| { |
| "best_global_step": null, |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 2.4681318681318682, |
| "eval_steps": 500, |
| "global_step": 375, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.006593406593406593, |
| "grad_norm": 1.195170283317566, |
| "learning_rate": 0.0, |
| "loss": 1.0821, |
| "step": 1 |
| }, |
| { |
| "epoch": 0.013186813186813187, |
| "grad_norm": 1.0331945419311523, |
| "learning_rate": 9.56e-05, |
| "loss": 1.0763, |
| "step": 2 |
| }, |
| { |
| "epoch": 0.01978021978021978, |
| "grad_norm": 1.1415084600448608, |
| "learning_rate": 0.0001912, |
| "loss": 1.0495, |
| "step": 3 |
| }, |
| { |
| "epoch": 0.026373626373626374, |
| "grad_norm": 1.063579797744751, |
| "learning_rate": 0.0002868, |
| "loss": 0.956, |
| "step": 4 |
| }, |
| { |
| "epoch": 0.03296703296703297, |
| "grad_norm": 0.6625626087188721, |
| "learning_rate": 0.0003824, |
| "loss": 0.8477, |
| "step": 5 |
| }, |
| { |
| "epoch": 0.03956043956043956, |
| "grad_norm": 0.6782878637313843, |
| "learning_rate": 0.000478, |
| "loss": 1.0343, |
| "step": 6 |
| }, |
| { |
| "epoch": 0.046153846153846156, |
| "grad_norm": 0.6161335110664368, |
| "learning_rate": 0.00047694013303769406, |
| "loss": 0.9727, |
| "step": 7 |
| }, |
| { |
| "epoch": 0.05274725274725275, |
| "grad_norm": 0.41984498500823975, |
| "learning_rate": 0.00047588026607538804, |
| "loss": 0.8698, |
| "step": 8 |
| }, |
| { |
| "epoch": 0.05934065934065934, |
| "grad_norm": 0.3116914927959442, |
| "learning_rate": 0.0004748203991130821, |
| "loss": 0.8446, |
| "step": 9 |
| }, |
| { |
| "epoch": 0.06593406593406594, |
| "grad_norm": 0.29025012254714966, |
| "learning_rate": 0.00047376053215077607, |
| "loss": 0.7877, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.07252747252747253, |
| "grad_norm": 0.39959046244621277, |
| "learning_rate": 0.00047270066518847006, |
| "loss": 0.8688, |
| "step": 11 |
| }, |
| { |
| "epoch": 0.07912087912087912, |
| "grad_norm": 0.5689778923988342, |
| "learning_rate": 0.0004716407982261641, |
| "loss": 0.8847, |
| "step": 12 |
| }, |
| { |
| "epoch": 0.08571428571428572, |
| "grad_norm": 0.40798476338386536, |
| "learning_rate": 0.00047058093126385814, |
| "loss": 0.7254, |
| "step": 13 |
| }, |
| { |
| "epoch": 0.09230769230769231, |
| "grad_norm": 0.6968982219696045, |
| "learning_rate": 0.0004695210643015521, |
| "loss": 0.8179, |
| "step": 14 |
| }, |
| { |
| "epoch": 0.0989010989010989, |
| "grad_norm": 0.5194700360298157, |
| "learning_rate": 0.00046846119733924617, |
| "loss": 0.7156, |
| "step": 15 |
| }, |
| { |
| "epoch": 0.1054945054945055, |
| "grad_norm": 0.43927809596061707, |
| "learning_rate": 0.00046740133037694015, |
| "loss": 0.8273, |
| "step": 16 |
| }, |
| { |
| "epoch": 0.11208791208791209, |
| "grad_norm": 0.3979283571243286, |
| "learning_rate": 0.00046634146341463414, |
| "loss": 0.7288, |
| "step": 17 |
| }, |
| { |
| "epoch": 0.11868131868131868, |
| "grad_norm": 0.4314859211444855, |
| "learning_rate": 0.0004652815964523282, |
| "loss": 0.7677, |
| "step": 18 |
| }, |
| { |
| "epoch": 0.12527472527472527, |
| "grad_norm": 0.3548160791397095, |
| "learning_rate": 0.0004642217294900222, |
| "loss": 0.6472, |
| "step": 19 |
| }, |
| { |
| "epoch": 0.13186813186813187, |
| "grad_norm": 0.38769328594207764, |
| "learning_rate": 0.0004631618625277162, |
| "loss": 0.6578, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.13846153846153847, |
| "grad_norm": 0.3017254173755646, |
| "learning_rate": 0.00046210199556541025, |
| "loss": 0.7163, |
| "step": 21 |
| }, |
| { |
| "epoch": 0.14505494505494507, |
| "grad_norm": 0.33315712213516235, |
| "learning_rate": 0.00046104212860310423, |
| "loss": 0.6951, |
| "step": 22 |
| }, |
| { |
| "epoch": 0.15164835164835164, |
| "grad_norm": 0.3334329426288605, |
| "learning_rate": 0.0004599822616407982, |
| "loss": 0.7464, |
| "step": 23 |
| }, |
| { |
| "epoch": 0.15824175824175823, |
| "grad_norm": 0.301981121301651, |
| "learning_rate": 0.00045892239467849226, |
| "loss": 0.667, |
| "step": 24 |
| }, |
| { |
| "epoch": 0.16483516483516483, |
| "grad_norm": 0.3768517076969147, |
| "learning_rate": 0.0004578625277161863, |
| "loss": 0.6719, |
| "step": 25 |
| }, |
| { |
| "epoch": 0.17142857142857143, |
| "grad_norm": 0.3810518980026245, |
| "learning_rate": 0.0004568026607538803, |
| "loss": 0.7687, |
| "step": 26 |
| }, |
| { |
| "epoch": 0.17802197802197803, |
| "grad_norm": 0.4689561128616333, |
| "learning_rate": 0.00045574279379157433, |
| "loss": 0.7114, |
| "step": 27 |
| }, |
| { |
| "epoch": 0.18461538461538463, |
| "grad_norm": 0.3895481824874878, |
| "learning_rate": 0.00045468292682926826, |
| "loss": 0.7328, |
| "step": 28 |
| }, |
| { |
| "epoch": 0.1912087912087912, |
| "grad_norm": 0.44972071051597595, |
| "learning_rate": 0.0004536230598669623, |
| "loss": 0.708, |
| "step": 29 |
| }, |
| { |
| "epoch": 0.1978021978021978, |
| "grad_norm": 0.39502742886543274, |
| "learning_rate": 0.00045256319290465634, |
| "loss": 0.7672, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.2043956043956044, |
| "grad_norm": 0.5479147434234619, |
| "learning_rate": 0.00045150332594235033, |
| "loss": 0.7583, |
| "step": 31 |
| }, |
| { |
| "epoch": 0.210989010989011, |
| "grad_norm": 0.33382317423820496, |
| "learning_rate": 0.00045044345898004437, |
| "loss": 0.7369, |
| "step": 32 |
| }, |
| { |
| "epoch": 0.2175824175824176, |
| "grad_norm": 0.49583446979522705, |
| "learning_rate": 0.0004493835920177384, |
| "loss": 0.8174, |
| "step": 33 |
| }, |
| { |
| "epoch": 0.22417582417582418, |
| "grad_norm": 0.3855836093425751, |
| "learning_rate": 0.00044832372505543234, |
| "loss": 0.7935, |
| "step": 34 |
| }, |
| { |
| "epoch": 0.23076923076923078, |
| "grad_norm": 0.45332643389701843, |
| "learning_rate": 0.0004472638580931264, |
| "loss": 0.7593, |
| "step": 35 |
| }, |
| { |
| "epoch": 0.23736263736263735, |
| "grad_norm": 0.3067397475242615, |
| "learning_rate": 0.0004462039911308204, |
| "loss": 0.6878, |
| "step": 36 |
| }, |
| { |
| "epoch": 0.24395604395604395, |
| "grad_norm": 0.3372962474822998, |
| "learning_rate": 0.0004451441241685144, |
| "loss": 0.7705, |
| "step": 37 |
| }, |
| { |
| "epoch": 0.25054945054945055, |
| "grad_norm": 0.40445494651794434, |
| "learning_rate": 0.00044408425720620845, |
| "loss": 0.7849, |
| "step": 38 |
| }, |
| { |
| "epoch": 0.2571428571428571, |
| "grad_norm": 0.3279916048049927, |
| "learning_rate": 0.0004430243902439025, |
| "loss": 0.715, |
| "step": 39 |
| }, |
| { |
| "epoch": 0.26373626373626374, |
| "grad_norm": 0.31464719772338867, |
| "learning_rate": 0.0004419645232815964, |
| "loss": 0.7322, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.2703296703296703, |
| "grad_norm": 0.27914178371429443, |
| "learning_rate": 0.00044090465631929047, |
| "loss": 0.6922, |
| "step": 41 |
| }, |
| { |
| "epoch": 0.27692307692307694, |
| "grad_norm": 0.27055102586746216, |
| "learning_rate": 0.0004398447893569845, |
| "loss": 0.7071, |
| "step": 42 |
| }, |
| { |
| "epoch": 0.2835164835164835, |
| "grad_norm": 0.30358147621154785, |
| "learning_rate": 0.0004387849223946785, |
| "loss": 0.7257, |
| "step": 43 |
| }, |
| { |
| "epoch": 0.29010989010989013, |
| "grad_norm": 0.3098717927932739, |
| "learning_rate": 0.00043772505543237253, |
| "loss": 0.6871, |
| "step": 44 |
| }, |
| { |
| "epoch": 0.2967032967032967, |
| "grad_norm": 0.32566770911216736, |
| "learning_rate": 0.0004366651884700666, |
| "loss": 0.7264, |
| "step": 45 |
| }, |
| { |
| "epoch": 0.3032967032967033, |
| "grad_norm": 0.3617362082004547, |
| "learning_rate": 0.0004356053215077605, |
| "loss": 0.6902, |
| "step": 46 |
| }, |
| { |
| "epoch": 0.3098901098901099, |
