "step": 2976 + }, + { + "epoch": 0.8587102411429292, + "grad_norm": 3.4437668323516846, + "learning_rate": 1.4871016691957514e-06, + "loss": 0.2529, + "step": 2979 + }, + { + "epoch": 0.8595750047291759, + "grad_norm": 1.9130094051361084, + "learning_rate": 1.4779969650986344e-06, + "loss": 0.2431, + "step": 2982 + }, + { + "epoch": 0.8604397683154226, + "grad_norm": 1.7369916439056396, + "learning_rate": 1.4688922610015175e-06, + "loss": 0.2131, + "step": 2985 + }, + { + "epoch": 0.8613045319016692, + "grad_norm": 3.240520477294922, + "learning_rate": 1.4597875569044007e-06, + "loss": 0.2401, + "step": 2988 + }, + { + "epoch": 0.8621692954879159, + "grad_norm": 2.4426064491271973, + "learning_rate": 1.4506828528072838e-06, + "loss": 0.2089, + "step": 2991 + }, + { + "epoch": 0.8630340590741625, + "grad_norm": 1.9682718515396118, + "learning_rate": 1.441578148710167e-06, + "loss": 0.1858, + "step": 2994 + }, + { + "epoch": 0.8638988226604092, + "grad_norm": 1.8037463426589966, + "learning_rate": 1.4324734446130503e-06, + "loss": 0.2126, + "step": 2997 + }, + { + "epoch": 0.8647635862466558, + "grad_norm": 3.8557345867156982, + "learning_rate": 1.4233687405159332e-06, + "loss": 0.2274, + "step": 3000 + }, + { + "epoch": 0.8647635862466558, + "eval_loss": 0.22426578402519226, + "eval_mse": 0.22426578066963704, + "eval_runtime": 64.2143, + "eval_samples_per_second": 15.573, + "eval_steps_per_second": 3.893, + "step": 3000 + }, + { + "epoch": 0.8656283498329025, + "grad_norm": 3.097010850906372, + "learning_rate": 1.4142640364188164e-06, + "loss": 0.229, + "step": 3003 + }, + { + "epoch": 0.8664931134191491, + "grad_norm": 3.1431732177734375, + "learning_rate": 1.4051593323216997e-06, + "loss": 0.2437, + "step": 3006 + }, + { + "epoch": 0.8673578770053958, + "grad_norm": 1.8530339002609253, + "learning_rate": 1.396054628224583e-06, + "loss": 0.2201, + "step": 3009 + }, + { + "epoch": 0.8682226405916424, + "grad_norm": 2.807565927505493, + "learning_rate": 1.3869499241274658e-06, + "loss": 0.2237, + "step": 3012 + }, + { + "epoch": 0.8690874041778891, + "grad_norm": 3.1385457515716553, + "learning_rate": 1.377845220030349e-06, + "loss": 0.2645, + "step": 3015 + }, + { + "epoch": 0.8699521677641358, + "grad_norm": 4.684901714324951, + "learning_rate": 1.3687405159332323e-06, + "loss": 0.2562, + "step": 3018 + }, + { + "epoch": 0.8708169313503824, + "grad_norm": 2.9381299018859863, + "learning_rate": 1.3596358118361156e-06, + "loss": 0.2408, + "step": 3021 + }, + { + "epoch": 0.8716816949366291, + "grad_norm": 1.9034892320632935, + "learning_rate": 1.3505311077389985e-06, + "loss": 0.2087, + "step": 3024 + }, + { + "epoch": 0.8725464585228757, + "grad_norm": 1.968497395515442, + "learning_rate": 1.3414264036418817e-06, + "loss": 0.2028, + "step": 3027 + }, + { + "epoch": 0.8734112221091224, + "grad_norm": 2.3359177112579346, + "learning_rate": 1.332321699544765e-06, + "loss": 0.2305, + "step": 3030 + }, + { + "epoch": 0.874275985695369, + "grad_norm": 1.7949774265289307, + "learning_rate": 1.3232169954476482e-06, + "loss": 0.229, + "step": 3033 + }, + { + "epoch": 0.8751407492816157, + "grad_norm": 2.1725058555603027, + "learning_rate": 1.314112291350531e-06, + "loss": 0.2158, + "step": 3036 + }, + { + "epoch": 