| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9989327641408752, | |
| "eval_steps": 500, | |
| "global_step": 819, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0012196981247141333, | |
| "grad_norm": 0.11007858067750931, | |
| "learning_rate": 7.317073170731708e-07, | |
| "loss": 1.3748, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.0024393962494282666, | |
| "grad_norm": 0.11017652601003647, | |
| "learning_rate": 1.4634146341463416e-06, | |
| "loss": 1.3711, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.0036590943741423997, | |
| "grad_norm": 0.11269191652536392, | |
| "learning_rate": 2.195121951219512e-06, | |
| "loss": 1.3847, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.004878792498856533, | |
| "grad_norm": 0.11109239608049393, | |
| "learning_rate": 2.9268292682926833e-06, | |
| "loss": 1.3495, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.006098490623570666, | |
| "grad_norm": 0.08448194712400436, | |
| "learning_rate": 3.6585365853658537e-06, | |
| "loss": 1.3583, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.007318188748284799, | |
| "grad_norm": 0.07260827720165253, | |
| "learning_rate": 4.390243902439024e-06, | |
| "loss": 1.3519, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.008537886872998933, | |
| "grad_norm": 0.09394502639770508, | |
| "learning_rate": 5.121951219512195e-06, | |
| "loss": 1.362, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.009757584997713066, | |
| "grad_norm": 0.10756496340036392, | |
| "learning_rate": 5.853658536585367e-06, | |
| "loss": 1.3543, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.0109772831224272, | |
| "grad_norm": 0.1778419017791748, | |
| "learning_rate": 6.585365853658537e-06, | |
| "loss": 1.352, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.012196981247141333, | |
| "grad_norm": 0.09264743328094482, | |
| "learning_rate": 7.317073170731707e-06, | |
| "loss": 1.3302, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.013416679371855466, | |
| "grad_norm": 0.05440306290984154, | |
| "learning_rate": 8.048780487804879e-06, | |
| "loss": 1.3428, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.014636377496569599, | |
| "grad_norm": 0.07098151743412018, | |
| "learning_rate": 8.780487804878048e-06, | |
| "loss": 1.3488, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.015856075621283734, | |
| "grad_norm": 0.07189884781837463, | |
| "learning_rate": 9.51219512195122e-06, | |
| "loss": 1.3465, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.017075773745997867, | |
| "grad_norm": 0.06418049335479736, | |
| "learning_rate": 1.024390243902439e-05, | |
| "loss": 1.3215, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.018295471870712, | |
| "grad_norm": 0.0515773668885231, | |
| "learning_rate": 1.097560975609756e-05, | |
| "loss": 1.3378, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.019515169995426133, | |
| "grad_norm": 0.048272013664245605, | |
| "learning_rate": 1.1707317073170733e-05, | |
| "loss": 1.3092, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.020734868120140266, | |
| "grad_norm": 0.07053564488887787, | |
| "learning_rate": 1.2439024390243903e-05, | |
| "loss": 1.307, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.0219545662448544, | |
| "grad_norm": 0.04769408702850342, | |
| "learning_rate": 1.3170731707317074e-05, | |
| "loss": 1.2981, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.023174264369568532, | |
| "grad_norm": 0.04310401529073715, | |
| "learning_rate": 1.3902439024390245e-05, | |
| "loss": 1.305, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.024393962494282665, | |
| "grad_norm": 0.047090884298086166, | |
| "learning_rate": 1.4634146341463415e-05, | |
| "loss": 1.3013, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.025613660618996798, | |
| "grad_norm": 0.04777524992823601, | |
| "learning_rate": 1.5365853658536586e-05, | |
| "loss": 1.3299, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.02683335874371093, | |
| "grad_norm": 0.037225186824798584, | |
| "learning_rate": 1.6097560975609757e-05, | |
| "loss": 1.2845, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.028053056868425064, | |
| "grad_norm": 0.042991526424884796, | |
| "learning_rate": 1.682926829268293e-05, | |
| "loss": 1.3141, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.029272754993139197, | |
| "grad_norm": 0.04716911166906357, | |
| "learning_rate": 1.7560975609756096e-05, | |
| "loss": 1.298, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.03049245311785333, | |
| "grad_norm": 0.03650735691189766, | |
| "learning_rate": 1.8292682926829268e-05, | |
| "loss": 1.3214, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03171215124256747, | |
| "grad_norm": 0.03615666925907135, | |
| "learning_rate": 1.902439024390244e-05, | |
| "loss": 1.3095, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.0329318493672816, | |
| "grad_norm": 0.04502608999609947, | |
| "learning_rate": 1.975609756097561e-05, | |
| "loss": 1.2741, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.03415154749199573, | |
| "grad_norm": 0.03916921466588974, | |
| "learning_rate": 2.048780487804878e-05, | |
| "loss": 1.2926, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.035371245616709866, | |
| "grad_norm": 0.03455301374197006, | |
| "learning_rate": 2.121951219512195e-05, | |
| "loss": 1.3066, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.036590943741424, | |
| "grad_norm": 0.040515027940273285, | |
| "learning_rate": 2.195121951219512e-05, | |
| "loss": 1.2743, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03781064186613813, | |
| "grad_norm": 0.04025571048259735, | |
| "learning_rate": 2.2682926829268295e-05, | |
| "loss": 1.2728, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.039030339990852266, | |
| "grad_norm": 0.03217988461256027, | |
| "learning_rate": 2.3414634146341466e-05, | |
| "loss": 1.2673, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.0402500381155664, | |
| "grad_norm": 0.036379508674144745, | |
| "learning_rate": 2.4146341463414638e-05, | |
| "loss": 1.2726, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.04146973624028053, | |
| "grad_norm": 0.03569620102643967, | |
| "learning_rate": 2.4878048780487805e-05, | |
| "loss": 1.2709, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.042689434364994665, | |
| "grad_norm": 0.034672535955905914, | |
| "learning_rate": 2.5609756097560977e-05, | |
| "loss": 1.2792, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.0439091324897088, | |
| "grad_norm": 0.032855115830898285, | |
| "learning_rate": 2.6341463414634148e-05, | |
| "loss": 1.2664, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.04512883061442293, | |
| "grad_norm": 0.038606300950050354, | |
| "learning_rate": 2.707317073170732e-05, | |
| "loss": 1.2779, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.046348528739137064, | |
| "grad_norm": 0.03479374945163727, | |
| "learning_rate": 2.780487804878049e-05, | |
| "loss": 1.2915, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.0475682268638512, | |
| "grad_norm": 0.033597253262996674, | |
| "learning_rate": 2.8536585365853658e-05, | |
| "loss": 1.293, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.04878792498856533, | |
| "grad_norm": 0.03426161780953407, | |
| "learning_rate": 2.926829268292683e-05, | |
| "loss": 1.2323, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.05000762311327946, | |
| "grad_norm": 0.035149700939655304, | |
| "learning_rate": 3e-05, | |
| "loss": 1.2484, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.051227321237993596, | |
| "grad_norm": 0.03139062970876694, | |
| "learning_rate": 2.9999877707105382e-05, | |
| "loss": 1.245, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.05244701936270773, | |
| "grad_norm": 0.036200787872076035, | |
| "learning_rate": 2.99995108304156e-05, | |
| "loss": 1.2582, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.05366671748742186, | |
| "grad_norm": 0.036149125546216965, | |
| "learning_rate": 2.9998899375912845e-05, | |
| "loss": 1.2596, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.054886415612135996, | |
| "grad_norm": 0.032857056707143784, | |
| "learning_rate": 2.9998043353567318e-05, | |
| "loss": 1.2699, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05610611373685013, | |
| "grad_norm": 0.03136944770812988, | |
| "learning_rate": 2.9996942777337083e-05, | |
| "loss": 1.2465, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.05732581186156426, | |
| "grad_norm": 0.03375684469938278, | |
| "learning_rate": 2.9995597665167824e-05, | |
| "loss": 1.2256, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.058545509986278395, | |
| "grad_norm": 0.03383234888315201, | |
| "learning_rate": 2.9994008038992565e-05, | |
| "loss": 1.234, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.05976520811099253, | |
| "grad_norm": 0.031171569600701332, | |
| "learning_rate": 2.9992173924731307e-05, | |
| "loss": 1.2404, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.06098490623570666, | |
| "grad_norm": 0.028516564518213272, | |
| "learning_rate": 2.999009535229059e-05, | |
| "loss": 1.2221, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.062204604360420794, | |
| "grad_norm": 0.033833008259534836, | |
| "learning_rate": 2.998777235556305e-05, | |
| "loss": 1.2688, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.06342430248513493, | |
| "grad_norm": 0.028970452025532722, | |
| "learning_rate": 2.9985204972426807e-05, | |
| "loss": 1.2191, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.06464400060984907, | |
| "grad_norm": 0.032843172550201416, | |
| "learning_rate": 2.9982393244744895e-05, | |
| "loss": 1.2558, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.0658636987345632, | |
| "grad_norm": 0.03238116577267647, | |
| "learning_rate": 2.9979337218364556e-05, | |
| "loss": 1.2639, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.06708339685927733, | |
| "grad_norm": 0.03204154595732689, | |
| "learning_rate": 2.9976036943116493e-05, | |
| "loss": 1.2531, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06830309498399147, | |