| "grad_norm": 0.43403375148773193, |
| "learning_rate": 0.00043454545454545455, |
| "loss": 0.7037, |
| "step": 47 |
| }, |
| { |
| "epoch": 0.31648351648351647, |
| "grad_norm": 0.33475279808044434, |
| "learning_rate": 0.0004334855875831486, |
| "loss": 0.7261, |
| "step": 48 |
| }, |
| { |
| "epoch": 0.3230769230769231, |
| "grad_norm": 0.34020453691482544, |
| "learning_rate": 0.0004324257206208426, |
| "loss": 0.6361, |
| "step": 49 |
| }, |
| { |
| "epoch": 0.32967032967032966, |
| "grad_norm": 0.29145631194114685, |
| "learning_rate": 0.0004313658536585366, |
| "loss": 0.6797, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.3362637362637363, |
| "grad_norm": 0.35258740186691284, |
| "learning_rate": 0.00043030598669623066, |
| "loss": 0.6811, |
| "step": 51 |
| }, |
| { |
| "epoch": 0.34285714285714286, |
| "grad_norm": 0.31551578640937805, |
| "learning_rate": 0.0004292461197339246, |
| "loss": 0.6662, |
| "step": 52 |
| }, |
| { |
| "epoch": 0.34945054945054943, |
| "grad_norm": 0.40033137798309326, |
| "learning_rate": 0.00042818625277161863, |
| "loss": 0.7066, |
| "step": 53 |
| }, |
| { |
| "epoch": 0.35604395604395606, |
| "grad_norm": 0.34928882122039795, |
| "learning_rate": 0.00042712638580931267, |
| "loss": 0.6704, |
| "step": 54 |
| }, |
| { |
| "epoch": 0.3626373626373626, |
| "grad_norm": 0.3890458345413208, |
| "learning_rate": 0.00042606651884700666, |
| "loss": 0.7074, |
| "step": 55 |
| }, |
| { |
| "epoch": 0.36923076923076925, |
| "grad_norm": 0.34585464000701904, |
| "learning_rate": 0.0004250066518847007, |
| "loss": 0.7147, |
| "step": 56 |
| }, |
| { |
| "epoch": 0.3758241758241758, |
| "grad_norm": 0.2896708548069, |
| "learning_rate": 0.00042394678492239474, |
| "loss": 0.7679, |
| "step": 57 |
| }, |
| { |
| "epoch": 0.3824175824175824, |
| "grad_norm": 0.3353179395198822, |
| "learning_rate": 0.00042288691796008867, |
| "loss": 0.6415, |
| "step": 58 |
| }, |
| { |
| "epoch": 0.389010989010989, |
| "grad_norm": 0.32367175817489624, |
| "learning_rate": 0.0004218270509977827, |
| "loss": 0.7697, |
| "step": 59 |
| }, |
| { |
| "epoch": 0.3956043956043956, |
| "grad_norm": 0.32652848958969116, |
| "learning_rate": 0.00042076718403547675, |
| "loss": 0.6835, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.4021978021978022, |
| "grad_norm": 0.2857896089553833, |
| "learning_rate": 0.00041970731707317074, |
| "loss": 0.7343, |
| "step": 61 |
| }, |
| { |
| "epoch": 0.4087912087912088, |
| "grad_norm": 0.35556867718696594, |
| "learning_rate": 0.0004186474501108648, |
| "loss": 0.6726, |
| "step": 62 |
| }, |
| { |
| "epoch": 0.4153846153846154, |
| "grad_norm": 0.3160596191883087, |
| "learning_rate": 0.0004175875831485588, |
| "loss": 0.7143, |
| "step": 63 |
| }, |
| { |
| "epoch": 0.421978021978022, |
| "grad_norm": 0.3572760224342346, |
| "learning_rate": 0.00041652771618625275, |
| "loss": 0.7587, |
| "step": 64 |
| }, |
| { |
| "epoch": 0.42857142857142855, |
| "grad_norm": 0.2954725921154022, |
| "learning_rate": 0.0004154678492239468, |
| "loss": 0.6812, |
| "step": 65 |
| }, |
| { |
| "epoch": 0.4351648351648352, |
| "grad_norm": 0.3745565116405487, |
| "learning_rate": 0.00041440798226164083, |
| "loss": 0.7258, |
| "step": 66 |
| }, |
| { |
| "epoch": 0.44175824175824174, |
| "grad_norm": 0.3284990191459656, |
| "learning_rate": 0.0004133481152993348, |
| "loss": 0.7153, |
| "step": 67 |
| }, |
| { |
| "epoch": 0.44835164835164837, |
| "grad_norm": 0.30010706186294556, |
| "learning_rate": 0.00041228824833702886, |
| "loss": 0.6984, |
| "step": 68 |
| }, |
| { |
| "epoch": 0.45494505494505494, |
| "grad_norm": 0.2915664613246918, |
| "learning_rate": 0.00041122838137472285, |
| "loss": 0.7052, |
| "step": 69 |
| }, |
| { |
| "epoch": 0.46153846153846156, |
| "grad_norm": 0.29902124404907227, |
| "learning_rate": 0.00041016851441241683, |
| "loss": 0.6934, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.46813186813186813, |
| "grad_norm": 0.3071542978286743, |
| "learning_rate": 0.00040910864745011087, |
| "loss": 0.708, |
| "step": 71 |
| }, |
| { |
| "epoch": 0.4747252747252747, |
| "grad_norm": 0.27577969431877136, |
| "learning_rate": 0.00040804878048780486, |
| "loss": 0.7428, |
| "step": 72 |
| }, |
| { |
| "epoch": 0.48131868131868133, |
| "grad_norm": 0.2982359826564789, |
| "learning_rate": 0.0004069889135254989, |
| "loss": 0.7779, |
| "step": 73 |
| }, |
| { |
| "epoch": 0.4879120879120879, |
| "grad_norm": 0.30666181445121765, |
| "learning_rate": 0.00040592904656319294, |
| "loss": 0.7067, |
| "step": 74 |
| }, |
| { |
| "epoch": 0.4945054945054945, |
| "grad_norm": 0.2994212508201599, |
| "learning_rate": 0.0004048691796008869, |
| "loss": 0.6904, |
| "step": 75 |
| }, |
| { |
| "epoch": 0.5010989010989011, |
| "grad_norm": 0.2912769317626953, |
| "learning_rate": 0.00040380931263858097, |
| "loss": 0.7374, |
| "step": 76 |
| }, |
| { |
| "epoch": 0.5076923076923077, |
| "grad_norm": 0.26267528533935547, |
| "learning_rate": 0.00040274944567627495, |
| "loss": 0.7415, |
| "step": 77 |
| }, |
| { |
| "epoch": 0.5142857142857142, |
| "grad_norm": 0.3228910565376282, |
| "learning_rate": 0.00040168957871396894, |
| "loss": 0.6345, |
| "step": 78 |
| }, |
| { |
| "epoch": 0.5208791208791209, |
| "grad_norm": 0.3353758752346039, |
| "learning_rate": 0.000400629711751663, |
| "loss": 0.6988, |
| "step": 79 |
| }, |
| { |
| "epoch": 0.5274725274725275, |
| "grad_norm": 0.34910428524017334, |
| "learning_rate": 0.000399569844789357, |
| "loss": 0.6955, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.5340659340659341, |
| "grad_norm": 0.35183653235435486, |
| "learning_rate": 0.000398509977827051, |
| "loss": 0.668, |
| "step": 81 |
| }, |
| { |
| "epoch": 0.5406593406593406, |
| "grad_norm": 0.2986193001270294, |
| "learning_rate": 0.00039745011086474505, |
| "loss": 0.7467, |
| "step": 82 |
| }, |
| { |
| "epoch": 0.5472527472527473, |
| "grad_norm": 0.3669707775115967, |
| "learning_rate": 0.00039639024390243904, |
| "loss": 0.7143, |
| "step": 83 |
| }, |
| { |
| "epoch": 0.5538461538461539, |
| "grad_norm": 0.3397374749183655, |
| "learning_rate": 0.000395330376940133, |
| "loss": 0.6948, |
| "step": 84 |
| }, |
| { |
| "epoch": 0.5604395604395604, |
| "grad_norm": 0.36349624395370483, |
| "learning_rate": 0.00039427050997782706, |
| "loss": 0.6381, |
| "step": 85 |
| }, |
| { |
| "epoch": 0.567032967032967, |
| "grad_norm": 0.3161599934101105, |
| "learning_rate": 0.0003932106430155211, |
| "loss": 0.7035, |
| "step": 86 |
| }, |
| { |
| "epoch": 0.5736263736263736, |
| "grad_norm": 0.34338268637657166, |
| "learning_rate": 0.0003921507760532151, |
| "loss": 0.6079, |
| "step": 87 |
| }, |
| { |
| "epoch": 0.5802197802197803, |
| "grad_norm": 0.3725844621658325, |
| "learning_rate": 0.00039109090909090913, |
| "loss": 0.6842, |
| "step": 88 |
| }, |
| { |
| "epoch": 0.5868131868131868, |
| "grad_norm": 0.3499543368816376, |
| "learning_rate": 0.0003900310421286031, |
| "loss": 0.6663, |
| "step": 89 |
| }, |
| { |
| "epoch": 0.5934065934065934, |
| "grad_norm": 0.27967825531959534, |
| "learning_rate": 0.0003889711751662971, |
| "loss": 0.7043, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.6, |
| "grad_norm": 0.42830947041511536, |
| "learning_rate": 0.00038791130820399114, |
| "loss": 0.6066, |
| "step": 91 |
| }, |
| { |
| "epoch": 0.6065934065934065, |
| "grad_norm": 0.2985455095767975, |
| "learning_rate": 0.0003868514412416852, |
| "loss": 0.7876, |
| "step": 92 |
| }, |
| { |
| "epoch": 0.6131868131868132, |
| "grad_norm": 0.31391894817352295, |
| "learning_rate": 0.00038579157427937917, |
| "loss": 0.7133, |
| "step": 93 |
| }, |
| { |
| "epoch": 0.6197802197802198, |
| "grad_norm": 0.3369094133377075, |