0.8760055128678623, + "grad_norm": 2.5111076831817627, + "learning_rate": 1.3050075872534144e-06, + "loss": 0.2032, + "step": 3039 + }, + { + "epoch": 0.876870276454109, + "grad_norm": 1.4585468769073486, + "learning_rate": 1.2959028831562976e-06, + "loss": 0.2064, + "step": 3042 + }, + { + "epoch": 0.8777350400403556, + "grad_norm": 2.4684135913848877, + "learning_rate": 1.2867981790591807e-06, + "loss": 0.2069, + "step": 3045 + }, + { + "epoch": 0.8785998036266023, + "grad_norm": 2.0519392490386963, + "learning_rate": 1.277693474962064e-06, + "loss": 0.2172, + "step": 3048 + }, + { + "epoch": 0.8794645672128489, + "grad_norm": 2.9718551635742188, + "learning_rate": 1.268588770864947e-06, + "loss": 0.2218, + "step": 3051 + }, + { + "epoch": 0.8803293307990956, + "grad_norm": 1.8732651472091675, + "learning_rate": 1.25948406676783e-06, + "loss": 0.2337, + "step": 3054 + }, + { + "epoch": 0.8811940943853422, + "grad_norm": 3.006303071975708, + "learning_rate": 1.2503793626707133e-06, + "loss": 0.2238, + "step": 3057 + }, + { + "epoch": 0.8820588579715889, + "grad_norm": 5.645898818969727, + "learning_rate": 1.2412746585735964e-06, + "loss": 0.2583, + "step": 3060 + }, + { + "epoch": 0.8829236215578355, + "grad_norm": 2.2179112434387207, + "learning_rate": 1.2321699544764796e-06, + "loss": 0.2308, + "step": 3063 + }, + { + "epoch": 0.8837883851440822, + "grad_norm": 3.6568005084991455, + "learning_rate": 1.2230652503793627e-06, + "loss": 0.2128, + "step": 3066 + }, + { + "epoch": 0.8846531487303289, + "grad_norm": 3.0393970012664795, + "learning_rate": 1.213960546282246e-06, + "loss": 0.2233, + "step": 3069 + }, + { + "epoch": 0.8855179123165755, + "grad_norm": 1.949182391166687, + "learning_rate": 1.204855842185129e-06, + "loss": 0.2409, + "step": 3072 + }, + { + "epoch": 0.8863826759028222, + "grad_norm": 1.6602507829666138, + "learning_rate": 1.1957511380880123e-06, + "loss": 0.2312, + "step": 3075 + }, + { + "epoch": 0.8872474394890688, + "grad_norm": 2.22861909866333, + "learning_rate": 1.1866464339908953e-06, + "loss": 0.2539, + "step": 3078 + }, + { + "epoch": 0.8881122030753155, + "grad_norm": 3.126702070236206, + "learning_rate": 1.1775417298937786e-06, + "loss": 0.2566, + "step": 3081 + }, + { + "epoch": 0.8889769666615621, + "grad_norm": 2.471876382827759, + "learning_rate": 1.1684370257966617e-06, + "loss": 0.2042, + "step": 3084 + }, + { + "epoch": 0.8898417302478088, + "grad_norm": 2.3772592544555664, + "learning_rate": 1.159332321699545e-06, + "loss": 0.2595, + "step": 3087 + }, + { + "epoch": 0.8907064938340554, + "grad_norm": 1.8966178894042969, + "learning_rate": 1.150227617602428e-06, + "loss": 0.2255, + "step": 3090 + }, + { + "epoch": 0.8915712574203021, + "grad_norm": 3.4512031078338623, + "learning_rate": 1.1411229135053112e-06, + "loss": 0.2417, + "step": 3093 + }, + { + "epoch": 0.8924360210065487, + "grad_norm": 9.493666648864746, + "learning_rate": 1.1320182094081943e-06, + "loss": 0.2244, + "step": 3096 + }, + { + "epoch": 0.8933007845927955, + "grad_norm": 2.2911179065704346, + "learning_rate": 1.1229135053110776e-06, + "loss": 0.2287, + "step": 3099 + }, + { + "epoch": 0.8935890391215443, + "eval_loss": 0.22029922902584076, + "eval_mse": 0.2202992255240679, + "eval_runtime": 64.4714, + "eval_samples_per_second": 