| "grad_norm": 0.029880031943321228, | |
| "learning_rate": 2.9972492472814072e-05, | |
| "loss": 1.2192, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.0695227931087056, | |
| "grad_norm": 0.030667617917060852, | |
| "learning_rate": 2.9968703865252434e-05, | |
| "loss": 1.2331, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.07074249123341973, | |
| "grad_norm": 0.032295312732458115, | |
| "learning_rate": 2.9964671182207543e-05, | |
| "loss": 1.2519, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.07196218935813387, | |
| "grad_norm": 0.032242096960544586, | |
| "learning_rate": 2.9960394489435202e-05, | |
| "loss": 1.2229, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.073181887482848, | |
| "grad_norm": 0.030505452305078506, | |
| "learning_rate": 2.9955873856669965e-05, | |
| "loss": 1.2274, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.07440158560756213, | |
| "grad_norm": 0.03479008749127388, | |
| "learning_rate": 2.9951109357623995e-05, | |
| "loss": 1.242, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.07562128373227626, | |
| "grad_norm": 0.03242231905460358, | |
| "learning_rate": 2.994610106998588e-05, | |
| "loss": 1.2352, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.0768409818569904, | |
| "grad_norm": 0.03107735514640808, | |
| "learning_rate": 2.9940849075419346e-05, | |
| "loss": 1.215, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.07806067998170453, | |
| "grad_norm": 0.03158826753497124, | |
| "learning_rate": 2.9935353459561953e-05, | |
| "loss": 1.2331, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.07928037810641866, | |
| "grad_norm": 0.03190038725733757, | |
| "learning_rate": 2.992961431202366e-05, | |
| "loss": 1.2365, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.0805000762311328, | |
| "grad_norm": 0.036105334758758545, | |
| "learning_rate": 2.99236317263854e-05, | |
| "loss": 1.2463, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.08171977435584693, | |
| "grad_norm": 0.03621091693639755, | |
| "learning_rate": 2.9917405800197534e-05, | |
| "loss": 1.2326, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.08293947248056106, | |
| "grad_norm": 0.03125251457095146, | |
| "learning_rate": 2.9910936634978266e-05, | |
| "loss": 1.2437, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.0841591706052752, | |
| "grad_norm": 0.038257889449596405, | |
| "learning_rate": 2.9904224336211998e-05, | |
| "loss": 1.2346, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.08537886872998933, | |
| "grad_norm": 0.03028116747736931, | |
| "learning_rate": 2.9897269013347573e-05, | |
| "loss": 1.2259, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.08659856685470346, | |
| "grad_norm": 0.03489763289690018, | |
| "learning_rate": 2.9890070779796544e-05, | |
| "loss": 1.2117, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.0878182649794176, | |
| "grad_norm": 0.03419747203588486, | |
| "learning_rate": 2.9882629752931283e-05, | |
| "loss": 1.1951, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.08903796310413173, | |
| "grad_norm": 0.028641505166888237, | |
| "learning_rate": 2.9874946054083088e-05, | |
| "loss": 1.2193, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.09025766122884586, | |
| "grad_norm": 0.033389799296855927, | |
| "learning_rate": 2.986701980854019e-05, | |
| "loss": 1.1966, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.09147735935356, | |
| "grad_norm": 0.03200199827551842, | |
| "learning_rate": 2.9858851145545728e-05, | |
| "loss": 1.1775, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.09269705747827413, | |
| "grad_norm": 0.030069107189774513, | |
| "learning_rate": 2.9850440198295628e-05, | |
| "loss": 1.2065, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.09391675560298826, | |
| "grad_norm": 0.041555605828762054, | |
| "learning_rate": 2.9841787103936432e-05, | |
| "loss": 1.2166, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.0951364537277024, | |
| "grad_norm": 0.02904364839196205, | |
| "learning_rate": 2.9832892003563073e-05, | |
| "loss": 1.2154, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.09635615185241653, | |
| "grad_norm": 0.04562120884656906, | |
| "learning_rate": 2.9823755042216557e-05, | |
| "loss": 1.1839, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.09757584997713066, | |
| "grad_norm": 0.04165040701627731, | |
| "learning_rate": 2.9814376368881606e-05, | |
| "loss": 1.2264, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.0987955481018448, | |
| "grad_norm": 0.03275958076119423, | |
| "learning_rate": 2.980475613648424e-05, | |
| "loss": 1.1865, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.10001524622655893, | |
| "grad_norm": 0.03784380853176117, | |
| "learning_rate": 2.979489450188927e-05, | |
| "loss": 1.1918, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.10123494435127306, | |
| "grad_norm": 0.03354591503739357, | |
| "learning_rate": 2.978479162589774e-05, | |
| "loss": 1.2097, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.10245464247598719, | |
| "grad_norm": 0.02976911887526512, | |
| "learning_rate": 2.9774447673244303e-05, | |
| "loss": 1.1791, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.10367434060070133, | |
| "grad_norm": 0.03253774717450142, | |
| "learning_rate": 2.9763862812594557e-05, | |
| "loss": 1.1684, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.10489403872541546, | |
| "grad_norm": 0.03416801989078522, | |
| "learning_rate": 2.975303721654226e-05, | |
| "loss": 1.183, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.10611373685012959, | |
| "grad_norm": 0.0328863188624382, | |
| "learning_rate": 2.9741971061606554e-05, | |
| "loss": 1.1678, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.10733343497484372, | |
| "grad_norm": 0.032211653888225555, | |
| "learning_rate": 2.9730664528229043e-05, | |
| "loss": 1.1882, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.10855313309955786, | |
| "grad_norm": 0.028323723003268242, | |
| "learning_rate": 2.971911780077089e-05, | |
| "loss": 1.1795, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.10977283122427199, | |
| "grad_norm": 0.031288620084524155, | |
| "learning_rate": 2.9707331067509794e-05, | |
| "loss": 1.2067, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.11099252934898612, | |
| "grad_norm": 0.030214037746191025, | |
| "learning_rate": 2.9695304520636913e-05, | |
| "loss": 1.1868, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.11221222747370026, | |
| "grad_norm": 0.031061869114637375, | |
| "learning_rate": 2.968303835625375e-05, | |
| "loss": 1.1628, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.11343192559841439, | |
| "grad_norm": 0.030998948961496353, | |
| "learning_rate": 2.967053277436894e-05, | |
| "loss": 1.1575, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.11465162372312852, | |
| "grad_norm": 0.029876932501792908, | |
| "learning_rate": 2.9657787978894985e-05, | |
| "loss": 1.1776, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.11587132184784266, | |
| "grad_norm": 0.03386124595999718, | |
| "learning_rate": 2.9644804177644942e-05, | |
| "loss": 1.1819, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.11709101997255679, | |
| "grad_norm": 0.02843078225851059, | |
| "learning_rate": 2.9631581582329033e-05, | |
| "loss": 1.1868, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.11831071809727092, | |
| "grad_norm": 0.030458344146609306, | |
| "learning_rate": 2.9618120408551185e-05, | |
| "loss": 1.1919, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.11953041622198506, | |
| "grad_norm": 0.030750496312975883, | |
| "learning_rate": 2.9604420875805517e-05, | |
| "loss": 1.1704, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.12075011434669919, | |
| "grad_norm": 0.030595768243074417, | |
| "learning_rate": 2.9590483207472766e-05, | |
| "loss": 1.1785, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.12196981247141332, | |
| "grad_norm": 0.03022405132651329, | |
| "learning_rate": 2.957630763081664e-05, | |
| "loss": 1.1759, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.12318951059612746, | |
| "grad_norm": 0.02857999876141548, | |
| "learning_rate": 2.9561894376980113e-05, | |
| "loss": 1.1738, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.12440920872084159, | |
| "grad_norm": 0.03123229742050171, | |
| "learning_rate": 2.954724368098165e-05, | |
| "loss": 1.1622, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.12562890684555572, | |
| "grad_norm": 0.028892606496810913, | |
| "learning_rate": 2.953235578171139e-05, | |
| "loss": 1.1929, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.12684860497026987, | |
| "grad_norm": 0.03238154202699661, | |
| "learning_rate": 2.9517230921927246e-05, | |
| "loss": 1.1829, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.128068303094984, | |
| "grad_norm": 0.02912278287112713, | |
| "learning_rate": 2.950186934825093e-05, | |
| "loss": 1.1829, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.12928800121969813, | |
| "grad_norm": 0.029590461403131485, | |
| "learning_rate": 2.9486271311163946e-05, | |
| "loss": 1.1486, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.13050769934441225, | |
| "grad_norm": 0.036476556211709976, | |
| "learning_rate": 2.9470437065003514e-05, | |
| "loss": 1.1603, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.1317273974691264, | |
| "grad_norm": 0.034086208790540695, | |
| "learning_rate": 2.9454366867958406e-05, | |
| "loss": 1.1665, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.13294709559384052, | |
| "grad_norm": 0.02712857723236084, | |
| "learning_rate": 2.9438060982064747e-05, | |
| "loss": 1.1681, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.13416679371855467, | |
| "grad_norm": 0.036111053079366684, | |
| "learning_rate": 2.942151967320173e-05, | |
| "loss": 1.1765, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.13538649184326879, | |
| "grad_norm": 0.034373193979263306, | |
| "learning_rate": 2.9404743211087293e-05, | |
| "loss": 1.1502, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.13660618996798293, | |