| "learning_rate": 0.0003847317073170732, |
| "loss": 0.6965, |
| "step": 94 |
| }, |
| { |
| "epoch": 0.6263736263736264, |
| "grad_norm": 0.35192903876304626, |
| "learning_rate": 0.0003836718403547672, |
| "loss": 0.738, |
| "step": 95 |
| }, |
| { |
| "epoch": 0.6329670329670329, |
| "grad_norm": 0.30551454424858093, |
| "learning_rate": 0.0003826119733924612, |
| "loss": 0.6792, |
| "step": 96 |
| }, |
| { |
| "epoch": 0.6395604395604395, |
| "grad_norm": 0.2987115681171417, |
| "learning_rate": 0.0003815521064301552, |
| "loss": 0.684, |
| "step": 97 |
| }, |
| { |
| "epoch": 0.6461538461538462, |
| "grad_norm": 0.2904681861400604, |
| "learning_rate": 0.00038049223946784927, |
| "loss": 0.7354, |
| "step": 98 |
| }, |
| { |
| "epoch": 0.6527472527472528, |
| "grad_norm": 0.3646203279495239, |
| "learning_rate": 0.00037943237250554325, |
| "loss": 0.7511, |
| "step": 99 |
| }, |
| { |
| "epoch": 0.6593406593406593, |
| "grad_norm": 0.27354785799980164, |
| "learning_rate": 0.0003783725055432373, |
| "loss": 0.6868, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.6659340659340659, |
| "grad_norm": 0.3103290796279907, |
| "learning_rate": 0.0003773126385809313, |
| "loss": 0.6935, |
| "step": 101 |
| }, |
| { |
| "epoch": 0.6725274725274726, |
| "grad_norm": 0.33116936683654785, |
| "learning_rate": 0.00037625277161862527, |
| "loss": 0.7088, |
| "step": 102 |
| }, |
| { |
| "epoch": 0.6791208791208792, |
| "grad_norm": 0.31001362204551697, |
| "learning_rate": 0.0003751929046563193, |
| "loss": 0.7556, |
| "step": 103 |
| }, |
| { |
| "epoch": 0.6857142857142857, |
| "grad_norm": 0.34533751010894775, |
| "learning_rate": 0.00037413303769401335, |
| "loss": 0.6535, |
| "step": 104 |
| }, |
| { |
| "epoch": 0.6923076923076923, |
| "grad_norm": 0.3335258364677429, |
| "learning_rate": 0.00037307317073170733, |
| "loss": 0.7302, |
| "step": 105 |
| }, |
| { |
| "epoch": 0.6989010989010989, |
| "grad_norm": 0.2870931923389435, |
| "learning_rate": 0.0003720133037694014, |
| "loss": 0.6758, |
| "step": 106 |
| }, |
| { |
| "epoch": 0.7054945054945055, |
| "grad_norm": 0.3254625201225281, |
| "learning_rate": 0.00037095343680709536, |
| "loss": 0.7593, |
| "step": 107 |
| }, |
| { |
| "epoch": 0.7120879120879121, |
| "grad_norm": 0.30663806200027466, |
| "learning_rate": 0.00036989356984478935, |
| "loss": 0.7143, |
| "step": 108 |
| }, |
| { |
| "epoch": 0.7186813186813187, |
| "grad_norm": 0.2793548107147217, |
| "learning_rate": 0.0003688337028824834, |
| "loss": 0.7234, |
| "step": 109 |
| }, |
| { |
| "epoch": 0.7252747252747253, |
| "grad_norm": 0.31374335289001465, |
| "learning_rate": 0.0003677738359201774, |
| "loss": 0.7025, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.7318681318681318, |
| "grad_norm": 0.3022034764289856, |
| "learning_rate": 0.0003667139689578714, |
| "loss": 0.6732, |
| "step": 111 |
| }, |
| { |
| "epoch": 0.7384615384615385, |
| "grad_norm": 0.3161036968231201, |
| "learning_rate": 0.00036565410199556546, |
| "loss": 0.7136, |
| "step": 112 |
| }, |
| { |
| "epoch": 0.7450549450549451, |
| "grad_norm": 0.30554062128067017, |
| "learning_rate": 0.0003645942350332594, |
| "loss": 0.6756, |
| "step": 113 |
| }, |
| { |
| "epoch": 0.7516483516483516, |
| "grad_norm": 0.29600027203559875, |
| "learning_rate": 0.00036353436807095343, |
| "loss": 0.7567, |
| "step": 114 |
| }, |
| { |
| "epoch": 0.7582417582417582, |
| "grad_norm": 0.29495969414711, |
| "learning_rate": 0.00036247450110864747, |
| "loss": 0.6993, |
| "step": 115 |
| }, |
| { |
| "epoch": 0.7648351648351648, |
| "grad_norm": 0.36187559366226196, |
| "learning_rate": 0.00036141463414634146, |
| "loss": 0.7093, |
| "step": 116 |
| }, |
| { |
| "epoch": 0.7714285714285715, |
| "grad_norm": 0.2875276505947113, |
| "learning_rate": 0.0003603547671840355, |
| "loss": 0.6432, |
| "step": 117 |
| }, |
| { |
| "epoch": 0.778021978021978, |
| "grad_norm": 0.3426610827445984, |
| "learning_rate": 0.00035929490022172954, |
| "loss": 0.6835, |
| "step": 118 |
| }, |
| { |
| "epoch": 0.7846153846153846, |
| "grad_norm": 0.3620908558368683, |
| "learning_rate": 0.00035823503325942347, |
| "loss": 0.756, |
| "step": 119 |
| }, |
| { |
| "epoch": 0.7912087912087912, |
| "grad_norm": 0.2915194630622864, |
| "learning_rate": 0.0003571751662971175, |
| "loss": 0.6883, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.7978021978021979, |
| "grad_norm": 0.32311534881591797, |
| "learning_rate": 0.00035611529933481155, |
| "loss": 0.727, |
| "step": 121 |
| }, |
| { |
| "epoch": 0.8043956043956044, |
| "grad_norm": 0.3259125053882599, |
| "learning_rate": 0.00035505543237250554, |
| "loss": 0.6523, |
| "step": 122 |
| }, |
| { |
| "epoch": 0.810989010989011, |
| "grad_norm": 0.33542436361312866, |
| "learning_rate": 0.0003539955654101996, |
| "loss": 0.6245, |
| "step": 123 |
| }, |
| { |
| "epoch": 0.8175824175824176, |
| "grad_norm": 0.4077654480934143, |
| "learning_rate": 0.0003529356984478936, |
| "loss": 0.7295, |
| "step": 124 |
| }, |
| { |
| "epoch": 0.8241758241758241, |
| "grad_norm": 0.35173124074935913, |
| "learning_rate": 0.00035187583148558755, |
| "loss": 0.7189, |
| "step": 125 |
| }, |
| { |
| "epoch": 0.8307692307692308, |
| "grad_norm": 0.3285554349422455, |
| "learning_rate": 0.0003508159645232816, |
| "loss": 0.7179, |
| "step": 126 |
| }, |
| { |
| "epoch": 0.8373626373626374, |
| "grad_norm": 0.3704049289226532, |
| "learning_rate": 0.00034975609756097563, |
| "loss": 0.7213, |
| "step": 127 |
| }, |
| { |
| "epoch": 0.843956043956044, |
| "grad_norm": 0.30321356654167175, |
| "learning_rate": 0.0003486962305986696, |
| "loss": 0.701, |
| "step": 128 |
| }, |
| { |
| "epoch": 0.8505494505494505, |
| "grad_norm": 0.25718748569488525, |
| "learning_rate": 0.00034763636363636366, |
| "loss": 0.7196, |
| "step": 129 |
| }, |
| { |
| "epoch": 0.8571428571428571, |
| "grad_norm": 0.38271740078926086, |
| "learning_rate": 0.0003465764966740577, |
| "loss": 0.7567, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.8637362637362638, |
| "grad_norm": 0.3291567862033844, |
| "learning_rate": 0.00034551662971175163, |
| "loss": 0.7078, |
| "step": 131 |
| }, |
| { |
| "epoch": 0.8703296703296703, |
| "grad_norm": 0.3053141236305237, |
| "learning_rate": 0.00034445676274944567, |
| "loss": 0.6136, |
| "step": 132 |
| }, |
| { |
| "epoch": 0.8769230769230769, |
| "grad_norm": 0.300508052110672, |
| "learning_rate": 0.0003433968957871397, |
| "loss": 0.7905, |
| "step": 133 |
| }, |
| { |
| "epoch": 0.8835164835164835, |
| "grad_norm": 0.3538469970226288, |
| "learning_rate": 0.0003423370288248337, |
| "loss": 0.6981, |
| "step": 134 |
| }, |
| { |
| "epoch": 0.8901098901098901, |
| "grad_norm": 0.285239040851593, |
| "learning_rate": 0.00034127716186252774, |
| "loss": 0.7405, |
| "step": 135 |
| }, |
| { |
| "epoch": 0.8967032967032967, |
| "grad_norm": 0.39319372177124023, |
| "learning_rate": 0.0003402172949002218, |
| "loss": 0.7222, |
| "step": 136 |
| }, |
| { |
| "epoch": 0.9032967032967033, |
| "grad_norm": 0.381517231464386, |
| "learning_rate": 0.00033915742793791577, |
| "loss": 0.712, |
| "step": 137 |
| }, |
| { |
| "epoch": 0.9098901098901099, |
| "grad_norm": 0.3278730511665344, |
| "learning_rate": 0.00033809756097560975, |
| "loss": 0.6998, |
| "step": 138 |
| }, |
| { |
| "epoch": 0.9164835164835164, |
| "grad_norm": 0.3165235221385956, |
| "learning_rate": 0.0003370376940133038, |
| "loss": 0.7314, |
| "step": 139 |
| }, |
| { |
| "epoch": 0.9230769230769231, |
| "grad_norm": 0.34131768345832825, |
| "learning_rate": 0.0003359778270509978, |
| "loss": 0.696, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.9296703296703297, |
| "grad_norm": 0.3008774518966675, |