15.511, + "eval_steps_per_second": 3.878, + "step": 3100 + }, + { + "epoch": 0.8941655481790421, + "grad_norm": 2.2213857173919678, + "learning_rate": 1.1138088012139606e-06, + "loss": 0.1838, + "step": 3102 + }, + { + "epoch": 0.8950303117652888, + "grad_norm": 1.7121168375015259, + "learning_rate": 1.1047040971168439e-06, + "loss": 0.2132, + "step": 3105 + }, + { + "epoch": 0.8958950753515355, + "grad_norm": 2.8772072792053223, + "learning_rate": 1.095599393019727e-06, + "loss": 0.223, + "step": 3108 + }, + { + "epoch": 0.8967598389377821, + "grad_norm": 1.947067141532898, + "learning_rate": 1.08649468892261e-06, + "loss": 0.2117, + "step": 3111 + }, + { + "epoch": 0.8976246025240288, + "grad_norm": 2.486713171005249, + "learning_rate": 1.0773899848254933e-06, + "loss": 0.2476, + "step": 3114 + }, + { + "epoch": 0.8984893661102754, + "grad_norm": 2.81208872795105, + "learning_rate": 1.0682852807283763e-06, + "loss": 0.2077, + "step": 3117 + }, + { + "epoch": 0.8993541296965221, + "grad_norm": 3.0056064128875732, + "learning_rate": 1.0591805766312596e-06, + "loss": 0.2419, + "step": 3120 + }, + { + "epoch": 0.9002188932827687, + "grad_norm": 1.8039710521697998, + "learning_rate": 1.0500758725341426e-06, + "loss": 0.2211, + "step": 3123 + }, + { + "epoch": 0.9010836568690154, + "grad_norm": 2.9514808654785156, + "learning_rate": 1.0409711684370259e-06, + "loss": 0.2669, + "step": 3126 + }, + { + "epoch": 0.901948420455262, + "grad_norm": 1.9579650163650513, + "learning_rate": 1.031866464339909e-06, + "loss": 0.2693, + "step": 3129 + }, + { + "epoch": 0.9028131840415087, + "grad_norm": 3.1238980293273926, + "learning_rate": 1.0227617602427922e-06, + "loss": 0.1991, + "step": 3132 + }, + { + "epoch": 0.9036779476277553, + "grad_norm": 4.342383861541748, + "learning_rate": 1.0136570561456753e-06, + "loss": 0.2167, + "step": 3135 + }, + { + "epoch": 0.904542711214002, + "grad_norm": 2.0083911418914795, + "learning_rate": 1.0045523520485585e-06, + "loss": 0.1924, + "step": 3138 + }, + { + "epoch": 0.9054074748002486, + "grad_norm": 2.090289354324341, + "learning_rate": 9.954476479514416e-07, + "loss": 0.1962, + "step": 3141 + }, + { + "epoch": 0.9062722383864953, + "grad_norm": 3.258606433868408, + "learning_rate": 9.863429438543249e-07, + "loss": 0.2439, + "step": 3144 + }, + { + "epoch": 0.9071370019727419, + "grad_norm": 2.009249210357666, + "learning_rate": 9.77238239757208e-07, + "loss": 0.2197, + "step": 3147 + }, + { + "epoch": 0.9080017655589886, + "grad_norm": 1.645767331123352, + "learning_rate": 9.681335356600912e-07, + "loss": 0.2283, + "step": 3150 + }, + { + "epoch": 0.9088665291452352, + "grad_norm": 2.6519992351531982, + "learning_rate": 9.590288315629742e-07, + "loss": 0.226, + "step": 3153 + }, + { + "epoch": 0.9097312927314819, + "grad_norm": 2.1228978633880615, + "learning_rate": 9.499241274658574e-07, + "loss": 0.2424, + "step": 3156 + }, + { + "epoch": 0.9105960563177286, + "grad_norm": 1.9438750743865967, + "learning_rate": 9.408194233687407e-07, + "loss": 0.2418, + "step": 3159 + }, + { + "epoch": 0.9114608199039752, + "grad_norm": 2.031235933303833, + "learning_rate": 9.317147192716237e-07, + "loss": 0.2006, + "step": 3162 + }, + { + "epoch": 0.9123255834902219, + "grad_norm": 