| "grad_norm": 0.02697036974132061, | |
| "learning_rate": 2.938773186927372e-05, | |
| "loss": 1.1531, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.13782588809269705, | |
| "grad_norm": 0.03094477578997612, | |
| "learning_rate": 2.9370485925143173e-05, | |
| "loss": 1.1705, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.1390455862174112, | |
| "grad_norm": 0.02840457856655121, | |
| "learning_rate": 2.9353005659903185e-05, | |
| "loss": 1.1765, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.14026528434212532, | |
| "grad_norm": 0.028384346514940262, | |
| "learning_rate": 2.9335291358582042e-05, | |
| "loss": 1.1684, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.14148498246683947, | |
| "grad_norm": 0.036891188472509384, | |
| "learning_rate": 2.9317343310024175e-05, | |
| "loss": 1.1627, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.14270468059155358, | |
| "grad_norm": 0.0326014943420887, | |
| "learning_rate": 2.9299161806885424e-05, | |
| "loss": 1.154, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.14392437871626773, | |
| "grad_norm": 0.0276651568710804, | |
| "learning_rate": 2.928074714562828e-05, | |
| "loss": 1.1755, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.14514407684098185, | |
| "grad_norm": 0.03405042737722397, | |
| "learning_rate": 2.926209962651703e-05, | |
| "loss": 1.1716, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.146363774965696, | |
| "grad_norm": 0.03377283364534378, | |
| "learning_rate": 2.9243219553612894e-05, | |
| "loss": 1.1514, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.14758347309041012, | |
| "grad_norm": 0.03034239448606968, | |
| "learning_rate": 2.9224107234769038e-05, | |
| "loss": 1.1789, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.14880317121512426, | |
| "grad_norm": 0.0301741361618042, | |
| "learning_rate": 2.920476298162557e-05, | |
| "loss": 1.1482, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.15002286933983838, | |
| "grad_norm": 0.0347026027739048, | |
| "learning_rate": 2.9185187109604445e-05, | |
| "loss": 1.1462, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.15124256746455253, | |
| "grad_norm": 0.030086705461144447, | |
| "learning_rate": 2.9165379937904346e-05, | |
| "loss": 1.1395, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.15246226558926665, | |
| "grad_norm": 0.03938613086938858, | |
| "learning_rate": 2.914534178949545e-05, | |
| "loss": 1.1499, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.1536819637139808, | |
| "grad_norm": 0.03637115657329559, | |
| "learning_rate": 2.9125072991114177e-05, | |
| "loss": 1.1522, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.15490166183869492, | |
| "grad_norm": 0.029471632093191147, | |
| "learning_rate": 2.910457387325787e-05, | |
| "loss": 1.1482, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.15612135996340906, | |
| "grad_norm": 0.03042975813150406, | |
| "learning_rate": 2.9083844770179383e-05, | |
| "loss": 1.156, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.15734105808812318, | |
| "grad_norm": 0.028736678883433342, | |
| "learning_rate": 2.906288601988166e-05, | |
| "loss": 1.1481, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.15856075621283733, | |
| "grad_norm": 0.03154094144701958, | |
| "learning_rate": 2.904169796411219e-05, | |
| "loss": 1.1755, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.15978045433755145, | |
| "grad_norm": 0.03382111340761185, | |
| "learning_rate": 2.9020280948357474e-05, | |
| "loss": 1.1416, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.1610001524622656, | |
| "grad_norm": 0.02835703268647194, | |
| "learning_rate": 2.899863532183735e-05, | |
| "loss": 1.1601, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.1622198505869797, | |
| "grad_norm": 0.029106829315423965, | |
| "learning_rate": 2.8976761437499334e-05, | |
| "loss": 1.1536, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.16343954871169386, | |
| "grad_norm": 0.03881795331835747, | |
| "learning_rate": 2.8954659652012837e-05, | |
| "loss": 1.1558, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.16465924683640798, | |
| "grad_norm": 0.03499700874090195, | |
| "learning_rate": 2.8932330325763375e-05, | |
| "loss": 1.1367, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.16587894496112213, | |
| "grad_norm": 0.0308191180229187, | |
| "learning_rate": 2.890977382284667e-05, | |
| "loss": 1.1431, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.16709864308583625, | |
| "grad_norm": 0.04107924550771713, | |
| "learning_rate": 2.888699051106272e-05, | |
| "loss": 1.1581, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.1683183412105504, | |
| "grad_norm": 0.03705988824367523, | |
| "learning_rate": 2.8863980761909828e-05, | |
| "loss": 1.1462, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.1695380393352645, | |
| "grad_norm": 0.028250785544514656, | |
| "learning_rate": 2.8840744950578493e-05, | |
| "loss": 1.163, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.17075773745997866, | |
| "grad_norm": 0.04115700721740723, | |
| "learning_rate": 2.881728345594533e-05, | |
| "loss": 1.1563, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.17197743558469278, | |
| "grad_norm": 0.0338493287563324, | |
| "learning_rate": 2.879359666056689e-05, | |
| "loss": 1.1541, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.17319713370940693, | |
| "grad_norm": 0.03268829733133316, | |
| "learning_rate": 2.8769684950673408e-05, | |
| "loss": 1.1172, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.17441683183412104, | |
| "grad_norm": 0.03336245194077492, | |
| "learning_rate": 2.8745548716162514e-05, | |
| "loss": 1.1607, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.1756365299588352, | |
| "grad_norm": 0.029322873800992966, | |
| "learning_rate": 2.872118835059287e-05, | |
| "loss": 1.1244, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.1768562280835493, | |
| "grad_norm": 0.035321250557899475, | |
| "learning_rate": 2.8696604251177758e-05, | |
| "loss": 1.1355, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.17807592620826346, | |
| "grad_norm": 0.02890695258975029, | |
| "learning_rate": 2.8671796818778603e-05, | |
| "loss": 1.1125, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.17929562433297758, | |
| "grad_norm": 0.03701947256922722, | |
| "learning_rate": 2.8646766457898434e-05, | |
| "loss": 1.1243, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.18051532245769172, | |
| "grad_norm": 0.028676442801952362, | |
| "learning_rate": 2.8621513576675293e-05, | |
| "loss": 1.1588, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.18173502058240584, | |
| "grad_norm": 0.031159276142716408, | |
| "learning_rate": 2.8596038586875562e-05, | |
| "loss": 1.1177, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.18295471870712, | |
| "grad_norm": 0.03400141000747681, | |
| "learning_rate": 2.8570341903887288e-05, | |
| "loss": 1.1381, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.1841744168318341, | |
| "grad_norm": 0.03328748047351837, | |
| "learning_rate": 2.8544423946713357e-05, | |
| "loss": 1.1301, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.18539411495654826, | |
| "grad_norm": 0.03428579494357109, | |
| "learning_rate": 2.851828513796471e-05, | |
| "loss": 1.134, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.18661381308126238, | |
| "grad_norm": 0.03697452321648598, | |
| "learning_rate": 2.849192590385342e-05, | |
| "loss": 1.1326, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.18783351120597652, | |
| "grad_norm": 0.04578408598899841, | |
| "learning_rate": 2.8465346674185767e-05, | |
| "loss": 1.1222, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.18905320933069064, | |
| "grad_norm": 0.034599971026182175, | |
| "learning_rate": 2.8438547882355205e-05, | |
| "loss": 1.1419, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.1902729074554048, | |
| "grad_norm": 0.057664740830659866, | |
| "learning_rate": 2.8411529965335305e-05, | |
| "loss": 1.1174, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.1914926055801189, | |
| "grad_norm": 0.04009220749139786, | |
| "learning_rate": 2.8384293363672643e-05, | |
| "loss": 1.1202, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.19271230370483305, | |
| "grad_norm": 0.0408347062766552, | |
| "learning_rate": 2.83568385214796e-05, | |
| "loss": 1.1404, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.19393200182954717, | |
| "grad_norm": 0.04737814515829086, | |
| "learning_rate": 2.8329165886427122e-05, | |
| "loss": 1.1072, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.19515169995426132, | |
| "grad_norm": 0.0306765865534544, | |
| "learning_rate": 2.8301275909737432e-05, | |
| "loss": 1.1218, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.19637139807897544, | |
| "grad_norm": 0.041958630084991455, | |
| "learning_rate": 2.8273169046176663e-05, | |
| "loss": 1.1263, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.1975910962036896, | |
| "grad_norm": 0.03273888677358627, | |
| "learning_rate": 2.8244845754047433e-05, | |
| "loss": 1.1334, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.19881079432840373, | |
| "grad_norm": 0.043546468019485474, | |
| "learning_rate": 2.82163064951814e-05, | |
| "loss": 1.1233, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.20003049245311785, | |
| "grad_norm": 0.03175269067287445, | |
| "learning_rate": 2.8187551734931708e-05, | |
| "loss": 1.1174, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.201250190577832, | |
| "grad_norm": 0.03650674968957901, | |
| "learning_rate": 2.81585819421654e-05, | |
| "loss": 1.1082, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.20246988870254612, | |
| "grad_norm": 0.03380831703543663, | |
| "learning_rate": 2.8129397589255787e-05, | |
| "loss": 1.1308, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.20368958682726027, | |
| "grad_norm": 0.03131740167737007, | |
| "learning_rate": 2.809999915207474e-05, | |
| "loss": 1.1377, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.20490928495197439, | |
| "grad_norm": 0.02918667159974575, | |