| "learning_rate": 0.0003349179600886918, |
| "loss": 0.6536, |
| "step": 141 |
| }, |
| { |
| "epoch": 0.9362637362637363, |
| "grad_norm": 0.3007766306400299, |
| "learning_rate": 0.00033385809312638586, |
| "loss": 0.6952, |
| "step": 142 |
| }, |
| { |
| "epoch": 0.9428571428571428, |
| "grad_norm": 0.3253723680973053, |
| "learning_rate": 0.00033279822616407985, |
| "loss": 0.7443, |
| "step": 143 |
| }, |
| { |
| "epoch": 0.9494505494505494, |
| "grad_norm": 0.31287214159965515, |
| "learning_rate": 0.00033173835920177384, |
| "loss": 0.7231, |
| "step": 144 |
| }, |
| { |
| "epoch": 0.9560439560439561, |
| "grad_norm": 0.31145626306533813, |
| "learning_rate": 0.0003306784922394679, |
| "loss": 0.6726, |
| "step": 145 |
| }, |
| { |
| "epoch": 0.9626373626373627, |
| "grad_norm": 0.3471812307834625, |
| "learning_rate": 0.00032961862527716186, |
| "loss": 0.685, |
| "step": 146 |
| }, |
| { |
| "epoch": 0.9692307692307692, |
| "grad_norm": 0.33790674805641174, |
| "learning_rate": 0.0003285587583148559, |
| "loss": 0.6287, |
| "step": 147 |
| }, |
| { |
| "epoch": 0.9758241758241758, |
| "grad_norm": 0.29381513595581055, |
| "learning_rate": 0.00032749889135254994, |
| "loss": 0.7269, |
| "step": 148 |
| }, |
| { |
| "epoch": 0.9824175824175824, |
| "grad_norm": 0.33163395524024963, |
| "learning_rate": 0.00032643902439024393, |
| "loss": 0.6688, |
| "step": 149 |
| }, |
| { |
| "epoch": 0.989010989010989, |
| "grad_norm": 0.28233274817466736, |
| "learning_rate": 0.0003253791574279379, |
| "loss": 0.7459, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.9956043956043956, |
| "grad_norm": 0.2697957456111908, |
| "learning_rate": 0.0003243192904656319, |
| "loss": 0.7204, |
| "step": 151 |
| }, |
| { |
| "epoch": 1.0, |
| "grad_norm": 0.3910374939441681, |
| "learning_rate": 0.00032325942350332594, |
| "loss": 0.7153, |
| "step": 152 |
| }, |
| { |
| "epoch": 1.0065934065934066, |
| "grad_norm": 0.31035199761390686, |
| "learning_rate": 0.00032219955654102, |
| "loss": 0.6372, |
| "step": 153 |
| }, |
| { |
| "epoch": 1.0131868131868131, |
| "grad_norm": 0.35336118936538696, |
| "learning_rate": 0.00032113968957871397, |
| "loss": 0.6743, |
| "step": 154 |
| }, |
| { |
| "epoch": 1.0197802197802197, |
| "grad_norm": 0.34364980459213257, |
| "learning_rate": 0.000320079822616408, |
| "loss": 0.6599, |
| "step": 155 |
| }, |
| { |
| "epoch": 1.0263736263736263, |
| "grad_norm": 0.43252575397491455, |
| "learning_rate": 0.000319019955654102, |
| "loss": 0.7043, |
| "step": 156 |
| }, |
| { |
| "epoch": 1.032967032967033, |
| "grad_norm": 0.3506278991699219, |
| "learning_rate": 0.000317960088691796, |
| "loss": 0.6128, |
| "step": 157 |
| }, |
| { |
| "epoch": 1.0395604395604396, |
| "grad_norm": 0.2789537012577057, |
| "learning_rate": 0.00031690022172949, |
| "loss": 0.5281, |
| "step": 158 |
| }, |
| { |
| "epoch": 1.0461538461538462, |
| "grad_norm": 0.35527005791664124, |
| "learning_rate": 0.00031584035476718407, |
| "loss": 0.6422, |
| "step": 159 |
| }, |
| { |
| "epoch": 1.0527472527472528, |
| "grad_norm": 0.31511110067367554, |
| "learning_rate": 0.00031478048780487805, |
| "loss": 0.5686, |
| "step": 160 |
| }, |
| { |
| "epoch": 1.0593406593406594, |
| "grad_norm": 0.3340548574924469, |
| "learning_rate": 0.0003137206208425721, |
| "loss": 0.6509, |
| "step": 161 |
| }, |
| { |
| "epoch": 1.065934065934066, |
| "grad_norm": 0.32808569073677063, |
| "learning_rate": 0.0003126607538802661, |
| "loss": 0.6608, |
| "step": 162 |
| }, |
| { |
| "epoch": 1.0725274725274725, |
| "grad_norm": 0.33119022846221924, |
| "learning_rate": 0.00031160088691796007, |
| "loss": 0.6298, |
| "step": 163 |
| }, |
| { |
| "epoch": 1.079120879120879, |
| "grad_norm": 0.3775770664215088, |
| "learning_rate": 0.0003105410199556541, |
| "loss": 0.656, |
| "step": 164 |
| }, |
| { |
| "epoch": 1.0857142857142856, |
| "grad_norm": 0.30977287888526917, |
| "learning_rate": 0.00030948115299334815, |
| "loss": 0.6113, |
| "step": 165 |
| }, |
| { |
| "epoch": 1.0923076923076924, |
| "grad_norm": 0.37175706028938293, |
| "learning_rate": 0.00030842128603104213, |
| "loss": 0.7013, |
| "step": 166 |
| }, |
| { |
| "epoch": 1.098901098901099, |
| "grad_norm": 0.351672887802124, |
| "learning_rate": 0.0003073614190687362, |
| "loss": 0.6631, |
| "step": 167 |
| }, |
| { |
| "epoch": 1.1054945054945056, |
| "grad_norm": 0.359859436750412, |
| "learning_rate": 0.00030630155210643016, |
| "loss": 0.6584, |
| "step": 168 |
| }, |
| { |
| "epoch": 1.1120879120879121, |
| "grad_norm": 0.4018190801143646, |
| "learning_rate": 0.00030524168514412415, |
| "loss": 0.6357, |
| "step": 169 |
| }, |
| { |
| "epoch": 1.1186813186813187, |
| "grad_norm": 0.30350229144096375, |
| "learning_rate": 0.0003041818181818182, |
| "loss": 0.6667, |
| "step": 170 |
| }, |
| { |
| "epoch": 1.1252747252747253, |
| "grad_norm": 0.35754507780075073, |
| "learning_rate": 0.00030312195121951223, |
| "loss": 0.6565, |
| "step": 171 |
| }, |
| { |
| "epoch": 1.1318681318681318, |
| "grad_norm": 0.399149090051651, |
| "learning_rate": 0.0003020620842572062, |
| "loss": 0.6746, |
| "step": 172 |
| }, |
| { |
| "epoch": 1.1384615384615384, |
| "grad_norm": 0.3958664834499359, |
| "learning_rate": 0.00030100221729490026, |
| "loss": 0.6506, |
| "step": 173 |
| }, |
| { |
| "epoch": 1.145054945054945, |
| "grad_norm": 0.32290762662887573, |
| "learning_rate": 0.00029994235033259424, |
| "loss": 0.6816, |
| "step": 174 |
| }, |
| { |
| "epoch": 1.1516483516483516, |
| "grad_norm": 0.48313409090042114, |
| "learning_rate": 0.00029888248337028823, |
| "loss": 0.6262, |
| "step": 175 |
| }, |
| { |
| "epoch": 1.1582417582417581, |
| "grad_norm": 0.35737237334251404, |
| "learning_rate": 0.00029782261640798227, |
| "loss": 0.654, |
| "step": 176 |
| }, |
| { |
| "epoch": 1.164835164835165, |
| "grad_norm": 0.3466680943965912, |
| "learning_rate": 0.0002967627494456763, |
| "loss": 0.6356, |
| "step": 177 |
| }, |
| { |
| "epoch": 1.1714285714285715, |
| "grad_norm": 0.39362478256225586, |
| "learning_rate": 0.0002957028824833703, |
| "loss": 0.6485, |
| "step": 178 |
| }, |
| { |
| "epoch": 1.178021978021978, |
| "grad_norm": 0.402873158454895, |
| "learning_rate": 0.00029464301552106434, |
| "loss": 0.6457, |
| "step": 179 |
| }, |
| { |
| "epoch": 1.1846153846153846, |
| "grad_norm": 0.3372310996055603, |
| "learning_rate": 0.0002935831485587583, |
| "loss": 0.6354, |
| "step": 180 |
| }, |
| { |
| "epoch": 1.1912087912087912, |
| "grad_norm": 0.3874308168888092, |
| "learning_rate": 0.0002925232815964523, |
| "loss": 0.7243, |
| "step": 181 |
| }, |
| { |
| "epoch": 1.1978021978021978, |
| "grad_norm": 0.4037889838218689, |
| "learning_rate": 0.00029146341463414635, |
| "loss": 0.6482, |
| "step": 182 |
| }, |
| { |
| "epoch": 1.2043956043956043, |
| "grad_norm": 0.37217509746551514, |
| "learning_rate": 0.0002904035476718404, |
| "loss": 0.7087, |
| "step": 183 |
| }, |
| { |
| "epoch": 1.210989010989011, |
| "grad_norm": 0.3493293821811676, |
| "learning_rate": 0.0002893436807095344, |
| "loss": 0.6694, |
| "step": 184 |
| }, |
| { |
| "epoch": 1.2175824175824177, |
| "grad_norm": 0.4025772213935852, |
| "learning_rate": 0.0002882838137472284, |
| "loss": 0.6237, |
| "step": 185 |
| }, |
| { |
| "epoch": 1.2241758241758243, |
| "grad_norm": 0.34033340215682983, |
| "learning_rate": 0.0002872239467849224, |
| "loss": 0.5906, |
| "step": 186 |
| }, |
| { |
| "epoch": 1.2307692307692308, |
| "grad_norm": 0.36151084303855896, |
| "learning_rate": 0.0002861640798226164, |
| "loss": 0.6349, |
| "step": 187 |
| }, |
| { |
| "epoch": 1.2373626373626374, |
| "grad_norm": 0.36955204606056213, |
| "learning_rate": 0.00028510421286031043, |