2.0259816646575928, + "learning_rate": 9.22610015174507e-07, + "loss": 0.2174, + "step": 3165 + }, + { + "epoch": 0.9131903470764685, + "grad_norm": 2.0144436359405518, + "learning_rate": 9.1350531107739e-07, + "loss": 0.2115, + "step": 3168 + }, + { + "epoch": 0.9140551106627152, + "grad_norm": 2.4154086112976074, + "learning_rate": 9.044006069802733e-07, + "loss": 0.2443, + "step": 3171 + }, + { + "epoch": 0.9149198742489618, + "grad_norm": 2.0454399585723877, + "learning_rate": 8.952959028831563e-07, + "loss": 0.2032, + "step": 3174 + }, + { + "epoch": 0.9157846378352085, + "grad_norm": 3.3808867931365967, + "learning_rate": 8.861911987860396e-07, + "loss": 0.2256, + "step": 3177 + }, + { + "epoch": 0.9166494014214551, + "grad_norm": 4.769220352172852, + "learning_rate": 8.770864946889227e-07, + "loss": 0.2395, + "step": 3180 + }, + { + "epoch": 0.9175141650077018, + "grad_norm": 3.655897855758667, + "learning_rate": 8.679817905918058e-07, + "loss": 0.2491, + "step": 3183 + }, + { + "epoch": 0.9183789285939484, + "grad_norm": 3.9675004482269287, + "learning_rate": 8.58877086494689e-07, + "loss": 0.2262, + "step": 3186 + }, + { + "epoch": 0.9192436921801951, + "grad_norm": 3.2647206783294678, + "learning_rate": 8.497723823975721e-07, + "loss": 0.222, + "step": 3189 + }, + { + "epoch": 0.9201084557664417, + "grad_norm": 2.699422597885132, + "learning_rate": 8.406676783004553e-07, + "loss": 0.225, + "step": 3192 + }, + { + "epoch": 0.9209732193526884, + "grad_norm": 2.65602970123291, + "learning_rate": 8.315629742033385e-07, + "loss": 0.249, + "step": 3195 + }, + { + "epoch": 0.921837982938935, + "grad_norm": 2.1228225231170654, + "learning_rate": 8.224582701062215e-07, + "loss": 0.261, + "step": 3198 + }, + { + "epoch": 0.9224144919964329, + "eval_loss": 0.21858178079128265, + "eval_mse": 0.21858179174736142, + "eval_runtime": 64.5701, + "eval_samples_per_second": 15.487, + "eval_steps_per_second": 3.872, + "step": 3200 + }, + { + "epoch": 0.9227027465251817, + "grad_norm": 3.9595842361450195, + "learning_rate": 8.133535660091048e-07, + "loss": 0.2243, + "step": 3201 + }, + { + "epoch": 0.9235675101114283, + "grad_norm": 2.7755258083343506, + "learning_rate": 8.042488619119878e-07, + "loss": 0.2175, + "step": 3204 + }, + { + "epoch": 0.924432273697675, + "grad_norm": 2.876634120941162, + "learning_rate": 7.951441578148711e-07, + "loss": 0.2263, + "step": 3207 + }, + { + "epoch": 0.9252970372839217, + "grad_norm": 1.809976577758789, + "learning_rate": 7.860394537177542e-07, + "loss": 0.2161, + "step": 3210 + }, + { + "epoch": 0.9261618008701683, + "grad_norm": 2.5012404918670654, + "learning_rate": 7.769347496206374e-07, + "loss": 0.2412, + "step": 3213 + }, + { + "epoch": 0.9270265644564151, + "grad_norm": 2.09769868850708, + "learning_rate": 7.678300455235206e-07, + "loss": 0.2556, + "step": 3216 + }, + { + "epoch": 0.9278913280426617, + "grad_norm": 2.4902303218841553, + "learning_rate": 7.587253414264036e-07, + "loss": 0.2184, + "step": 3219 + }, + { + "epoch": 0.9287560916289084, + "grad_norm": 1.705115795135498, + "learning_rate": 7.496206373292869e-07, + "loss": 0.2104, + "step": 3222 + }, + { + "epoch": 0.929620855215155, + "grad_norm": 2.0965850353240967, + "learning_rate": 7.4051593323217e-07, + "loss": 