| "learning_rate": 2.8070387109984917e-05, | |
| "loss": 1.1392, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.20612898307668853, | |
| "grad_norm": 0.029100164771080017, | |
| "learning_rate": 2.804056194583197e-05, | |
| "loss": 1.1064, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.20734868120140265, | |
| "grad_norm": 0.03485820069909096, | |
| "learning_rate": 2.801052414593664e-05, | |
| "loss": 1.1385, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.2085683793261168, | |
| "grad_norm": 0.02814539521932602, | |
| "learning_rate": 2.7980274200086874e-05, | |
| "loss": 1.113, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.20978807745083092, | |
| "grad_norm": 0.030358154326677322, | |
| "learning_rate": 2.7949812601529793e-05, | |
| "loss": 1.111, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.21100777557554506, | |
| "grad_norm": 0.029699979349970818, | |
| "learning_rate": 2.7919139846963665e-05, | |
| "loss": 1.128, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.21222747370025918, | |
| "grad_norm": 0.0343812070786953, | |
| "learning_rate": 2.788825643652983e-05, | |
| "loss": 1.104, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.21344717182497333, | |
| "grad_norm": 0.03476931154727936, | |
| "learning_rate": 2.7857162873804495e-05, | |
| "loss": 1.0914, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.21466686994968745, | |
| "grad_norm": 0.03074513003230095, | |
| "learning_rate": 2.7825859665790575e-05, | |
| "loss": 1.117, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.2158865680744016, | |
| "grad_norm": 0.043227337300777435, | |
| "learning_rate": 2.7794347322909384e-05, | |
| "loss": 1.1066, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.21710626619911572, | |
| "grad_norm": 0.029317837208509445, | |
| "learning_rate": 2.7762626358992348e-05, | |
| "loss": 1.1128, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.21832596432382986, | |
| "grad_norm": 0.036613281816244125, | |
| "learning_rate": 2.773069729127259e-05, | |
| "loss": 1.1227, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.21954566244854398, | |
| "grad_norm": 0.030658643692731857, | |
| "learning_rate": 2.7698560640376532e-05, | |
| "loss": 1.1125, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.22076536057325813, | |
| "grad_norm": 0.039888713508844376, | |
| "learning_rate": 2.766621693031538e-05, | |
| "loss": 1.1012, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.22198505869797225, | |
| "grad_norm": 0.036936912685632706, | |
| "learning_rate": 2.7633666688476595e-05, | |
| "loss": 1.1217, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.2232047568226864, | |
| "grad_norm": 0.03285688906908035, | |
| "learning_rate": 2.7600910445615278e-05, | |
| "loss": 1.0936, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.22442445494740051, | |
| "grad_norm": 0.035456281155347824, | |
| "learning_rate": 2.7567948735845533e-05, | |
| "loss": 1.1047, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.22564415307211466, | |
| "grad_norm": 0.03127971664071083, | |
| "learning_rate": 2.7534782096631745e-05, | |
| "loss": 1.1108, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.22686385119682878, | |
| "grad_norm": 0.03643341362476349, | |
| "learning_rate": 2.7501411068779818e-05, | |
| "loss": 1.0863, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.22808354932154293, | |
| "grad_norm": 0.030930623412132263, | |
| "learning_rate": 2.7467836196428374e-05, | |
| "loss": 1.1075, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.22930324744625705, | |
| "grad_norm": 0.03585926443338394, | |
| "learning_rate": 2.743405802703985e-05, | |
| "loss": 1.1092, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.2305229455709712, | |
| "grad_norm": 0.03263048455119133, | |
| "learning_rate": 2.740007711139159e-05, | |
| "loss": 1.1117, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.2317426436956853, | |
| "grad_norm": 0.030576784163713455, | |
| "learning_rate": 2.736589400356688e-05, | |
| "loss": 1.0929, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.23296234182039946, | |
| "grad_norm": 0.03622313588857651, | |
| "learning_rate": 2.733150926094587e-05, | |
| "loss": 1.1064, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.23418203994511358, | |
| "grad_norm": 0.029499880969524384, | |
| "learning_rate": 2.7296923444196514e-05, | |
| "loss": 1.0921, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.23540173806982773, | |
| "grad_norm": 0.03302260860800743, | |
| "learning_rate": 2.7262137117265446e-05, | |
| "loss": 1.0902, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.23662143619454185, | |
| "grad_norm": 0.029528412967920303, | |
| "learning_rate": 2.722715084736874e-05, | |
| "loss": 1.1117, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.237841134319256, | |
| "grad_norm": 0.030244572088122368, | |
| "learning_rate": 2.719196520498269e-05, | |
| "loss": 1.1147, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.2390608324439701, | |
| "grad_norm": 0.03036593459546566, | |
| "learning_rate": 2.7156580763834507e-05, | |
| "loss": 1.1154, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.24028053056868426, | |
| "grad_norm": 0.04007107391953468, | |
| "learning_rate": 2.7120998100892952e-05, | |
| "loss": 1.1093, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.24150022869339838, | |
| "grad_norm": 0.031021859496831894, | |
| "learning_rate": 2.7085217796358946e-05, | |
| "loss": 1.0911, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.24271992681811252, | |
| "grad_norm": 0.03384002298116684, | |
| "learning_rate": 2.7049240433656083e-05, | |
| "loss": 1.1047, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.24393962494282664, | |
| "grad_norm": 0.035958196967840195, | |
| "learning_rate": 2.701306659942114e-05, | |
| "loss": 1.0825, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2451593230675408, | |
| "grad_norm": 0.028023509308695793, | |
| "learning_rate": 2.697669688349451e-05, | |
| "loss": 1.1065, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.2463790211922549, | |
| "grad_norm": 0.030478816479444504, | |
| "learning_rate": 2.6940131878910563e-05, | |
| "loss": 1.1112, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.24759871931696906, | |
| "grad_norm": 0.033006150275468826, | |
| "learning_rate": 2.6903372181888003e-05, | |
| "loss": 1.1128, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.24881841744168318, | |
| "grad_norm": 0.02733665332198143, | |
| "learning_rate": 2.6866418391820135e-05, | |
| "loss": 1.0898, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.2500381155663973, | |
| "grad_norm": 0.030616983771324158, | |
| "learning_rate": 2.6829271111265074e-05, | |
| "loss": 1.0953, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.25125781369111144, | |
| "grad_norm": 0.030758537352085114, | |
| "learning_rate": 2.6791930945935965e-05, | |
| "loss": 1.1027, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.25247751181582556, | |
| "grad_norm": 0.029804008081555367, | |
| "learning_rate": 2.6754398504691046e-05, | |
| "loss": 1.0804, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.25369720994053974, | |
| "grad_norm": 0.03151213005185127, | |
| "learning_rate": 2.671667439952378e-05, | |
| "loss": 1.0952, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.25491690806525386, | |
| "grad_norm": 0.0282082911580801, | |
| "learning_rate": 2.6678759245552833e-05, | |
| "loss": 1.0809, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.256136606189968, | |
| "grad_norm": 0.030734723433852196, | |
| "learning_rate": 2.6640653661012056e-05, | |
| "loss": 1.1, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.2573563043146821, | |
| "grad_norm": 0.028539832681417465, | |
| "learning_rate": 2.660235826724042e-05, | |
| "loss": 1.0993, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.25857600243939627, | |
| "grad_norm": 0.03310566022992134, | |
| "learning_rate": 2.656387368867186e-05, | |
| "loss": 1.0972, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.2597957005641104, | |
| "grad_norm": 0.026676960289478302, | |
| "learning_rate": 2.6525200552825124e-05, | |
| "loss": 1.1059, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.2610153986888245, | |
| "grad_norm": 0.02887815237045288, | |
| "learning_rate": 2.6486339490293493e-05, | |
| "loss": 1.071, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.2622350968135386, | |
| "grad_norm": 0.029342928901314735, | |
| "learning_rate": 2.644729113473455e-05, | |
| "loss": 1.0786, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.2634547949382528, | |
| "grad_norm": 0.027089102193713188, | |
| "learning_rate": 2.6408056122859823e-05, | |
| "loss": 1.091, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.2646744930629669, | |
| "grad_norm": 0.028973670676350594, | |
| "learning_rate": 2.63686350944244e-05, | |
| "loss": 1.0968, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.26589419118768104, | |
| "grad_norm": 0.030958445742726326, | |
| "learning_rate": 2.6329028692216495e-05, | |
| "loss": 1.0914, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.26711388931239516, | |
| "grad_norm": 0.02644866518676281, | |
| "learning_rate": 2.6289237562046998e-05, | |
| "loss": 1.0615, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.26833358743710933, | |
| "grad_norm": 0.029000701382756233, | |
| "learning_rate": 2.6249262352738898e-05, | |
| "loss": 1.074, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.26955328556182345, | |
| "grad_norm": 0.02928287722170353, | |
| "learning_rate": 2.620910371611674e-05, | |
| "loss": 1.0615, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.27077298368653757, | |
| "grad_norm": 0.02831627056002617, | |
| "learning_rate": 2.616876230699598e-05, | |
| "loss": 1.0847, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.2719926818112517, | |
| "grad_norm": 0.02923707105219364, | |
| "learning_rate": 2.612823878317231e-05, | |
| "loss": 1.0967, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.27321237993596587, | |
| "grad_norm": 0.027700254693627357, | |
| "learning_rate": 2.608753380541093e-05, | |