| "loss": 0.673, |
| "step": 188 |
| }, |
| { |
| "epoch": 1.243956043956044, |
| "grad_norm": 0.3759375810623169, |
| "learning_rate": 0.0002840443458980045, |
| "loss": 0.6669, |
| "step": 189 |
| }, |
| { |
| "epoch": 1.2505494505494505, |
| "grad_norm": 0.3453136086463928, |
| "learning_rate": 0.00028298447893569846, |
| "loss": 0.6001, |
| "step": 190 |
| }, |
| { |
| "epoch": 1.2571428571428571, |
| "grad_norm": 0.39012014865875244, |
| "learning_rate": 0.0002819246119733925, |
| "loss": 0.6824, |
| "step": 191 |
| }, |
| { |
| "epoch": 1.2637362637362637, |
| "grad_norm": 0.35880038142204285, |
| "learning_rate": 0.00028086474501108643, |
| "loss": 0.6539, |
| "step": 192 |
| }, |
| { |
| "epoch": 1.2703296703296703, |
| "grad_norm": 0.4133313000202179, |
| "learning_rate": 0.0002798048780487805, |
| "loss": 0.6706, |
| "step": 193 |
| }, |
| { |
| "epoch": 1.2769230769230768, |
| "grad_norm": 0.39098864793777466, |
| "learning_rate": 0.0002787450110864745, |
| "loss": 0.5938, |
| "step": 194 |
| }, |
| { |
| "epoch": 1.2835164835164834, |
| "grad_norm": 0.32598677277565, |
| "learning_rate": 0.0002776851441241685, |
| "loss": 0.7272, |
| "step": 195 |
| }, |
| { |
| "epoch": 1.2901098901098902, |
| "grad_norm": 0.3618911802768707, |
| "learning_rate": 0.00027662527716186254, |
| "loss": 0.676, |
| "step": 196 |
| }, |
| { |
| "epoch": 1.2967032967032968, |
| "grad_norm": 0.37141621112823486, |
| "learning_rate": 0.0002755654101995566, |
| "loss": 0.626, |
| "step": 197 |
| }, |
| { |
| "epoch": 1.3032967032967033, |
| "grad_norm": 0.3062952160835266, |
| "learning_rate": 0.0002745055432372505, |
| "loss": 0.5656, |
| "step": 198 |
| }, |
| { |
| "epoch": 1.30989010989011, |
| "grad_norm": 0.37736329436302185, |
| "learning_rate": 0.00027344567627494455, |
| "loss": 0.7134, |
| "step": 199 |
| }, |
| { |
| "epoch": 1.3164835164835165, |
| "grad_norm": 0.3816686272621155, |
| "learning_rate": 0.0002723858093126386, |
| "loss": 0.6379, |
| "step": 200 |
| }, |
| { |
| "epoch": 1.323076923076923, |
| "grad_norm": 0.3832955062389374, |
| "learning_rate": 0.0002713259423503326, |
| "loss": 0.6502, |
| "step": 201 |
| }, |
| { |
| "epoch": 1.3296703296703296, |
| "grad_norm": 0.33459657430648804, |
| "learning_rate": 0.0002702660753880266, |
| "loss": 0.659, |
| "step": 202 |
| }, |
| { |
| "epoch": 1.3362637362637364, |
| "grad_norm": 0.42061227560043335, |
| "learning_rate": 0.00026920620842572066, |
| "loss": 0.6586, |
| "step": 203 |
| }, |
| { |
| "epoch": 1.342857142857143, |
| "grad_norm": 0.3629121780395508, |
| "learning_rate": 0.00026814634146341465, |
| "loss": 0.5816, |
| "step": 204 |
| }, |
| { |
| "epoch": 1.3494505494505495, |
| "grad_norm": 0.3140220642089844, |
| "learning_rate": 0.00026708647450110864, |
| "loss": 0.6568, |
| "step": 205 |
| }, |
| { |
| "epoch": 1.3560439560439561, |
| "grad_norm": 0.36066094040870667, |
| "learning_rate": 0.0002660266075388027, |
| "loss": 0.6666, |
| "step": 206 |
| }, |
| { |
| "epoch": 1.3626373626373627, |
| "grad_norm": 0.3712823987007141, |
| "learning_rate": 0.00026496674057649666, |
| "loss": 0.6371, |
| "step": 207 |
| }, |
| { |
| "epoch": 1.3692307692307693, |
| "grad_norm": 0.4108193516731262, |
| "learning_rate": 0.0002639068736141907, |
| "loss": 0.6652, |
| "step": 208 |
| }, |
| { |
| "epoch": 1.3758241758241758, |
| "grad_norm": 0.3651275932788849, |
| "learning_rate": 0.00026284700665188474, |
| "loss": 0.6161, |
| "step": 209 |
| }, |
| { |
| "epoch": 1.3824175824175824, |
| "grad_norm": 0.3415852189064026, |
| "learning_rate": 0.00026178713968957873, |
| "loss": 0.6737, |
| "step": 210 |
| }, |
| { |
| "epoch": 1.389010989010989, |
| "grad_norm": 0.4050087034702301, |
| "learning_rate": 0.0002607272727272727, |
| "loss": 0.5946, |
| "step": 211 |
| }, |
| { |
| "epoch": 1.3956043956043955, |
| "grad_norm": 0.3588634431362152, |
| "learning_rate": 0.00025966740576496676, |
| "loss": 0.6445, |
| "step": 212 |
| }, |
| { |
| "epoch": 1.402197802197802, |
| "grad_norm": 0.35990527272224426, |
| "learning_rate": 0.00025860753880266074, |
| "loss": 0.6944, |
| "step": 213 |
| }, |
| { |
| "epoch": 1.4087912087912087, |
| "grad_norm": 0.3686452805995941, |
| "learning_rate": 0.0002575476718403548, |
| "loss": 0.6887, |
| "step": 214 |
| }, |
| { |
| "epoch": 1.4153846153846155, |
| "grad_norm": 0.3962976932525635, |
| "learning_rate": 0.0002564878048780488, |
| "loss": 0.6894, |
| "step": 215 |
| }, |
| { |
| "epoch": 1.421978021978022, |
| "grad_norm": 0.3887828588485718, |
| "learning_rate": 0.0002554279379157428, |
| "loss": 0.6455, |
| "step": 216 |
| }, |
| { |
| "epoch": 1.4285714285714286, |
| "grad_norm": 0.3367188274860382, |
| "learning_rate": 0.0002543680709534368, |
| "loss": 0.5887, |
| "step": 217 |
| }, |
| { |
| "epoch": 1.4351648351648352, |
| "grad_norm": 0.35716238617897034, |
| "learning_rate": 0.00025330820399113084, |
| "loss": 0.6222, |
| "step": 218 |
| }, |
| { |
| "epoch": 1.4417582417582417, |
| "grad_norm": 0.36131733655929565, |
| "learning_rate": 0.0002522483370288248, |
| "loss": 0.6142, |
| "step": 219 |
| }, |
| { |
| "epoch": 1.4483516483516483, |
| "grad_norm": 0.40967121720314026, |
| "learning_rate": 0.00025118847006651887, |
| "loss": 0.6257, |
| "step": 220 |
| }, |
| { |
| "epoch": 1.4549450549450549, |
| "grad_norm": 0.35617414116859436, |
| "learning_rate": 0.0002501286031042129, |
| "loss": 0.6287, |
| "step": 221 |
| }, |
| { |
| "epoch": 1.4615384615384617, |
| "grad_norm": 0.3525332808494568, |
| "learning_rate": 0.0002490687361419069, |
| "loss": 0.6664, |
| "step": 222 |
| }, |
| { |
| "epoch": 1.4681318681318682, |
| "grad_norm": 0.39492669701576233, |
| "learning_rate": 0.0002480088691796009, |
| "loss": 0.6249, |
| "step": 223 |
| }, |
| { |
| "epoch": 1.4747252747252748, |
| "grad_norm": 0.35548609495162964, |
| "learning_rate": 0.0002469490022172949, |
| "loss": 0.5817, |
| "step": 224 |
| }, |
| { |
| "epoch": 1.4813186813186814, |
| "grad_norm": 0.38659876585006714, |
| "learning_rate": 0.0002458891352549889, |
| "loss": 0.6588, |
| "step": 225 |
| }, |
| { |
| "epoch": 1.487912087912088, |
| "grad_norm": 0.3900056779384613, |
| "learning_rate": 0.00024482926829268295, |
| "loss": 0.6637, |
| "step": 226 |
| }, |
| { |
| "epoch": 1.4945054945054945, |
| "grad_norm": 0.4056378901004791, |
| "learning_rate": 0.00024376940133037696, |
| "loss": 0.5983, |
| "step": 227 |
| }, |
| { |
| "epoch": 1.501098901098901, |
| "grad_norm": 0.3593887984752655, |
| "learning_rate": 0.00024270953436807097, |
| "loss": 0.6192, |
| "step": 228 |
| }, |
| { |
| "epoch": 1.5076923076923077, |
| "grad_norm": 0.3599669933319092, |
| "learning_rate": 0.00024164966740576496, |
| "loss": 0.5867, |
| "step": 229 |
| }, |
| { |
| "epoch": 1.5142857142857142, |
| "grad_norm": 0.3734591007232666, |
| "learning_rate": 0.00024058980044345897, |
| "loss": 0.6821, |
| "step": 230 |
| }, |
| { |
| "epoch": 1.5208791208791208, |
| "grad_norm": 0.3447280824184418, |
| "learning_rate": 0.000239529933481153, |
| "loss": 0.6375, |
| "step": 231 |
| }, |
| { |
| "epoch": 1.5274725274725274, |
| "grad_norm": 0.36874809861183167, |
| "learning_rate": 0.00023847006651884703, |
| "loss": 0.6779, |
| "step": 232 |
| }, |
| { |
| "epoch": 1.534065934065934, |
| "grad_norm": 0.38847875595092773, |
| "learning_rate": 0.00023741019955654104, |
| "loss": 0.6667, |
| "step": 233 |
| }, |
| { |
| "epoch": 1.5406593406593405, |
| "grad_norm": 0.39220407605171204, |
| "learning_rate": 0.00023635033259423503, |
| "loss": 0.6733, |
| "step": 234 |
| }, |
| { |
| "epoch": 1.5472527472527473, |
| "grad_norm": 0.44259729981422424, |
| "learning_rate": 0.00023529046563192907, |
| "loss": 0.6365, |