0.2282, + "step": 3225 + }, + { + "epoch": 0.9304856188014017, + "grad_norm": 1.9065147638320923, + "learning_rate": 7.314112291350532e-07, + "loss": 0.2261, + "step": 3228 + }, + { + "epoch": 0.9313503823876483, + "grad_norm": 1.8584680557250977, + "learning_rate": 7.223065250379363e-07, + "loss": 0.1955, + "step": 3231 + }, + { + "epoch": 0.932215145973895, + "grad_norm": 2.8803727626800537, + "learning_rate": 7.132018209408196e-07, + "loss": 0.2133, + "step": 3234 + }, + { + "epoch": 0.9330799095601416, + "grad_norm": 2.6942360401153564, + "learning_rate": 7.040971168437026e-07, + "loss": 0.21, + "step": 3237 + }, + { + "epoch": 0.9339446731463883, + "grad_norm": 3.0036349296569824, + "learning_rate": 6.949924127465859e-07, + "loss": 0.2172, + "step": 3240 + }, + { + "epoch": 0.934809436732635, + "grad_norm": 3.479217290878296, + "learning_rate": 6.858877086494689e-07, + "loss": 0.222, + "step": 3243 + }, + { + "epoch": 0.9356742003188816, + "grad_norm": 8.07664680480957, + "learning_rate": 6.767830045523521e-07, + "loss": 0.2236, + "step": 3246 + }, + { + "epoch": 0.9365389639051283, + "grad_norm": 3.184757709503174, + "learning_rate": 6.676783004552352e-07, + "loss": 0.2471, + "step": 3249 + }, + { + "epoch": 0.9374037274913749, + "grad_norm": 1.7666175365447998, + "learning_rate": 6.585735963581184e-07, + "loss": 0.2145, + "step": 3252 + }, + { + "epoch": 0.9382684910776216, + "grad_norm": 1.856346607208252, + "learning_rate": 6.494688922610016e-07, + "loss": 0.2334, + "step": 3255 + }, + { + "epoch": 0.9391332546638682, + "grad_norm": 2.367562770843506, + "learning_rate": 6.403641881638847e-07, + "loss": 0.2301, + "step": 3258 + }, + { + "epoch": 0.9399980182501149, + "grad_norm": 2.3909690380096436, + "learning_rate": 6.312594840667678e-07, + "loss": 0.2305, + "step": 3261 + }, + { + "epoch": 0.9408627818363615, + "grad_norm": 2.0632777214050293, + "learning_rate": 6.22154779969651e-07, + "loss": 0.2083, + "step": 3264 + }, + { + "epoch": 0.9417275454226082, + "grad_norm": 2.854069471359253, + "learning_rate": 6.130500758725342e-07, + "loss": 0.2112, + "step": 3267 + }, + { + "epoch": 0.9425923090088548, + "grad_norm": 3.0743472576141357, + "learning_rate": 6.039453717754174e-07, + "loss": 0.225, + "step": 3270 + }, + { + "epoch": 0.9434570725951015, + "grad_norm": 1.8354507684707642, + "learning_rate": 5.948406676783005e-07, + "loss": 0.2067, + "step": 3273 + }, + { + "epoch": 0.9443218361813481, + "grad_norm": 7.717159748077393, + "learning_rate": 5.857359635811837e-07, + "loss": 0.2307, + "step": 3276 + }, + { + "epoch": 0.9451865997675948, + "grad_norm": 2.406531572341919, + "learning_rate": 5.766312594840668e-07, + "loss": 0.2245, + "step": 3279 + }, + { + "epoch": 0.9460513633538414, + "grad_norm": 2.0075788497924805, + "learning_rate": 5.675265553869499e-07, + "loss": 0.227, + "step": 3282 + }, + { + "epoch": 0.9469161269400881, + "grad_norm": 1.8937417268753052, + "learning_rate": 5.584218512898331e-07, + "loss": 0.214, + "step": 3285 + }, + { + "epoch": 0.9477808905263347, + "grad_norm": 1.6452524662017822, + "learning_rate": 5.493171471927162e-07, + "loss": 0.2088, + "step": 3288 + }, + { + "epoch": 0.9486456541125814, + "grad_norm": 1.942367672920227, + "learning_rate": 5.402124430955994e-07, + "loss": 