| "loss": 1.108, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.27443207806068, | |
| "grad_norm": 0.028196847066283226, | |
| "learning_rate": 2.604664803743579e-05, | |
| "loss": 1.06, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.2756517761853941, | |
| "grad_norm": 0.028219493106007576, | |
| "learning_rate": 2.6005582145918736e-05, | |
| "loss": 1.1228, | |
| "step": 226 | |
| }, | |
| { | |
| "epoch": 0.2768714743101082, | |
| "grad_norm": 0.02854679338634014, | |
| "learning_rate": 2.596433680046867e-05, | |
| "loss": 1.1146, | |
| "step": 227 | |
| }, | |
| { | |
| "epoch": 0.2780911724348224, | |
| "grad_norm": 0.028835901990532875, | |
| "learning_rate": 2.592291267362062e-05, | |
| "loss": 1.0701, | |
| "step": 228 | |
| }, | |
| { | |
| "epoch": 0.2793108705595365, | |
| "grad_norm": 0.02892499603331089, | |
| "learning_rate": 2.5881310440824765e-05, | |
| "loss": 1.0719, | |
| "step": 229 | |
| }, | |
| { | |
| "epoch": 0.28053056868425064, | |
| "grad_norm": 0.030248796567320824, | |
| "learning_rate": 2.5839530780435437e-05, | |
| "loss": 1.1006, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.28175026680896476, | |
| "grad_norm": 0.02989150956273079, | |
| "learning_rate": 2.5797574373700048e-05, | |
| "loss": 1.0779, | |
| "step": 231 | |
| }, | |
| { | |
| "epoch": 0.28296996493367893, | |
| "grad_norm": 0.026601247489452362, | |
| "learning_rate": 2.5755441904747993e-05, | |
| "loss": 1.0428, | |
| "step": 232 | |
| }, | |
| { | |
| "epoch": 0.28418966305839305, | |
| "grad_norm": 0.03424813225865364, | |
| "learning_rate": 2.5713134060579477e-05, | |
| "loss": 1.0608, | |
| "step": 233 | |
| }, | |
| { | |
| "epoch": 0.28540936118310717, | |
| "grad_norm": 0.03180866315960884, | |
| "learning_rate": 2.5670651531054337e-05, | |
| "loss": 1.1121, | |
| "step": 234 | |
| }, | |
| { | |
| "epoch": 0.2866290593078213, | |
| "grad_norm": 0.02862912230193615, | |
| "learning_rate": 2.562799500888077e-05, | |
| "loss": 1.0984, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.28784875743253546, | |
| "grad_norm": 0.033668022602796555, | |
| "learning_rate": 2.5585165189604057e-05, | |
| "loss": 1.0766, | |
| "step": 236 | |
| }, | |
| { | |
| "epoch": 0.2890684555572496, | |
| "grad_norm": 0.03088454343378544, | |
| "learning_rate": 2.5542162771595197e-05, | |
| "loss": 1.085, | |
| "step": 237 | |
| }, | |
| { | |
| "epoch": 0.2902881536819637, | |
| "grad_norm": 0.028033507987856865, | |
| "learning_rate": 2.5498988456039555e-05, | |
| "loss": 1.0664, | |
| "step": 238 | |
| }, | |
| { | |
| "epoch": 0.2915078518066778, | |
| "grad_norm": 0.02833806350827217, | |
| "learning_rate": 2.5455642946925402e-05, | |
| "loss": 1.1103, | |
| "step": 239 | |
| }, | |
| { | |
| "epoch": 0.292727549931392, | |
| "grad_norm": 0.030071113258600235, | |
| "learning_rate": 2.541212695103244e-05, | |
| "loss": 1.0781, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.2939472480561061, | |
| "grad_norm": 0.028680888935923576, | |
| "learning_rate": 2.5368441177920275e-05, | |
| "loss": 1.0794, | |
| "step": 241 | |
| }, | |
| { | |
| "epoch": 0.29516694618082023, | |
| "grad_norm": 0.02729710005223751, | |
| "learning_rate": 2.532458633991687e-05, | |
| "loss": 1.0705, | |
| "step": 242 | |
| }, | |
| { | |
| "epoch": 0.2963866443055344, | |
| "grad_norm": 0.03282102569937706, | |
| "learning_rate": 2.5280563152106896e-05, | |
| "loss": 1.1015, | |
| "step": 243 | |
| }, | |
| { | |
| "epoch": 0.2976063424302485, | |
| "grad_norm": 0.02904895320534706, | |
| "learning_rate": 2.52363723323201e-05, | |
| "loss": 1.0614, | |
| "step": 244 | |
| }, | |
| { | |
| "epoch": 0.29882604055496265, | |
| "grad_norm": 0.028720369562506676, | |
| "learning_rate": 2.5192014601119584e-05, | |
| "loss": 1.0428, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.30004573867967677, | |
| "grad_norm": 0.029231993481516838, | |
| "learning_rate": 2.5147490681790062e-05, | |
| "loss": 1.0875, | |
| "step": 246 | |
| }, | |
| { | |
| "epoch": 0.30126543680439094, | |
| "grad_norm": 0.026574166491627693, | |
| "learning_rate": 2.5102801300326057e-05, | |
| "loss": 1.0603, | |
| "step": 247 | |
| }, | |
| { | |
| "epoch": 0.30248513492910506, | |
| "grad_norm": 0.028949176892638206, | |
| "learning_rate": 2.505794718542008e-05, | |
| "loss": 1.0955, | |
| "step": 248 | |
| }, | |
| { | |
| "epoch": 0.3037048330538192, | |
| "grad_norm": 0.027590245008468628, | |
| "learning_rate": 2.5012929068450743e-05, | |
| "loss": 1.0627, | |
| "step": 249 | |
| }, | |
| { | |
| "epoch": 0.3049245311785333, | |
| "grad_norm": 0.02780095301568508, | |
| "learning_rate": 2.496774768347082e-05, | |
| "loss": 1.0935, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.3061442293032475, | |
| "grad_norm": 0.027106380090117455, | |
| "learning_rate": 2.4922403767195296e-05, | |
| "loss": 1.049, | |
| "step": 251 | |
| }, | |
| { | |
| "epoch": 0.3073639274279616, | |
| "grad_norm": 0.02786438539624214, | |
| "learning_rate": 2.4876898058989333e-05, | |
| "loss": 1.0752, | |
| "step": 252 | |
| }, | |
| { | |
| "epoch": 0.3085836255526757, | |
| "grad_norm": 0.029437003657221794, | |
| "learning_rate": 2.483123130085624e-05, | |
| "loss": 1.093, | |
| "step": 253 | |
| }, | |
| { | |
| "epoch": 0.30980332367738983, | |
| "grad_norm": 0.030303029343485832, | |
| "learning_rate": 2.4785404237425354e-05, | |
| "loss": 1.0498, | |
| "step": 254 | |
| }, | |
| { | |
| "epoch": 0.311023021802104, | |
| "grad_norm": 0.02828768640756607, | |
| "learning_rate": 2.4739417615939905e-05, | |
| "loss": 1.0699, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.3122427199268181, | |
| "grad_norm": 0.03015846386551857, | |
| "learning_rate": 2.4693272186244835e-05, | |
| "loss": 1.0729, | |
| "step": 256 | |
| }, | |
| { | |
| "epoch": 0.31346241805153224, | |
| "grad_norm": 0.02840234898030758, | |
| "learning_rate": 2.464696870077457e-05, | |
| "loss": 1.0468, | |
| "step": 257 | |
| }, | |
| { | |
| "epoch": 0.31468211617624636, | |
| "grad_norm": 0.026751426979899406, | |
| "learning_rate": 2.460050791454075e-05, | |
| "loss": 1.0592, | |
| "step": 258 | |
| }, | |
| { | |
| "epoch": 0.31590181430096054, | |
| "grad_norm": 0.028034351766109467, | |
| "learning_rate": 2.45538905851199e-05, | |
| "loss": 1.047, | |
| "step": 259 | |
| }, | |
| { | |
| "epoch": 0.31712151242567466, | |
| "grad_norm": 0.027636881917715073, | |
| "learning_rate": 2.4507117472641113e-05, | |
| "loss": 1.0771, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.3183412105503888, | |
| "grad_norm": 0.027936888858675957, | |
| "learning_rate": 2.446018933977363e-05, | |
| "loss": 1.0623, | |
| "step": 261 | |
| }, | |
| { | |
| "epoch": 0.3195609086751029, | |
| "grad_norm": 0.03554129973053932, | |
| "learning_rate": 2.441310695171442e-05, | |
| "loss": 1.0522, | |
| "step": 262 | |
| }, | |
| { | |
| "epoch": 0.32078060679981707, | |
| "grad_norm": 0.028000889346003532, | |
| "learning_rate": 2.4365871076175676e-05, | |
| "loss": 1.0391, | |
| "step": 263 | |
| }, | |
| { | |
| "epoch": 0.3220003049245312, | |
| "grad_norm": 0.031914420425891876, | |
| "learning_rate": 2.431848248337233e-05, | |
| "loss": 1.0564, | |
| "step": 264 | |
| }, | |
| { | |
| "epoch": 0.3232200030492453, | |
| "grad_norm": 0.030601803213357925, | |
| "learning_rate": 2.4270941946009466e-05, | |
| "loss": 1.057, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.3244397011739594, | |
| "grad_norm": 0.03154106065630913, | |
| "learning_rate": 2.422325023926975e-05, | |
| "loss": 1.0722, | |
| "step": 266 | |
| }, | |
| { | |
| "epoch": 0.3256593992986736, | |
| "grad_norm": 0.03405509889125824, | |
| "learning_rate": 2.4175408140800766e-05, | |
| "loss": 1.0626, | |
| "step": 267 | |
| }, | |
| { | |
| "epoch": 0.3268790974233877, | |
| "grad_norm": 0.02672364003956318, | |
| "learning_rate": 2.4127416430702328e-05, | |
| "loss": 1.0642, | |
| "step": 268 | |
| }, | |
| { | |
| "epoch": 0.32809879554810184, | |
| "grad_norm": 0.028287194669246674, | |
| "learning_rate": 2.40792758915138e-05, | |
| "loss": 1.0446, | |
| "step": 269 | |
| }, | |
| { | |
| "epoch": 0.32931849367281596, | |
| "grad_norm": 0.029666807502508163, | |
| "learning_rate": 2.40309873082013e-05, | |
| "loss": 1.0457, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.33053819179753013, | |
| "grad_norm": 0.02643524296581745, | |
| "learning_rate": 2.3982551468144907e-05, | |
| "loss": 1.0563, | |
| "step": 271 | |
| }, | |
| { | |
| "epoch": 0.33175788992224425, | |
| "grad_norm": 0.028444331139326096, | |
| "learning_rate": 2.3933969161125833e-05, | |
| "loss": 1.0687, | |
| "step": 272 | |
| }, | |
| { | |
| "epoch": 0.3329775880469584, | |
| "grad_norm": 0.029489219188690186, | |
| "learning_rate": 2.388524117931354e-05, | |
| "loss": 1.0498, | |
| "step": 273 | |
| }, | |
| { | |
| "epoch": 0.3341972861716725, | |
| "grad_norm": 0.02988230623304844, | |
| "learning_rate": 2.3836368317252816e-05, | |
| "loss": 1.0598, | |
| "step": 274 | |
| }, | |
| { | |
| "epoch": 0.33541698429638667, | |
| "grad_norm": 0.028853530064225197, | |
| "learning_rate": 2.378735137185084e-05, | |
| "loss": 1.0549, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.3366366824211008, | |
| "grad_norm": 0.031677935272455215, | |
| "learning_rate": 2.373819114236416e-05, | |
| "loss": 1.0615, | |
| "step": 276 | |
| }, | |
| { | |
| "epoch": 0.3378563805458149, | |
| "grad_norm": 0.030767912045121193, | |
| "learning_rate": 2.368888843038567e-05, | |
| "loss": 1.0685, | |
| "step": 277 | |
| }, | |
| { | |
| "epoch": 0.339076078670529, | |
| "grad_norm": 0.027159489691257477, | |
| "learning_rate": 2.3639444039831558e-05, | |
| "loss": 1.0786, | |
| "step": 278 | |
| }, | |
| { | |
| "epoch": 0.3402957767952432, | |
| "grad_norm": 0.028248904272913933, | |
| "learning_rate": 2.3589858776928185e-05, | |
| "loss": 1.0732, | |
| "step": 279 | |
| }, | |
| { | |
| "epoch": 0.3415154749199573, | |
| "grad_norm": 0.026549790054559708, | |
| "learning_rate": 2.354013345019891e-05, | |
| "loss": 1.0608, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.34273517304467144, | |