| "step": 235 |
| }, |
| { |
| "epoch": 1.5538461538461539, |
| "grad_norm": 0.37200653553009033, |
| "learning_rate": 0.00023423059866962308, |
| "loss": 0.6352, |
| "step": 236 |
| }, |
| { |
| "epoch": 1.5604395604395604, |
| "grad_norm": 0.3853394389152527, |
| "learning_rate": 0.00023317073170731707, |
| "loss": 0.6432, |
| "step": 237 |
| }, |
| { |
| "epoch": 1.567032967032967, |
| "grad_norm": 0.37795448303222656, |
| "learning_rate": 0.0002321108647450111, |
| "loss": 0.5782, |
| "step": 238 |
| }, |
| { |
| "epoch": 1.5736263736263736, |
| "grad_norm": 0.3939788043498993, |
| "learning_rate": 0.00023105099778270512, |
| "loss": 0.6513, |
| "step": 239 |
| }, |
| { |
| "epoch": 1.5802197802197804, |
| "grad_norm": 0.37062007188796997, |
| "learning_rate": 0.0002299911308203991, |
| "loss": 0.6167, |
| "step": 240 |
| }, |
| { |
| "epoch": 1.586813186813187, |
| "grad_norm": 0.3661113679409027, |
| "learning_rate": 0.00022893126385809315, |
| "loss": 0.6161, |
| "step": 241 |
| }, |
| { |
| "epoch": 1.5934065934065935, |
| "grad_norm": 0.3689832091331482, |
| "learning_rate": 0.00022787139689578716, |
| "loss": 0.6776, |
| "step": 242 |
| }, |
| { |
| "epoch": 1.6, |
| "grad_norm": 0.3579269349575043, |
| "learning_rate": 0.00022681152993348115, |
| "loss": 0.6498, |
| "step": 243 |
| }, |
| { |
| "epoch": 1.6065934065934067, |
| "grad_norm": 0.3285681903362274, |
| "learning_rate": 0.00022575166297117516, |
| "loss": 0.6519, |
| "step": 244 |
| }, |
| { |
| "epoch": 1.6131868131868132, |
| "grad_norm": 0.34521231055259705, |
| "learning_rate": 0.0002246917960088692, |
| "loss": 0.6447, |
| "step": 245 |
| }, |
| { |
| "epoch": 1.6197802197802198, |
| "grad_norm": 0.3420107066631317, |
| "learning_rate": 0.0002236319290465632, |
| "loss": 0.6388, |
| "step": 246 |
| }, |
| { |
| "epoch": 1.6263736263736264, |
| "grad_norm": 0.35361817479133606, |
| "learning_rate": 0.0002225720620842572, |
| "loss": 0.6077, |
| "step": 247 |
| }, |
| { |
| "epoch": 1.632967032967033, |
| "grad_norm": 0.38164547085762024, |
| "learning_rate": 0.00022151219512195125, |
| "loss": 0.5999, |
| "step": 248 |
| }, |
| { |
| "epoch": 1.6395604395604395, |
| "grad_norm": 0.3841310441493988, |
| "learning_rate": 0.00022045232815964523, |
| "loss": 0.6645, |
| "step": 249 |
| }, |
| { |
| "epoch": 1.646153846153846, |
| "grad_norm": 0.36244288086891174, |
| "learning_rate": 0.00021939246119733925, |
| "loss": 0.6422, |
| "step": 250 |
| }, |
| { |
| "epoch": 1.6527472527472526, |
| "grad_norm": 0.35995614528656006, |
| "learning_rate": 0.0002183325942350333, |
| "loss": 0.6243, |
| "step": 251 |
| }, |
| { |
| "epoch": 1.6593406593406592, |
| "grad_norm": 0.3498387336730957, |
| "learning_rate": 0.00021727272727272727, |
| "loss": 0.7294, |
| "step": 252 |
| }, |
| { |
| "epoch": 1.6659340659340658, |
| "grad_norm": 0.3545849919319153, |
| "learning_rate": 0.0002162128603104213, |
| "loss": 0.6, |
| "step": 253 |
| }, |
| { |
| "epoch": 1.6725274725274726, |
| "grad_norm": 0.3808715045452118, |
| "learning_rate": 0.00021515299334811533, |
| "loss": 0.6017, |
| "step": 254 |
| }, |
| { |
| "epoch": 1.6791208791208792, |
| "grad_norm": 0.3720131814479828, |
| "learning_rate": 0.00021409312638580931, |
| "loss": 0.698, |
| "step": 255 |
| }, |
| { |
| "epoch": 1.6857142857142857, |
| "grad_norm": 0.365560382604599, |
| "learning_rate": 0.00021303325942350333, |
| "loss": 0.6549, |
| "step": 256 |
| }, |
| { |
| "epoch": 1.6923076923076923, |
| "grad_norm": 0.3739791810512543, |
| "learning_rate": 0.00021197339246119737, |
| "loss": 0.6619, |
| "step": 257 |
| }, |
| { |
| "epoch": 1.6989010989010989, |
| "grad_norm": 0.3811318278312683, |
| "learning_rate": 0.00021091352549889135, |
| "loss": 0.6728, |
| "step": 258 |
| }, |
| { |
| "epoch": 1.7054945054945057, |
| "grad_norm": 0.37915971875190735, |
| "learning_rate": 0.00020985365853658537, |
| "loss": 0.6009, |
| "step": 259 |
| }, |
| { |
| "epoch": 1.7120879120879122, |
| "grad_norm": 0.43078261613845825, |
| "learning_rate": 0.0002087937915742794, |
| "loss": 0.5865, |
| "step": 260 |
| }, |
| { |
| "epoch": 1.7186813186813188, |
| "grad_norm": 0.39934995770454407, |
| "learning_rate": 0.0002077339246119734, |
| "loss": 0.5964, |
| "step": 261 |
| }, |
| { |
| "epoch": 1.7252747252747254, |
| "grad_norm": 0.3778096139431, |
| "learning_rate": 0.0002066740576496674, |
| "loss": 0.7449, |
| "step": 262 |
| }, |
| { |
| "epoch": 1.731868131868132, |
| "grad_norm": 0.37463709712028503, |
| "learning_rate": 0.00020561419068736142, |
| "loss": 0.5972, |
| "step": 263 |
| }, |
| { |
| "epoch": 1.7384615384615385, |
| "grad_norm": 0.41654863953590393, |
| "learning_rate": 0.00020455432372505544, |
| "loss": 0.5918, |
| "step": 264 |
| }, |
| { |
| "epoch": 1.745054945054945, |
| "grad_norm": 0.3658749759197235, |
| "learning_rate": 0.00020349445676274945, |
| "loss": 0.58, |
| "step": 265 |
| }, |
| { |
| "epoch": 1.7516483516483516, |
| "grad_norm": 0.3733176290988922, |
| "learning_rate": 0.00020243458980044346, |
| "loss": 0.744, |
| "step": 266 |
| }, |
| { |
| "epoch": 1.7582417582417582, |
| "grad_norm": 0.38651400804519653, |
| "learning_rate": 0.00020137472283813748, |
| "loss": 0.6206, |
| "step": 267 |
| }, |
| { |
| "epoch": 1.7648351648351648, |
| "grad_norm": 0.377961665391922, |
| "learning_rate": 0.0002003148558758315, |
| "loss": 0.7244, |
| "step": 268 |
| }, |
| { |
| "epoch": 1.7714285714285714, |
| "grad_norm": 0.3353257477283478, |
| "learning_rate": 0.0001992549889135255, |
| "loss": 0.6618, |
| "step": 269 |
| }, |
| { |
| "epoch": 1.778021978021978, |
| "grad_norm": 0.3616909384727478, |
| "learning_rate": 0.00019819512195121952, |
| "loss": 0.6544, |
| "step": 270 |
| }, |
| { |
| "epoch": 1.7846153846153845, |
| "grad_norm": 0.35055965185165405, |
| "learning_rate": 0.00019713525498891353, |
| "loss": 0.62, |
| "step": 271 |
| }, |
| { |
| "epoch": 1.791208791208791, |
| "grad_norm": 0.3660716414451599, |
| "learning_rate": 0.00019607538802660754, |
| "loss": 0.6236, |
| "step": 272 |
| }, |
| { |
| "epoch": 1.7978021978021979, |
| "grad_norm": 0.37623634934425354, |
| "learning_rate": 0.00019501552106430156, |
| "loss": 0.6487, |
| "step": 273 |
| }, |
| { |
| "epoch": 1.8043956043956044, |
| "grad_norm": 0.3791641592979431, |
| "learning_rate": 0.00019395565410199557, |
| "loss": 0.6505, |
| "step": 274 |
| }, |
| { |
| "epoch": 1.810989010989011, |
| "grad_norm": 0.4153837263584137, |
| "learning_rate": 0.00019289578713968959, |
| "loss": 0.6357, |
| "step": 275 |
| }, |
| { |
| "epoch": 1.8175824175824176, |
| "grad_norm": 0.3974490165710449, |
| "learning_rate": 0.0001918359201773836, |
| "loss": 0.635, |
| "step": 276 |
| }, |
| { |
| "epoch": 1.8241758241758241, |
| "grad_norm": 0.41546180844306946, |
| "learning_rate": 0.0001907760532150776, |
| "loss": 0.5926, |
| "step": 277 |
| }, |
| { |
| "epoch": 1.830769230769231, |
| "grad_norm": 0.36638110876083374, |
| "learning_rate": 0.00018971618625277163, |
| "loss": 0.6514, |
| "step": 278 |
| }, |
| { |
| "epoch": 1.8373626373626375, |
| "grad_norm": 0.3790123462677002, |
| "learning_rate": 0.00018865631929046564, |
| "loss": 0.6576, |
| "step": 279 |
| }, |
| { |
| "epoch": 1.843956043956044, |
| "grad_norm": 0.374115526676178, |
| "learning_rate": 0.00018759645232815965, |
| "loss": 0.6328, |
| "step": 280 |
| }, |
| { |
| "epoch": 1.8505494505494506, |
| "grad_norm": 0.36899662017822266, |
| "learning_rate": 0.00018653658536585367, |
| "loss": 0.6007, |
| "step": 281 |
| }, |
| { |
| "epoch": 1.8571428571428572, |
| "grad_norm": 0.36647939682006836, |
| "learning_rate": 0.00018547671840354768, |
| "loss": 0.6552, |
| "step": 282 |
| }, |
| { |