0.2008, + "step": 3291 + }, + { + "epoch": 0.949510417698828, + "grad_norm": 1.6608052253723145, + "learning_rate": 5.311077389984825e-07, + "loss": 0.2131, + "step": 3294 + }, + { + "epoch": 0.9503751812850747, + "grad_norm": 1.681344747543335, + "learning_rate": 5.220030349013658e-07, + "loss": 0.2093, + "step": 3297 + }, + { + "epoch": 0.9512399448713214, + "grad_norm": 2.0465240478515625, + "learning_rate": 5.12898330804249e-07, + "loss": 0.2187, + "step": 3300 + }, + { + "epoch": 0.9512399448713214, + "eval_loss": 0.21758121252059937, + "eval_mse": 0.21758120648143814, + "eval_runtime": 64.738, + "eval_samples_per_second": 15.447, + "eval_steps_per_second": 3.862, + "step": 3300 + }, + { + "epoch": 0.952104708457568, + "grad_norm": 2.127420425415039, + "learning_rate": 5.037936267071321e-07, + "loss": 0.267, + "step": 3303 + }, + { + "epoch": 0.9529694720438147, + "grad_norm": 1.9527215957641602, + "learning_rate": 4.946889226100153e-07, + "loss": 0.2304, + "step": 3306 + }, + { + "epoch": 0.9538342356300613, + "grad_norm": 2.5131490230560303, + "learning_rate": 4.855842185128983e-07, + "loss": 0.1996, + "step": 3309 + }, + { + "epoch": 0.954698999216308, + "grad_norm": 1.555931568145752, + "learning_rate": 4.7647951441578155e-07, + "loss": 0.2054, + "step": 3312 + }, + { + "epoch": 0.9555637628025546, + "grad_norm": 2.3513290882110596, + "learning_rate": 4.673748103186647e-07, + "loss": 0.2015, + "step": 3315 + }, + { + "epoch": 0.9564285263888013, + "grad_norm": 4.119522571563721, + "learning_rate": 4.582701062215478e-07, + "loss": 0.1802, + "step": 3318 + }, + { + "epoch": 0.9572932899750479, + "grad_norm": 2.0488264560699463, + "learning_rate": 4.49165402124431e-07, + "loss": 0.218, + "step": 3321 + }, + { + "epoch": 0.9581580535612946, + "grad_norm": 4.189717769622803, + "learning_rate": 4.4006069802731414e-07, + "loss": 0.2439, + "step": 3324 + }, + { + "epoch": 0.9590228171475412, + "grad_norm": 3.0399768352508545, + "learning_rate": 4.309559939301973e-07, + "loss": 0.2606, + "step": 3327 + }, + { + "epoch": 0.9598875807337879, + "grad_norm": 3.3942065238952637, + "learning_rate": 4.2185128983308046e-07, + "loss": 0.2137, + "step": 3330 + }, + { + "epoch": 0.9607523443200346, + "grad_norm": 2.329697608947754, + "learning_rate": 4.127465857359636e-07, + "loss": 0.2363, + "step": 3333 + }, + { + "epoch": 0.9616171079062813, + "grad_norm": 1.8820949792861938, + "learning_rate": 4.0364188163884673e-07, + "loss": 0.2208, + "step": 3336 + }, + { + "epoch": 0.962481871492528, + "grad_norm": 2.777941942214966, + "learning_rate": 3.945371775417299e-07, + "loss": 0.2254, + "step": 3339 + }, + { + "epoch": 0.9633466350787746, + "grad_norm": 3.551192045211792, + "learning_rate": 3.8543247344461305e-07, + "loss": 0.2155, + "step": 3342 + }, + { + "epoch": 0.9642113986650213, + "grad_norm": 1.5239380598068237, + "learning_rate": 3.763277693474962e-07, + "loss": 0.2336, + "step": 3345 + }, + { + "epoch": 0.9650761622512679, + "grad_norm": 3.446213722229004, + "learning_rate": 3.6722306525037937e-07, + "loss": 0.2139, + "step": 3348 + }, + { + "epoch": 0.9659409258375146, + "grad_norm": 2.0579702854156494, + "learning_rate": 3.581183611532626e-07, + "loss": 0.2032, + "step": 3351 + }, + { + "epoch": 