| "grad_norm": 0.03185888007283211, | |
| "learning_rate": 2.349026887045097e-05, | |
| "loss": 1.0541, | |
| "step": 281 | |
| }, | |
| { | |
| "epoch": 0.34395487116938556, | |
| "grad_norm": 0.02891666628420353, | |
| "learning_rate": 2.3440265850762187e-05, | |
| "loss": 1.0399, | |
| "step": 282 | |
| }, | |
| { | |
| "epoch": 0.34517456929409973, | |
| "grad_norm": 0.027899250388145447, | |
| "learning_rate": 2.3390125206467777e-05, | |
| "loss": 1.0364, | |
| "step": 283 | |
| }, | |
| { | |
| "epoch": 0.34639426741881385, | |
| "grad_norm": 0.029317913576960564, | |
| "learning_rate": 2.3339847755147005e-05, | |
| "loss": 1.0502, | |
| "step": 284 | |
| }, | |
| { | |
| "epoch": 0.34761396554352797, | |
| "grad_norm": 0.02855260670185089, | |
| "learning_rate": 2.3289434316609878e-05, | |
| "loss": 1.0609, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.3488336636682421, | |
| "grad_norm": 0.029661063104867935, | |
| "learning_rate": 2.3238885712883777e-05, | |
| "loss": 1.0615, | |
| "step": 286 | |
| }, | |
| { | |
| "epoch": 0.35005336179295626, | |
| "grad_norm": 0.02779245376586914, | |
| "learning_rate": 2.3188202768200036e-05, | |
| "loss": 1.0471, | |
| "step": 287 | |
| }, | |
| { | |
| "epoch": 0.3512730599176704, | |
| "grad_norm": 0.030106334015727043, | |
| "learning_rate": 2.3137386308980534e-05, | |
| "loss": 1.05, | |
| "step": 288 | |
| }, | |
| { | |
| "epoch": 0.3524927580423845, | |
| "grad_norm": 0.030160751193761826, | |
| "learning_rate": 2.308643716382418e-05, | |
| "loss": 1.0738, | |
| "step": 289 | |
| }, | |
| { | |
| "epoch": 0.3537124561670986, | |
| "grad_norm": 0.028361307457089424, | |
| "learning_rate": 2.3035356163493438e-05, | |
| "loss": 1.0654, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.3549321542918128, | |
| "grad_norm": 0.028084101155400276, | |
| "learning_rate": 2.2984144140900762e-05, | |
| "loss": 1.0599, | |
| "step": 291 | |
| }, | |
| { | |
| "epoch": 0.3561518524165269, | |
| "grad_norm": 0.02607366070151329, | |
| "learning_rate": 2.2932801931095015e-05, | |
| "loss": 1.0417, | |
| "step": 292 | |
| }, | |
| { | |
| "epoch": 0.35737155054124103, | |
| "grad_norm": 0.026301255449652672, | |
| "learning_rate": 2.288133037124786e-05, | |
| "loss": 1.0509, | |
| "step": 293 | |
| }, | |
| { | |
| "epoch": 0.35859124866595515, | |
| "grad_norm": 0.03088521771132946, | |
| "learning_rate": 2.282973030064009e-05, | |
| "loss": 1.0394, | |
| "step": 294 | |
| }, | |
| { | |
| "epoch": 0.35981094679066933, | |
| "grad_norm": 0.02662101574242115, | |
| "learning_rate": 2.2778002560647984e-05, | |
| "loss": 1.06, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.36103064491538345, | |
| "grad_norm": 0.02891872636973858, | |
| "learning_rate": 2.2726147994729548e-05, | |
| "loss": 1.0553, | |
| "step": 296 | |
| }, | |
| { | |
| "epoch": 0.36225034304009757, | |
| "grad_norm": 0.027647551149129868, | |
| "learning_rate": 2.2674167448410772e-05, | |
| "loss": 1.06, | |
| "step": 297 | |
| }, | |
| { | |
| "epoch": 0.3634700411648117, | |
| "grad_norm": 0.027648145332932472, | |
| "learning_rate": 2.2622061769271858e-05, | |
| "loss": 1.0458, | |
| "step": 298 | |
| }, | |
| { | |
| "epoch": 0.36468973928952586, | |
| "grad_norm": 0.026272861286997795, | |
| "learning_rate": 2.2569831806933378e-05, | |
| "loss": 1.0445, | |
| "step": 299 | |
| }, | |
| { | |
| "epoch": 0.36590943741424, | |
| "grad_norm": 0.025647873058915138, | |
| "learning_rate": 2.251747841304244e-05, | |
| "loss": 1.0286, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.3671291355389541, | |
| "grad_norm": 0.029211081564426422, | |
| "learning_rate": 2.246500244125879e-05, | |
| "loss": 1.0483, | |
| "step": 301 | |
| }, | |
| { | |
| "epoch": 0.3683488336636682, | |
| "grad_norm": 0.0274583138525486, | |
| "learning_rate": 2.2412404747240885e-05, | |
| "loss": 1.0414, | |
| "step": 302 | |
| }, | |
| { | |
| "epoch": 0.3695685317883824, | |
| "grad_norm": 0.03541995584964752, | |
| "learning_rate": 2.2359686188631964e-05, | |
| "loss": 1.0424, | |
| "step": 303 | |
| }, | |
| { | |
| "epoch": 0.3707882299130965, | |
| "grad_norm": 0.032834675163030624, | |
| "learning_rate": 2.230684762504605e-05, | |
| "loss": 1.0772, | |
| "step": 304 | |
| }, | |
| { | |
| "epoch": 0.37200792803781063, | |
| "grad_norm": 0.02792850323021412, | |
| "learning_rate": 2.2253889918053916e-05, | |
| "loss": 1.0414, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.37322762616252475, | |
| "grad_norm": 0.03247504308819771, | |
| "learning_rate": 2.2200813931169077e-05, | |
| "loss": 1.0624, | |
| "step": 306 | |
| }, | |
| { | |
| "epoch": 0.3744473242872389, | |
| "grad_norm": 0.029376089572906494, | |
| "learning_rate": 2.214762052983367e-05, | |
| "loss": 1.0495, | |
| "step": 307 | |
| }, | |
| { | |
| "epoch": 0.37566702241195304, | |
| "grad_norm": 0.02701037935912609, | |
| "learning_rate": 2.2094310581404367e-05, | |
| "loss": 1.0523, | |
| "step": 308 | |
| }, | |
| { | |
| "epoch": 0.37688672053666716, | |
| "grad_norm": 0.032518189400434494, | |
| "learning_rate": 2.2040884955138218e-05, | |
| "loss": 1.0414, | |
| "step": 309 | |
| }, | |
| { | |
| "epoch": 0.3781064186613813, | |
| "grad_norm": 0.027747979387640953, | |
| "learning_rate": 2.1987344522178494e-05, | |
| "loss": 1.0199, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.37932611678609546, | |
| "grad_norm": 0.029136350378394127, | |
| "learning_rate": 2.1933690155540456e-05, | |
| "loss": 1.052, | |
| "step": 311 | |
| }, | |
| { | |
| "epoch": 0.3805458149108096, | |
| "grad_norm": 0.03846918046474457, | |
| "learning_rate": 2.1879922730097156e-05, | |
| "loss": 1.0643, | |
| "step": 312 | |
| }, | |
| { | |
| "epoch": 0.3817655130355237, | |
| "grad_norm": 0.025697479024529457, | |
| "learning_rate": 2.182604312256513e-05, | |
| "loss": 1.0379, | |
| "step": 313 | |
| }, | |
| { | |
| "epoch": 0.3829852111602378, | |
| "grad_norm": 0.030951574444770813, | |
| "learning_rate": 2.1772052211490137e-05, | |
| "loss": 1.0502, | |
| "step": 314 | |
| }, | |
| { | |
| "epoch": 0.384204909284952, | |
| "grad_norm": 0.02786581963300705, | |
| "learning_rate": 2.1717950877232824e-05, | |
| "loss": 1.0489, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.3854246074096661, | |
| "grad_norm": 0.02592206373810768, | |
| "learning_rate": 2.1663740001954346e-05, | |
| "loss": 1.028, | |
| "step": 316 | |
| }, | |
| { | |
| "epoch": 0.38664430553438023, | |
| "grad_norm": 0.028019361197948456, | |
| "learning_rate": 2.1609420469602027e-05, | |
| "loss": 1.0442, | |
| "step": 317 | |
| }, | |
| { | |
| "epoch": 0.38786400365909435, | |
| "grad_norm": 0.02656111679971218, | |
| "learning_rate": 2.1554993165894913e-05, | |
| "loss": 1.0259, | |
| "step": 318 | |
| }, | |
| { | |
| "epoch": 0.3890837017838085, | |
| "grad_norm": 0.026575613766908646, | |
| "learning_rate": 2.1500458978309336e-05, | |
| "loss": 1.038, | |
| "step": 319 | |
| }, | |
| { | |
| "epoch": 0.39030339990852264, | |
| "grad_norm": 0.029259830713272095, | |
| "learning_rate": 2.1445818796064454e-05, | |
| "loss": 1.0615, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.39152309803323676, | |
| "grad_norm": 0.02846096083521843, | |
| "learning_rate": 2.1391073510107732e-05, | |
| "loss": 1.0356, | |
| "step": 321 | |
| }, | |
| { | |
| "epoch": 0.3927427961579509, | |
| "grad_norm": 0.03128277510404587, | |
| "learning_rate": 2.1336224013100445e-05, | |
| "loss": 1.0476, | |
| "step": 322 | |
| }, | |
| { | |
| "epoch": 0.39396249428266505, | |
| "grad_norm": 0.0283402930945158, | |
| "learning_rate": 2.128127119940309e-05, | |
| "loss": 1.0287, | |
| "step": 323 | |
| }, | |
| { | |
| "epoch": 0.3951821924073792, | |
| "grad_norm": 0.030746258795261383, | |
| "learning_rate": 2.122621596506082e-05, | |
| "loss": 1.0468, | |
| "step": 324 | |
| }, | |
| { | |
| "epoch": 0.3964018905320933, | |
| "grad_norm": 0.027544867247343063, | |
| "learning_rate": 2.1171059207788833e-05, | |
| "loss": 1.0354, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.39762158865680747, | |
| "grad_norm": 0.02882738783955574, | |
| "learning_rate": 2.1115801826957722e-05, | |
| "loss": 1.0402, | |
| "step": 326 | |
| }, | |
| { | |
| "epoch": 0.3988412867815216, | |
| "grad_norm": 0.029325004667043686, | |
| "learning_rate": 2.106044472357884e-05, | |
| "loss": 1.04, | |
| "step": 327 | |
| }, | |
| { | |
| "epoch": 0.4000609849062357, | |
| "grad_norm": 0.028363360092043877, | |
| "learning_rate": 2.100498880028957e-05, | |
| "loss": 1.0577, | |
| "step": 328 | |
| }, | |
| { | |
| "epoch": 0.4012806830309498, | |
| "grad_norm": 0.028542932122945786, | |
| "learning_rate": 2.0949434961338623e-05, | |
| "loss": 1.0329, | |
| "step": 329 | |
| }, | |
| { | |
| "epoch": 0.402500381155664, | |
| "grad_norm": 0.028052939102053642, | |
| "learning_rate": 2.089378411257131e-05, | |
| "loss": 1.0642, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.4037200792803781, | |
| "grad_norm": 0.02711203508079052, | |
| "learning_rate": 2.0838037161414743e-05, | |
| "loss": 1.0109, | |
| "step": 331 | |
| }, | |
| { | |
| "epoch": 0.40493977740509224, | |
| "grad_norm": 0.026364503428339958, | |
| "learning_rate": 2.0782195016863063e-05, | |
| "loss": 1.0337, | |
| "step": 332 | |
| }, | |
| { | |
| "epoch": 0.40615947552980636, | |
| "grad_norm": 0.028002282604575157, | |
| "learning_rate": 2.07262585894626e-05, | |
| "loss": 1.0329, | |
| "step": 333 | |
| }, | |
| { | |
| "epoch": 0.40737917365452053, | |
| "grad_norm": 0.027146434411406517, | |
| "learning_rate": 2.0670228791297045e-05, | |
| "loss": 1.029, | |
| "step": 334 | |
| }, | |
| { | |
| "epoch": 0.40859887177923465, | |
| "grad_norm": 0.029334858059883118, | |
| "learning_rate": 2.061410653597254e-05, | |
| "loss": 1.0393, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.40981856990394877, | |
| "grad_norm": 0.026179639622569084, | |
| "learning_rate": 2.0557892738602847e-05, | |
| "loss": 1.0162, | |
| "step": 336 | |
| }, | |
| { | |
| "epoch": 0.4110382680286629, | |
| "grad_norm": 0.028885357081890106, | |