| "epoch": 1.8637362637362638, |
| "grad_norm": 0.37169116735458374, |
| "learning_rate": 0.0001844168514412417, |
| "loss": 0.591, |
| "step": 283 |
| }, |
| { |
| "epoch": 1.8703296703296703, |
| "grad_norm": 0.40471574664115906, |
| "learning_rate": 0.0001833569844789357, |
| "loss": 0.6495, |
| "step": 284 |
| }, |
| { |
| "epoch": 1.876923076923077, |
| "grad_norm": 0.3741428852081299, |
| "learning_rate": 0.0001822971175166297, |
| "loss": 0.599, |
| "step": 285 |
| }, |
| { |
| "epoch": 1.8835164835164835, |
| "grad_norm": 0.3474360406398773, |
| "learning_rate": 0.00018123725055432373, |
| "loss": 0.6384, |
| "step": 286 |
| }, |
| { |
| "epoch": 1.89010989010989, |
| "grad_norm": 0.3605997562408447, |
| "learning_rate": 0.00018017738359201775, |
| "loss": 0.6477, |
| "step": 287 |
| }, |
| { |
| "epoch": 1.8967032967032966, |
| "grad_norm": 0.34898510575294495, |
| "learning_rate": 0.00017911751662971173, |
| "loss": 0.6447, |
| "step": 288 |
| }, |
| { |
| "epoch": 1.9032967032967032, |
| "grad_norm": 0.39736175537109375, |
| "learning_rate": 0.00017805764966740578, |
| "loss": 0.7392, |
| "step": 289 |
| }, |
| { |
| "epoch": 1.9098901098901098, |
| "grad_norm": 0.3412422239780426, |
| "learning_rate": 0.0001769977827050998, |
| "loss": 0.6967, |
| "step": 290 |
| }, |
| { |
| "epoch": 1.9164835164835163, |
| "grad_norm": 0.35411298274993896, |
| "learning_rate": 0.00017593791574279378, |
| "loss": 0.701, |
| "step": 291 |
| }, |
| { |
| "epoch": 1.9230769230769231, |
| "grad_norm": 0.3787080645561218, |
| "learning_rate": 0.00017487804878048782, |
| "loss": 0.5901, |
| "step": 292 |
| }, |
| { |
| "epoch": 1.9296703296703297, |
| "grad_norm": 0.4306442439556122, |
| "learning_rate": 0.00017381818181818183, |
| "loss": 0.652, |
| "step": 293 |
| }, |
| { |
| "epoch": 1.9362637362637363, |
| "grad_norm": 0.43030846118927, |
| "learning_rate": 0.00017275831485587582, |
| "loss": 0.7147, |
| "step": 294 |
| }, |
| { |
| "epoch": 1.9428571428571428, |
| "grad_norm": 0.3750913739204407, |
| "learning_rate": 0.00017169844789356986, |
| "loss": 0.6517, |
| "step": 295 |
| }, |
| { |
| "epoch": 1.9494505494505494, |
| "grad_norm": 0.3687937557697296, |
| "learning_rate": 0.00017063858093126387, |
| "loss": 0.6481, |
| "step": 296 |
| }, |
| { |
| "epoch": 1.9560439560439562, |
| "grad_norm": 0.37245458364486694, |
| "learning_rate": 0.00016957871396895788, |
| "loss": 0.703, |
| "step": 297 |
| }, |
| { |
| "epoch": 1.9626373626373628, |
| "grad_norm": 0.35127994418144226, |
| "learning_rate": 0.0001685188470066519, |
| "loss": 0.6116, |
| "step": 298 |
| }, |
| { |
| "epoch": 1.9692307692307693, |
| "grad_norm": 0.3911903500556946, |
| "learning_rate": 0.0001674589800443459, |
| "loss": 0.6367, |
| "step": 299 |
| }, |
| { |
| "epoch": 1.975824175824176, |
| "grad_norm": 0.3940306305885315, |
| "learning_rate": 0.00016639911308203992, |
| "loss": 0.6765, |
| "step": 300 |
| }, |
| { |
| "epoch": 1.9824175824175825, |
| "grad_norm": 0.40002891421318054, |
| "learning_rate": 0.00016533924611973394, |
| "loss": 0.6013, |
| "step": 301 |
| }, |
| { |
| "epoch": 1.989010989010989, |
| "grad_norm": 0.42048192024230957, |
| "learning_rate": 0.00016427937915742795, |
| "loss": 0.6515, |
| "step": 302 |
| }, |
| { |
| "epoch": 1.9956043956043956, |
| "grad_norm": 0.3791607618331909, |
| "learning_rate": 0.00016321951219512197, |
| "loss": 0.5838, |
| "step": 303 |
| }, |
| { |
| "epoch": 2.0, |
| "grad_norm": 0.5695094466209412, |
| "learning_rate": 0.00016215964523281595, |
| "loss": 0.6227, |
| "step": 304 |
| }, |
| { |
| "epoch": 2.0065934065934066, |
| "grad_norm": 0.3927454948425293, |
| "learning_rate": 0.00016109977827051, |
| "loss": 0.5797, |
| "step": 305 |
| }, |
| { |
| "epoch": 2.013186813186813, |
| "grad_norm": 0.37926599383354187, |
| "learning_rate": 0.000160039911308204, |
| "loss": 0.5397, |
| "step": 306 |
| }, |
| { |
| "epoch": 2.0197802197802197, |
| "grad_norm": 0.3964325189590454, |
| "learning_rate": 0.000158980044345898, |
| "loss": 0.5796, |
| "step": 307 |
| }, |
| { |
| "epoch": 2.0263736263736263, |
| "grad_norm": 0.38841575384140015, |
| "learning_rate": 0.00015792017738359203, |
| "loss": 0.4925, |
| "step": 308 |
| }, |
| { |
| "epoch": 2.032967032967033, |
| "grad_norm": 0.410722941160202, |
| "learning_rate": 0.00015686031042128605, |
| "loss": 0.5662, |
| "step": 309 |
| }, |
| { |
| "epoch": 2.0395604395604394, |
| "grad_norm": 0.45456835627555847, |
| "learning_rate": 0.00015580044345898003, |
| "loss": 0.591, |
| "step": 310 |
| }, |
| { |
| "epoch": 2.046153846153846, |
| "grad_norm": 0.42649292945861816, |
| "learning_rate": 0.00015474057649667407, |
| "loss": 0.5464, |
| "step": 311 |
| }, |
| { |
| "epoch": 2.0527472527472526, |
| "grad_norm": 0.4023902714252472, |
| "learning_rate": 0.0001536807095343681, |
| "loss": 0.5437, |
| "step": 312 |
| }, |
| { |
| "epoch": 2.059340659340659, |
| "grad_norm": 0.3795386850833893, |
| "learning_rate": 0.00015262084257206207, |
| "loss": 0.4971, |
| "step": 313 |
| }, |
| { |
| "epoch": 2.065934065934066, |
| "grad_norm": 0.40091997385025024, |
| "learning_rate": 0.00015156097560975611, |
| "loss": 0.5157, |
| "step": 314 |
| }, |
| { |
| "epoch": 2.0725274725274727, |
| "grad_norm": 0.3627980649471283, |
| "learning_rate": 0.00015050110864745013, |
| "loss": 0.501, |
| "step": 315 |
| }, |
| { |
| "epoch": 2.0791208791208793, |
| "grad_norm": 0.41095221042633057, |
| "learning_rate": 0.00014944124168514411, |
| "loss": 0.5733, |
| "step": 316 |
| }, |
| { |
| "epoch": 2.085714285714286, |
| "grad_norm": 0.46346715092658997, |
| "learning_rate": 0.00014838137472283816, |
| "loss": 0.5563, |
| "step": 317 |
| }, |
| { |
| "epoch": 2.0923076923076924, |
| "grad_norm": 0.4056820571422577, |
| "learning_rate": 0.00014732150776053217, |
| "loss": 0.5715, |
| "step": 318 |
| }, |
| { |
| "epoch": 2.098901098901099, |
| "grad_norm": 0.4651278257369995, |
| "learning_rate": 0.00014626164079822616, |
| "loss": 0.4923, |
| "step": 319 |
| }, |
| { |
| "epoch": 2.1054945054945056, |
| "grad_norm": 0.4463025629520416, |
| "learning_rate": 0.0001452017738359202, |
| "loss": 0.5858, |
| "step": 320 |
| }, |
| { |
| "epoch": 2.112087912087912, |
| "grad_norm": 0.44867852330207825, |
| "learning_rate": 0.0001441419068736142, |
| "loss": 0.558, |
| "step": 321 |
| }, |
| { |
| "epoch": 2.1186813186813187, |
| "grad_norm": 0.41333332657814026, |
| "learning_rate": 0.0001430820399113082, |
| "loss": 0.5505, |
| "step": 322 |
| }, |
| { |
| "epoch": 2.1252747252747253, |
| "grad_norm": 0.4120679795742035, |
| "learning_rate": 0.00014202217294900224, |
| "loss": 0.5335, |
| "step": 323 |
| }, |
| { |
| "epoch": 2.131868131868132, |
| "grad_norm": 0.4014519453048706, |
| "learning_rate": 0.00014096230598669625, |
| "loss": 0.522, |
| "step": 324 |
| }, |
| { |
| "epoch": 2.1384615384615384, |
| "grad_norm": 0.4796731472015381, |
| "learning_rate": 0.00013990243902439024, |
| "loss": 0.5727, |
| "step": 325 |
| }, |
| { |
| "epoch": 2.145054945054945, |
| "grad_norm": 0.44720908999443054, |
| "learning_rate": 0.00013884257206208425, |
| "loss": 0.5739, |
| "step": 326 |
| }, |
| { |
| "epoch": 2.1516483516483516, |
| "grad_norm": 0.49437999725341797, |
| "learning_rate": 0.0001377827050997783, |
| "loss": 0.567, |
| "step": 327 |
| }, |
| { |
| "epoch": 2.158241758241758, |
| "grad_norm": 0.45990318059921265, |
| "learning_rate": 0.00013672283813747228, |
| "loss": 0.5099, |
| "step": 328 |
| }, |
| { |
| "epoch": 2.1648351648351647, |
| "grad_norm": 0.48705539107322693, |
| "learning_rate": 0.0001356629711751663, |
| "loss": 0.5058, |
| "step": 329 |
| }, |
| { |
| "epoch": 2.1714285714285713, |
| "grad_norm": 0.42516469955444336, |