0.9668056894237612, + "grad_norm": 1.746421456336975, + "learning_rate": 3.4901365705614574e-07, + "loss": 0.1885, + "step": 3354 + }, + { + "epoch": 0.9676704530100079, + "grad_norm": 3.787752628326416, + "learning_rate": 3.399089529590289e-07, + "loss": 0.2145, + "step": 3357 + }, + { + "epoch": 0.9685352165962545, + "grad_norm": 2.8552401065826416, + "learning_rate": 3.3080424886191206e-07, + "loss": 0.1699, + "step": 3360 + }, + { + "epoch": 0.9693999801825012, + "grad_norm": 2.1320273876190186, + "learning_rate": 3.2169954476479517e-07, + "loss": 0.1813, + "step": 3363 + }, + { + "epoch": 0.9702647437687478, + "grad_norm": 2.03324031829834, + "learning_rate": 3.1259484066767833e-07, + "loss": 0.2193, + "step": 3366 + }, + { + "epoch": 0.9711295073549945, + "grad_norm": 1.8292734622955322, + "learning_rate": 3.034901365705615e-07, + "loss": 0.2111, + "step": 3369 + }, + { + "epoch": 0.9719942709412411, + "grad_norm": 2.5490074157714844, + "learning_rate": 2.9438543247344465e-07, + "loss": 0.218, + "step": 3372 + }, + { + "epoch": 0.9728590345274878, + "grad_norm": 3.7310287952423096, + "learning_rate": 2.852807283763278e-07, + "loss": 0.2132, + "step": 3375 + }, + { + "epoch": 0.9737237981137344, + "grad_norm": 1.762787938117981, + "learning_rate": 2.7617602427921097e-07, + "loss": 0.2054, + "step": 3378 + }, + { + "epoch": 0.9745885616999811, + "grad_norm": 2.4129533767700195, + "learning_rate": 2.670713201820941e-07, + "loss": 0.2089, + "step": 3381 + }, + { + "epoch": 0.9754533252862277, + "grad_norm": 2.115504026412964, + "learning_rate": 2.5796661608497724e-07, + "loss": 0.2308, + "step": 3384 + }, + { + "epoch": 0.9763180888724744, + "grad_norm": 1.7608245611190796, + "learning_rate": 2.488619119878604e-07, + "loss": 0.2233, + "step": 3387 + }, + { + "epoch": 0.977182852458721, + "grad_norm": 2.5845463275909424, + "learning_rate": 2.3975720789074356e-07, + "loss": 0.2285, + "step": 3390 + }, + { + "epoch": 0.9780476160449677, + "grad_norm": 10.719658851623535, + "learning_rate": 2.3065250379362674e-07, + "loss": 0.2317, + "step": 3393 + }, + { + "epoch": 0.9789123796312144, + "grad_norm": 1.8989415168762207, + "learning_rate": 2.215477996965099e-07, + "loss": 0.1977, + "step": 3396 + }, + { + "epoch": 0.979777143217461, + "grad_norm": 2.626805305480957, + "learning_rate": 2.1244309559939304e-07, + "loss": 0.2069, + "step": 3399 + }, + { + "epoch": 0.9800653977462099, + "eval_loss": 0.21775692701339722, + "eval_mse": 0.21775692150488976, + "eval_runtime": 64.3973, + "eval_samples_per_second": 15.529, + "eval_steps_per_second": 3.882, + "step": 3400 + }, + { + "epoch": 0.9806419068037077, + "grad_norm": 3.636981964111328, + "learning_rate": 2.033383915022762e-07, + "loss": 0.2272, + "step": 3402 + }, + { + "epoch": 0.9815066703899543, + "grad_norm": 2.3272805213928223, + "learning_rate": 1.9423368740515936e-07, + "loss": 0.215, + "step": 3405 + }, + { + "epoch": 0.982371433976201, + "grad_norm": 3.822941303253174, + "learning_rate": 1.851289833080425e-07, + "loss": 0.2589, + "step": 3408 + }, + { + "epoch": 0.9832361975624476, + "grad_norm": 1.864588975906372, + "learning_rate": 1.7602427921092565e-07, + "loss": 0.2135, + "step": 3411 + }, + { + "epoch": 0.9841009611486943, + "grad_norm": 2.8255016803741455, + "learning_rate": 