| "learning_rate": 2.0501588315794358e-05, | |
| "loss": 1.0374, | |
| "step": 337 | |
| }, | |
| { | |
| "epoch": 0.41225796615337706, | |
| "grad_norm": 0.027485452592372894, | |
| "learning_rate": 2.044519418563118e-05, | |
| "loss": 1.0517, | |
| "step": 338 | |
| }, | |
| { | |
| "epoch": 0.4134776642780912, | |
| "grad_norm": 0.02789158746600151, | |
| "learning_rate": 2.0388711267660172e-05, | |
| "loss": 1.0415, | |
| "step": 339 | |
| }, | |
| { | |
| "epoch": 0.4146973624028053, | |
| "grad_norm": 0.028159894049167633, | |
| "learning_rate": 2.0332140482875937e-05, | |
| "loss": 1.0301, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.4159170605275194, | |
| "grad_norm": 0.02805998921394348, | |
| "learning_rate": 2.0275482753705824e-05, | |
| "loss": 1.0369, | |
| "step": 341 | |
| }, | |
| { | |
| "epoch": 0.4171367586522336, | |
| "grad_norm": 0.028342561796307564, | |
| "learning_rate": 2.021873900399484e-05, | |
| "loss": 1.0372, | |
| "step": 342 | |
| }, | |
| { | |
| "epoch": 0.4183564567769477, | |
| "grad_norm": 0.02799578197300434, | |
| "learning_rate": 2.016191015899065e-05, | |
| "loss": 1.0191, | |
| "step": 343 | |
| }, | |
| { | |
| "epoch": 0.41957615490166184, | |
| "grad_norm": 0.028251998126506805, | |
| "learning_rate": 2.0104997145328445e-05, | |
| "loss": 1.0297, | |
| "step": 344 | |
| }, | |
| { | |
| "epoch": 0.42079585302637595, | |
| "grad_norm": 0.029293730854988098, | |
| "learning_rate": 2.004800089101586e-05, | |
| "loss": 1.0218, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.42201555115109013, | |
| "grad_norm": 0.026963412761688232, | |
| "learning_rate": 1.99909223254178e-05, | |
| "loss": 1.0227, | |
| "step": 346 | |
| }, | |
| { | |
| "epoch": 0.42323524927580425, | |
| "grad_norm": 0.02611950971186161, | |
| "learning_rate": 1.993376237924134e-05, | |
| "loss": 1.0245, | |
| "step": 347 | |
| }, | |
| { | |
| "epoch": 0.42445494740051837, | |
| "grad_norm": 0.027899425476789474, | |
| "learning_rate": 1.987652198452052e-05, | |
| "loss": 1.0447, | |
| "step": 348 | |
| }, | |
| { | |
| "epoch": 0.4256746455252325, | |
| "grad_norm": 0.027212733402848244, | |
| "learning_rate": 1.9819202074601147e-05, | |
| "loss": 1.0241, | |
| "step": 349 | |
| }, | |
| { | |
| "epoch": 0.42689434364994666, | |
| "grad_norm": 0.02669115737080574, | |
| "learning_rate": 1.9761803584125578e-05, | |
| "loss": 1.0205, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.4281140417746608, | |
| "grad_norm": 0.026474419981241226, | |
| "learning_rate": 1.970432744901749e-05, | |
| "loss": 1.0539, | |
| "step": 351 | |
| }, | |
| { | |
| "epoch": 0.4293337398993749, | |
| "grad_norm": 0.029460536316037178, | |
| "learning_rate": 1.96467746064666e-05, | |
| "loss": 0.9922, | |
| "step": 352 | |
| }, | |
| { | |
| "epoch": 0.430553438024089, | |
| "grad_norm": 0.02762787789106369, | |
| "learning_rate": 1.9589145994913412e-05, | |
| "loss": 1.0476, | |
| "step": 353 | |
| }, | |
| { | |
| "epoch": 0.4317731361488032, | |
| "grad_norm": 0.026386640965938568, | |
| "learning_rate": 1.953144255403389e-05, | |
| "loss": 1.0159, | |
| "step": 354 | |
| }, | |
| { | |
| "epoch": 0.4329928342735173, | |
| "grad_norm": 0.028949646279215813, | |
| "learning_rate": 1.9473665224724133e-05, | |
| "loss": 1.048, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.43421253239823143, | |
| "grad_norm": 0.026307718828320503, | |
| "learning_rate": 1.9415814949085055e-05, | |
| "loss": 0.9858, | |
| "step": 356 | |
| }, | |
| { | |
| "epoch": 0.43543223052294555, | |
| "grad_norm": 0.03090105764567852, | |
| "learning_rate": 1.9357892670407016e-05, | |
| "loss": 1.0156, | |
| "step": 357 | |
| }, | |
| { | |
| "epoch": 0.4366519286476597, | |
| "grad_norm": 0.027751069515943527, | |
| "learning_rate": 1.9299899333154435e-05, | |
| "loss": 1.0232, | |
| "step": 358 | |
| }, | |
| { | |
| "epoch": 0.43787162677237385, | |
| "grad_norm": 0.028816018253564835, | |
| "learning_rate": 1.924183588295038e-05, | |
| "loss": 1.0184, | |
| "step": 359 | |
| }, | |
| { | |
| "epoch": 0.43909132489708796, | |
| "grad_norm": 0.02776290290057659, | |
| "learning_rate": 1.9183703266561176e-05, | |
| "loss": 1.0075, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.4403110230218021, | |
| "grad_norm": 0.026354895904660225, | |
| "learning_rate": 1.9125502431880945e-05, | |
| "loss": 1.0145, | |
| "step": 361 | |
| }, | |
| { | |
| "epoch": 0.44153072114651626, | |
| "grad_norm": 0.0271601639688015, | |
| "learning_rate": 1.9067234327916163e-05, | |
| "loss": 1.0, | |
| "step": 362 | |
| }, | |
| { | |
| "epoch": 0.4427504192712304, | |
| "grad_norm": 0.02735433541238308, | |
| "learning_rate": 1.9008899904770173e-05, | |
| "loss": 1.0455, | |
| "step": 363 | |
| }, | |
| { | |
| "epoch": 0.4439701173959445, | |
| "grad_norm": 0.026855722069740295, | |
| "learning_rate": 1.8950500113627702e-05, | |
| "loss": 1.0231, | |
| "step": 364 | |
| }, | |
| { | |
| "epoch": 0.4451898155206586, | |
| "grad_norm": 0.028405100107192993, | |
| "learning_rate": 1.8892035906739358e-05, | |
| "loss": 1.0451, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.4464095136453728, | |
| "grad_norm": 0.0320996418595314, | |
| "learning_rate": 1.883350823740608e-05, | |
| "loss": 1.0261, | |
| "step": 366 | |
| }, | |
| { | |
| "epoch": 0.4476292117700869, | |
| "grad_norm": 0.027466582134366035, | |
| "learning_rate": 1.8774918059963615e-05, | |
| "loss": 1.0192, | |
| "step": 367 | |
| }, | |
| { | |
| "epoch": 0.44884890989480103, | |
| "grad_norm": 0.03680932894349098, | |
| "learning_rate": 1.871626632976695e-05, | |
| "loss": 1.003, | |
| "step": 368 | |
| }, | |
| { | |
| "epoch": 0.45006860801951515, | |
| "grad_norm": 0.02659592032432556, | |
| "learning_rate": 1.8657554003174734e-05, | |
| "loss": 1.044, | |
| "step": 369 | |
| }, | |
| { | |
| "epoch": 0.4512883061442293, | |
| "grad_norm": 0.031448498368263245, | |
| "learning_rate": 1.859878203753368e-05, | |
| "loss": 1.0235, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.45250800426894344, | |
| "grad_norm": 0.028744159266352654, | |
| "learning_rate": 1.8539951391162964e-05, | |
| "loss": 1.0069, | |
| "step": 371 | |
| }, | |
| { | |
| "epoch": 0.45372770239365756, | |
| "grad_norm": 0.028684042394161224, | |
| "learning_rate": 1.848106302333859e-05, | |
| "loss": 1.0199, | |
| "step": 372 | |
| }, | |
| { | |
| "epoch": 0.4549474005183717, | |
| "grad_norm": 0.02726101502776146, | |
| "learning_rate": 1.842211789427775e-05, | |
| "loss": 1.0369, | |
| "step": 373 | |
| }, | |
| { | |
| "epoch": 0.45616709864308586, | |
| "grad_norm": 0.02678980678319931, | |
| "learning_rate": 1.8363116965123178e-05, | |
| "loss": 1.0352, | |
| "step": 374 | |
| }, | |
| { | |
| "epoch": 0.4573867967678, | |
| "grad_norm": 0.028863143175840378, | |
| "learning_rate": 1.830406119792746e-05, | |
| "loss": 1.0272, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.4586064948925141, | |
| "grad_norm": 0.02574865333735943, | |
| "learning_rate": 1.8244951555637355e-05, | |
| "loss": 1.0671, | |
| "step": 376 | |
| }, | |
| { | |
| "epoch": 0.4598261930172282, | |
| "grad_norm": 0.026144713163375854, | |
| "learning_rate": 1.8185789002078098e-05, | |
| "loss": 1.0393, | |
| "step": 377 | |
| }, | |
| { | |
| "epoch": 0.4610458911419424, | |
| "grad_norm": 0.026150301098823547, | |
| "learning_rate": 1.8126574501937676e-05, | |
| "loss": 1.0182, | |
| "step": 378 | |
| }, | |
| { | |
| "epoch": 0.4622655892666565, | |
| "grad_norm": 0.028266610577702522, | |
| "learning_rate": 1.8067309020751115e-05, | |
| "loss": 1.016, | |
| "step": 379 | |
| }, | |
| { | |
| "epoch": 0.4634852873913706, | |
| "grad_norm": 0.02818026766180992, | |
| "learning_rate": 1.800799352488471e-05, | |
| "loss": 1.004, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.46470498551608475, | |
| "grad_norm": 0.03137146309018135, | |
| "learning_rate": 1.794862898152028e-05, | |
| "loss": 1.0038, | |
| "step": 381 | |
| }, | |
| { | |
| "epoch": 0.4659246836407989, | |
| "grad_norm": 0.025779929012060165, | |
| "learning_rate": 1.7889216358639413e-05, | |
| "loss": 1.0268, | |
| "step": 382 | |
| }, | |
| { | |
| "epoch": 0.46714438176551304, | |
| "grad_norm": 0.02844846062362194, | |
| "learning_rate": 1.7829756625007654e-05, | |
| "loss": 1.0336, | |
| "step": 383 | |
| }, | |
| { | |
| "epoch": 0.46836407989022716, | |
| "grad_norm": 0.026752308011054993, | |
| "learning_rate": 1.777025075015873e-05, | |
| "loss": 1.0267, | |
| "step": 384 | |
| }, | |
| { | |
| "epoch": 0.4695837780149413, | |
| "grad_norm": 0.03013698197901249, | |
| "learning_rate": 1.7710699704378735e-05, | |
| "loss": 1.0192, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.47080347613965545, | |
| "grad_norm": 0.025083104148507118, | |
| "learning_rate": 1.7651104458690298e-05, | |
| "loss": 1.0475, | |
| "step": 386 | |
| }, | |
| { | |
| "epoch": 0.47202317426436957, | |
| "grad_norm": 0.029982833191752434, | |
| "learning_rate": 1.7591465984836768e-05, | |
| "loss": 0.9808, | |
| "step": 387 | |
| }, | |
| { | |
| "epoch": 0.4732428723890837, | |
| "grad_norm": 0.02694685570895672, | |
| "learning_rate": 1.753178525526636e-05, | |
| "loss": 1.0002, | |
| "step": 388 | |
| }, | |
| { | |
| "epoch": 0.4744625705137978, | |
| "grad_norm": 0.027812331914901733, | |
| "learning_rate": 1.74720632431163e-05, | |
| "loss": 0.9852, | |
| "step": 389 | |
| }, | |
| { | |
| "epoch": 0.475682268638512, | |
| "grad_norm": 0.025453535839915276, | |
| "learning_rate": 1.741230092219695e-05, | |
| "loss": 1.0003, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.4769019667632261, | |
| "grad_norm": 0.02900811843574047, | |
| "learning_rate": 1.7352499266975933e-05, | |
| "loss": 1.0149, | |
| "step": 391 | |
| }, | |
| { | |
| "epoch": 0.4781216648879402, | |
| "grad_norm": 0.025282319635152817, | |
| "learning_rate": 1.7292659252562256e-05, | |
| "loss": 1.0163, | |
| "step": 392 | |
| }, | |
| { | |
| "epoch": 0.47934136301265434, | |
| "grad_norm": 0.026716770604252815, | |
| "learning_rate": 1.72327818546904e-05, | |
| "loss": 1.0094, | |
| "step": 393 | |
| }, | |