| "learning_rate": 0.00013460310421286033, |
| "loss": 0.5034, |
| "step": 330 |
| }, |
| { |
| "epoch": 2.178021978021978, |
| "grad_norm": 0.4437274932861328, |
| "learning_rate": 0.00013354323725055432, |
| "loss": 0.538, |
| "step": 331 |
| }, |
| { |
| "epoch": 2.184615384615385, |
| "grad_norm": 0.48122531175613403, |
| "learning_rate": 0.00013248337028824833, |
| "loss": 0.5472, |
| "step": 332 |
| }, |
| { |
| "epoch": 2.1912087912087914, |
| "grad_norm": 0.5141119956970215, |
| "learning_rate": 0.00013142350332594237, |
| "loss": 0.5172, |
| "step": 333 |
| }, |
| { |
| "epoch": 2.197802197802198, |
| "grad_norm": 0.48253554105758667, |
| "learning_rate": 0.00013036363636363636, |
| "loss": 0.5747, |
| "step": 334 |
| }, |
| { |
| "epoch": 2.2043956043956046, |
| "grad_norm": 0.5002180337905884, |
| "learning_rate": 0.00012930376940133037, |
| "loss": 0.5413, |
| "step": 335 |
| }, |
| { |
| "epoch": 2.210989010989011, |
| "grad_norm": 0.5182358622550964, |
| "learning_rate": 0.0001282439024390244, |
| "loss": 0.5553, |
| "step": 336 |
| }, |
| { |
| "epoch": 2.2175824175824177, |
| "grad_norm": 0.49602043628692627, |
| "learning_rate": 0.0001271840354767184, |
| "loss": 0.5302, |
| "step": 337 |
| }, |
| { |
| "epoch": 2.2241758241758243, |
| "grad_norm": 0.4516575038433075, |
| "learning_rate": 0.0001261241685144124, |
| "loss": 0.5145, |
| "step": 338 |
| }, |
| { |
| "epoch": 2.230769230769231, |
| "grad_norm": 0.5268491506576538, |
| "learning_rate": 0.00012506430155210645, |
| "loss": 0.6026, |
| "step": 339 |
| }, |
| { |
| "epoch": 2.2373626373626374, |
| "grad_norm": 0.5656492710113525, |
| "learning_rate": 0.00012400443458980044, |
| "loss": 0.6146, |
| "step": 340 |
| }, |
| { |
| "epoch": 2.243956043956044, |
| "grad_norm": 0.4348342716693878, |
| "learning_rate": 0.00012294456762749445, |
| "loss": 0.5895, |
| "step": 341 |
| }, |
| { |
| "epoch": 2.2505494505494505, |
| "grad_norm": 0.4370437562465668, |
| "learning_rate": 0.00012188470066518848, |
| "loss": 0.5207, |
| "step": 342 |
| }, |
| { |
| "epoch": 2.257142857142857, |
| "grad_norm": 0.49735766649246216, |
| "learning_rate": 0.00012082483370288248, |
| "loss": 0.5473, |
| "step": 343 |
| }, |
| { |
| "epoch": 2.2637362637362637, |
| "grad_norm": 0.45202648639678955, |
| "learning_rate": 0.0001197649667405765, |
| "loss": 0.5432, |
| "step": 344 |
| }, |
| { |
| "epoch": 2.2703296703296703, |
| "grad_norm": 0.44890251755714417, |
| "learning_rate": 0.00011870509977827052, |
| "loss": 0.498, |
| "step": 345 |
| }, |
| { |
| "epoch": 2.276923076923077, |
| "grad_norm": 0.4461687207221985, |
| "learning_rate": 0.00011764523281596453, |
| "loss": 0.4767, |
| "step": 346 |
| }, |
| { |
| "epoch": 2.2835164835164834, |
| "grad_norm": 0.48121407628059387, |
| "learning_rate": 0.00011658536585365853, |
| "loss": 0.5976, |
| "step": 347 |
| }, |
| { |
| "epoch": 2.29010989010989, |
| "grad_norm": 0.4563685953617096, |
| "learning_rate": 0.00011552549889135256, |
| "loss": 0.5828, |
| "step": 348 |
| }, |
| { |
| "epoch": 2.2967032967032965, |
| "grad_norm": 0.5286670327186584, |
| "learning_rate": 0.00011446563192904658, |
| "loss": 0.5899, |
| "step": 349 |
| }, |
| { |
| "epoch": 2.303296703296703, |
| "grad_norm": 0.49744054675102234, |
| "learning_rate": 0.00011340576496674058, |
| "loss": 0.5124, |
| "step": 350 |
| }, |
| { |
| "epoch": 2.3098901098901097, |
| "grad_norm": 0.46971404552459717, |
| "learning_rate": 0.0001123458980044346, |
| "loss": 0.5052, |
| "step": 351 |
| }, |
| { |
| "epoch": 2.3164835164835162, |
| "grad_norm": 0.44935399293899536, |
| "learning_rate": 0.0001112860310421286, |
| "loss": 0.5604, |
| "step": 352 |
| }, |
| { |
| "epoch": 2.3230769230769233, |
| "grad_norm": 0.4866276681423187, |
| "learning_rate": 0.00011022616407982262, |
| "loss": 0.53, |
| "step": 353 |
| }, |
| { |
| "epoch": 2.32967032967033, |
| "grad_norm": 0.4510449171066284, |
| "learning_rate": 0.00010916629711751664, |
| "loss": 0.535, |
| "step": 354 |
| }, |
| { |
| "epoch": 2.3362637362637364, |
| "grad_norm": 0.5384944081306458, |
| "learning_rate": 0.00010810643015521064, |
| "loss": 0.5308, |
| "step": 355 |
| }, |
| { |
| "epoch": 2.342857142857143, |
| "grad_norm": 0.5352367162704468, |
| "learning_rate": 0.00010704656319290466, |
| "loss": 0.5214, |
| "step": 356 |
| }, |
| { |
| "epoch": 2.3494505494505495, |
| "grad_norm": 0.4872523248195648, |
| "learning_rate": 0.00010598669623059868, |
| "loss": 0.5396, |
| "step": 357 |
| }, |
| { |
| "epoch": 2.356043956043956, |
| "grad_norm": 0.4921906888484955, |
| "learning_rate": 0.00010492682926829268, |
| "loss": 0.5246, |
| "step": 358 |
| }, |
| { |
| "epoch": 2.3626373626373627, |
| "grad_norm": 0.5419618487358093, |
| "learning_rate": 0.0001038669623059867, |
| "loss": 0.5967, |
| "step": 359 |
| }, |
| { |
| "epoch": 2.3692307692307693, |
| "grad_norm": 0.5435441136360168, |
| "learning_rate": 0.00010280709534368071, |
| "loss": 0.5464, |
| "step": 360 |
| }, |
| { |
| "epoch": 2.375824175824176, |
| "grad_norm": 0.5189798474311829, |
| "learning_rate": 0.00010174722838137472, |
| "loss": 0.5533, |
| "step": 361 |
| }, |
| { |
| "epoch": 2.3824175824175824, |
| "grad_norm": 0.5280422568321228, |
| "learning_rate": 0.00010068736141906874, |
| "loss": 0.536, |
| "step": 362 |
| }, |
| { |
| "epoch": 2.389010989010989, |
| "grad_norm": 0.5015032887458801, |
| "learning_rate": 9.962749445676275e-05, |
| "loss": 0.5259, |
| "step": 363 |
| }, |
| { |
| "epoch": 2.3956043956043955, |
| "grad_norm": 0.4896168112754822, |
| "learning_rate": 9.856762749445677e-05, |
| "loss": 0.546, |
| "step": 364 |
| }, |
| { |
| "epoch": 2.402197802197802, |
| "grad_norm": 0.5274070501327515, |
| "learning_rate": 9.750776053215078e-05, |
| "loss": 0.5342, |
| "step": 365 |
| }, |
| { |
| "epoch": 2.4087912087912087, |
| "grad_norm": 0.47004714608192444, |
| "learning_rate": 9.644789356984479e-05, |
| "loss": 0.5574, |
| "step": 366 |
| }, |
| { |
| "epoch": 2.4153846153846152, |
| "grad_norm": 0.4602804481983185, |
| "learning_rate": 9.53880266075388e-05, |
| "loss": 0.5335, |
| "step": 367 |
| }, |
| { |
| "epoch": 2.421978021978022, |
| "grad_norm": 0.5090558528900146, |
| "learning_rate": 9.432815964523282e-05, |
| "loss": 0.5146, |
| "step": 368 |
| }, |
| { |
| "epoch": 2.4285714285714284, |
| "grad_norm": 0.46477729082107544, |
| "learning_rate": 9.326829268292683e-05, |
| "loss": 0.5094, |
| "step": 369 |
| }, |
| { |
| "epoch": 2.4351648351648354, |
| "grad_norm": 0.4689581096172333, |
| "learning_rate": 9.220842572062085e-05, |
| "loss": 0.4868, |
| "step": 370 |
| }, |
| { |
| "epoch": 2.441758241758242, |
| "grad_norm": 0.5514542460441589, |
| "learning_rate": 9.114855875831485e-05, |
| "loss": 0.5778, |
| "step": 371 |
| }, |
| { |
| "epoch": 2.4483516483516485, |
| "grad_norm": 0.538215696811676, |
| "learning_rate": 9.008869179600887e-05, |
| "loss": 0.5387, |
| "step": 372 |
| }, |
| { |
| "epoch": 2.454945054945055, |
| "grad_norm": 0.5533526539802551, |
| "learning_rate": 8.902882483370289e-05, |
| "loss": 0.5511, |
| "step": 373 |
| }, |
| { |
| "epoch": 2.4615384615384617, |
| "grad_norm": 0.5638222098350525, |
| "learning_rate": 8.796895787139689e-05, |
| "loss": 0.5237, |
| "step": 374 |
| }, |
| { |
| "epoch": 2.4681318681318682, |
| "grad_norm": 0.5352548360824585, |
| "learning_rate": 8.690909090909091e-05, |
| "loss": 0.537, |
| "step": 375 |
| } |
| ], |
| "logging_steps": 1, |
| "max_steps": 456, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 3, |
| "save_steps": 15, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 1.716463320435118e+17, |
| "train_batch_size": 22, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