1.669195751138088e-07, + "loss": 0.2505, + "step": 3414 + }, + { + "epoch": 0.9849657247349409, + "grad_norm": 2.431203603744507, + "learning_rate": 1.5781487101669194e-07, + "loss": 0.2892, + "step": 3417 + }, + { + "epoch": 0.9858304883211876, + "grad_norm": 4.612681865692139, + "learning_rate": 1.4871016691957513e-07, + "loss": 0.2465, + "step": 3420 + }, + { + "epoch": 0.9866952519074342, + "grad_norm": 2.1581344604492188, + "learning_rate": 1.3960546282245826e-07, + "loss": 0.2233, + "step": 3423 + }, + { + "epoch": 0.9875600154936809, + "grad_norm": 2.2907607555389404, + "learning_rate": 1.3050075872534145e-07, + "loss": 0.2177, + "step": 3426 + }, + { + "epoch": 0.9884247790799275, + "grad_norm": 1.9750571250915527, + "learning_rate": 1.2139605462822459e-07, + "loss": 0.1949, + "step": 3429 + }, + { + "epoch": 0.9892895426661742, + "grad_norm": 2.6737992763519287, + "learning_rate": 1.1229135053110775e-07, + "loss": 0.2263, + "step": 3432 + }, + { + "epoch": 0.9901543062524208, + "grad_norm": 1.981949806213379, + "learning_rate": 1.031866464339909e-07, + "loss": 0.2285, + "step": 3435 + }, + { + "epoch": 0.9910190698386675, + "grad_norm": 2.5654916763305664, + "learning_rate": 9.408194233687405e-08, + "loss": 0.2308, + "step": 3438 + }, + { + "epoch": 0.9918838334249142, + "grad_norm": 1.8888391256332397, + "learning_rate": 8.497723823975723e-08, + "loss": 0.2311, + "step": 3441 + }, + { + "epoch": 0.9927485970111608, + "grad_norm": 1.7521294355392456, + "learning_rate": 7.587253414264037e-08, + "loss": 0.1957, + "step": 3444 + }, + { + "epoch": 0.9936133605974075, + "grad_norm": 2.322394609451294, + "learning_rate": 6.676783004552352e-08, + "loss": 0.2198, + "step": 3447 + }, + { + "epoch": 0.9944781241836542, + "grad_norm": 1.9925882816314697, + "learning_rate": 5.7663125948406686e-08, + "loss": 0.2426, + "step": 3450 + }, + { + "epoch": 0.9953428877699009, + "grad_norm": 2.0008606910705566, + "learning_rate": 4.855842185128984e-08, + "loss": 0.2125, + "step": 3453 + }, + { + "epoch": 0.9962076513561475, + "grad_norm": 4.1277570724487305, + "learning_rate": 3.9453717754172986e-08, + "loss": 0.2444, + "step": 3456 + }, + { + "epoch": 0.9970724149423942, + "grad_norm": 2.3507277965545654, + "learning_rate": 3.0349013657056146e-08, + "loss": 0.2183, + "step": 3459 + }, + { + "epoch": 0.9979371785286408, + "grad_norm": 2.143399238586426, + "learning_rate": 2.1244309559939306e-08, + "loss": 0.2128, + "step": 3462 + }, + { + "epoch": 0.9988019421148875, + "grad_norm": 1.850144863128662, + "learning_rate": 1.213960546282246e-08, + "loss": 0.1883, + "step": 3465 + }, + { + "epoch": 0.9996667057011341, + "grad_norm": 1.8591262102127075, + "learning_rate": 3.034901365705615e-09, + "loss": 0.2093, + "step": 3468 + }, + { + "epoch": 0.999954960229883, + "step": 3469, + "total_flos": 9.410742865058857e+17, + "train_loss": 0.2716184546260883, + "train_runtime": 39469.9974, + "train_samples_per_second": 11.25, + "train_steps_per_second": 0.088 + } + ], + "logging_steps": 3, + "max_steps": 3469, + "num_input_tokens_seen": 0, + "num_train_epochs": 1, + "save_steps": 100, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 9.410742865058857e+17, + "train_batch_size": 4, + "trial_name": null, + "trial_params": null +}