| { | |
| "epoch": 0.4805610611373685, | |
| "grad_norm": 0.025360623374581337, | |
| "learning_rate": 1.7172868049704397e-05, | |
| "loss": 0.9918, | |
| "step": 394 | |
| }, | |
| { | |
| "epoch": 0.48178075926208264, | |
| "grad_norm": 0.02906343713402748, | |
| "learning_rate": 1.7112918814541945e-05, | |
| "loss": 1.0166, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.48300045738679676, | |
| "grad_norm": 0.025819873437285423, | |
| "learning_rate": 1.705293512671843e-05, | |
| "loss": 1.0009, | |
| "step": 396 | |
| }, | |
| { | |
| "epoch": 0.4842201555115109, | |
| "grad_norm": 0.02552095428109169, | |
| "learning_rate": 1.6992917964311036e-05, | |
| "loss": 1.0072, | |
| "step": 397 | |
| }, | |
| { | |
| "epoch": 0.48543985363622505, | |
| "grad_norm": 0.025735564529895782, | |
| "learning_rate": 1.693286830594277e-05, | |
| "loss": 1.0395, | |
| "step": 398 | |
| }, | |
| { | |
| "epoch": 0.48665955176093917, | |
| "grad_norm": 0.03076878748834133, | |
| "learning_rate": 1.6872787130766488e-05, | |
| "loss": 0.9987, | |
| "step": 399 | |
| }, | |
| { | |
| "epoch": 0.4878792498856533, | |
| "grad_norm": 0.0290908794850111, | |
| "learning_rate": 1.6812675418448978e-05, | |
| "loss": 1.0318, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.4890989480103674, | |
| "grad_norm": 0.026456333696842194, | |
| "learning_rate": 1.6752534149154945e-05, | |
| "loss": 1.0141, | |
| "step": 401 | |
| }, | |
| { | |
| "epoch": 0.4903186461350816, | |
| "grad_norm": 0.026630204170942307, | |
| "learning_rate": 1.6692364303531046e-05, | |
| "loss": 1.0107, | |
| "step": 402 | |
| }, | |
| { | |
| "epoch": 0.4915383442597957, | |
| "grad_norm": 0.025951065123081207, | |
| "learning_rate": 1.6632166862689887e-05, | |
| "loss": 1.0171, | |
| "step": 403 | |
| }, | |
| { | |
| "epoch": 0.4927580423845098, | |
| "grad_norm": 0.02653127908706665, | |
| "learning_rate": 1.657194280819404e-05, | |
| "loss": 1.0274, | |
| "step": 404 | |
| }, | |
| { | |
| "epoch": 0.493977740509224, | |
| "grad_norm": 0.02548621967434883, | |
| "learning_rate": 1.651169312204004e-05, | |
| "loss": 0.9966, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.4951974386339381, | |
| "grad_norm": 0.024926092475652695, | |
| "learning_rate": 1.645141878664235e-05, | |
| "loss": 1.0059, | |
| "step": 406 | |
| }, | |
| { | |
| "epoch": 0.49641713675865223, | |
| "grad_norm": 0.025980288162827492, | |
| "learning_rate": 1.639112078481736e-05, | |
| "loss": 1.0295, | |
| "step": 407 | |
| }, | |
| { | |
| "epoch": 0.49763683488336635, | |
| "grad_norm": 0.025473380461335182, | |
| "learning_rate": 1.6330800099767365e-05, | |
| "loss": 1.0013, | |
| "step": 408 | |
| }, | |
| { | |
| "epoch": 0.4988565330080805, | |
| "grad_norm": 0.025365322828292847, | |
| "learning_rate": 1.6270457715064525e-05, | |
| "loss": 1.0038, | |
| "step": 409 | |
| }, | |
| { | |
| "epoch": 0.5000762311327946, | |
| "grad_norm": 0.025861822068691254, | |
| "learning_rate": 1.621009461463482e-05, | |
| "loss": 0.9942, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.5012959292575088, | |
| "grad_norm": 0.02603691816329956, | |
| "learning_rate": 1.614971178274203e-05, | |
| "loss": 1.0073, | |
| "step": 411 | |
| }, | |
| { | |
| "epoch": 0.5025156273822229, | |
| "grad_norm": 0.025546738877892494, | |
| "learning_rate": 1.608931020397165e-05, | |
| "loss": 0.9982, | |
| "step": 412 | |
| }, | |
| { | |
| "epoch": 0.503735325506937, | |
| "grad_norm": 0.02535770833492279, | |
| "learning_rate": 1.6028890863214872e-05, | |
| "loss": 1.0141, | |
| "step": 413 | |
| }, | |
| { | |
| "epoch": 0.5049550236316511, | |
| "grad_norm": 0.02483360469341278, | |
| "learning_rate": 1.5968454745652508e-05, | |
| "loss": 1.0107, | |
| "step": 414 | |
| }, | |
| { | |
| "epoch": 0.5061747217563654, | |
| "grad_norm": 0.02530517615377903, | |
| "learning_rate": 1.5908002836738925e-05, | |
| "loss": 1.0364, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.5073944198810795, | |
| "grad_norm": 0.026432055979967117, | |
| "learning_rate": 1.5847536122185975e-05, | |
| "loss": 1.0106, | |
| "step": 416 | |
| }, | |
| { | |
| "epoch": 0.5086141180057936, | |
| "grad_norm": 0.024275679141283035, | |
| "learning_rate": 1.5787055587946934e-05, | |
| "loss": 1.008, | |
| "step": 417 | |
| }, | |
| { | |
| "epoch": 0.5098338161305077, | |
| "grad_norm": 0.025925274938344955, | |
| "learning_rate": 1.572656222020042e-05, | |
| "loss": 1.0137, | |
| "step": 418 | |
| }, | |
| { | |
| "epoch": 0.5110535142552218, | |
| "grad_norm": 0.02538963407278061, | |
| "learning_rate": 1.56660570053343e-05, | |
| "loss": 1.0053, | |
| "step": 419 | |
| }, | |
| { | |
| "epoch": 0.512273212379936, | |
| "grad_norm": 0.025234252214431763, | |
| "learning_rate": 1.560554092992963e-05, | |
| "loss": 1.0188, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.5134929105046501, | |
| "grad_norm": 0.026904789730906487, | |
| "learning_rate": 1.5545014980744534e-05, | |
| "loss": 1.014, | |
| "step": 421 | |
| }, | |
| { | |
| "epoch": 0.5147126086293642, | |
| "grad_norm": 0.02695068158209324, | |
| "learning_rate": 1.548448014469816e-05, | |
| "loss": 1.0478, | |
| "step": 422 | |
| }, | |
| { | |
| "epoch": 0.5159323067540784, | |
| "grad_norm": 0.02476106956601143, | |
| "learning_rate": 1.542393740885455e-05, | |
| "loss": 1.0087, | |
| "step": 423 | |
| }, | |
| { | |
| "epoch": 0.5171520048787925, | |
| "grad_norm": 0.026810606941580772, | |
| "learning_rate": 1.5363387760406556e-05, | |
| "loss": 0.9909, | |
| "step": 424 | |
| }, | |
| { | |
| "epoch": 0.5183717030035067, | |
| "grad_norm": 0.02679542452096939, | |
| "learning_rate": 1.5302832186659752e-05, | |
| "loss": 1.0262, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.5195914011282208, | |
| "grad_norm": 0.02460315451025963, | |
| "learning_rate": 1.5242271675016324e-05, | |
| "loss": 0.9866, | |
| "step": 426 | |
| }, | |
| { | |
| "epoch": 0.5208110992529349, | |
| "grad_norm": 0.026209082454442978, | |
| "learning_rate": 1.5181707212958972e-05, | |
| "loss": 1.0173, | |
| "step": 427 | |
| }, | |
| { | |
| "epoch": 0.522030797377649, | |
| "grad_norm": 0.026854293420910835, | |
| "learning_rate": 1.5121139788034814e-05, | |
| "loss": 0.9897, | |
| "step": 428 | |
| }, | |
| { | |
| "epoch": 0.5232504955023631, | |
| "grad_norm": 0.02499808743596077, | |
| "learning_rate": 1.5060570387839277e-05, | |
| "loss": 1.0058, | |
| "step": 429 | |
| }, | |
| { | |
| "epoch": 0.5244701936270773, | |
| "grad_norm": 0.02579757198691368, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.0037, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.5256898917517915, | |
| "grad_norm": 0.027696194127202034, | |
| "learning_rate": 1.4939429612160727e-05, | |
| "loss": 1.0087, | |
| "step": 431 | |
| }, | |
| { | |
| "epoch": 0.5269095898765056, | |
| "grad_norm": 0.02441985532641411, | |
| "learning_rate": 1.4878860211965189e-05, | |
| "loss": 1.0003, | |
| "step": 432 | |
| }, | |
| { | |
| "epoch": 0.5281292880012197, | |
| "grad_norm": 0.02803633362054825, | |
| "learning_rate": 1.4818292787041032e-05, | |
| "loss": 0.9976, | |
| "step": 433 | |
| }, | |
| { | |
| "epoch": 0.5293489861259338, | |
| "grad_norm": 0.02653639204800129, | |
| "learning_rate": 1.4757728324983675e-05, | |
| "loss": 0.9898, | |
| "step": 434 | |
| }, | |
| { | |
| "epoch": 0.530568684250648, | |
| "grad_norm": 0.027481012046337128, | |
| "learning_rate": 1.4697167813340249e-05, | |
| "loss": 0.9907, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.5317883823753621, | |
| "grad_norm": 0.02567669376730919, | |
| "learning_rate": 1.4636612239593443e-05, | |
| "loss": 1.0057, | |
| "step": 436 | |
| }, | |
| { | |
| "epoch": 0.5330080805000762, | |
| "grad_norm": 0.026430847123265266, | |
| "learning_rate": 1.4576062591145451e-05, | |
| "loss": 1.0259, | |
| "step": 437 | |
| }, | |
| { | |
| "epoch": 0.5342277786247903, | |
| "grad_norm": 0.026311656460165977, | |
| "learning_rate": 1.451551985530184e-05, | |
| "loss": 0.99, | |
| "step": 438 | |
| }, | |
| { | |
| "epoch": 0.5354474767495045, | |
| "grad_norm": 0.029337909072637558, | |
| "learning_rate": 1.445498501925547e-05, | |
| "loss": 0.9983, | |
| "step": 439 | |
| }, | |
| { | |
| "epoch": 0.5366671748742187, | |
| "grad_norm": 0.024333585053682327, | |
| "learning_rate": 1.4394459070070376e-05, | |
| "loss": 1.0143, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.5378868729989328, | |
| "grad_norm": 0.026066364720463753, | |
| "learning_rate": 1.4333942994665703e-05, | |
| "loss": 0.9947, | |
| "step": 441 | |
| }, | |
| { | |
| "epoch": 0.5391065711236469, | |
| "grad_norm": 0.025209229439496994, | |
| "learning_rate": 1.427343777979958e-05, | |
| "loss": 1.0012, | |
| "step": 442 | |
| }, | |
| { | |
| "epoch": 0.540326269248361, | |
| "grad_norm": 0.025801435112953186, | |
| "learning_rate": 1.4212944412053066e-05, | |
| "loss": 1.0099, | |
| "step": 443 | |
| }, | |
| { | |
| "epoch": 0.5415459673730751, | |
| "grad_norm": 0.027363212779164314, | |
| "learning_rate": 1.415246387781403e-05, | |
| "loss": 1.0089, | |
| "step": 444 | |
| }, | |
| { | |
| "epoch": 0.5427656654977893, | |
| "grad_norm": 0.02772754244506359, | |
| "learning_rate": 1.409199716326108e-05, | |
| "loss": 1.0058, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.5439853636225034, | |
| "grad_norm": 0.023262690752744675, | |
| "learning_rate": 1.4031545254347496e-05, | |
| "loss": 0.9897, | |
| "step": 446 | |
| }, | |
| { | |
| "epoch": 0.5452050617472176, | |
| "grad_norm": 0.02593029849231243, | |
| "learning_rate": 1.3971109136785128e-05, | |
| "loss": 1.0003, | |
| "step": 447 | |
| }, | |
| { | |
| "epoch": 0.5464247598719317, | |
| "grad_norm": 0.025894835591316223, | |
| "learning_rate": 1.3910689796028354e-05, | |
| "loss": 1.0342, | |
| "step": 448 | |
| }, | |
| { | |
| "epoch": 0.5476444579966459, | |
| "grad_norm": 0.02416595071554184, | |
| "learning_rate": 1.3850288217257977e-05, | |
| "loss": 1.0191, | |
| "step": 449 | |
| }, | |
| { | |
| "epoch": 0.54886415612136, | |
| "grad_norm": 0.024584488943219185, | |
| "learning_rate": 1.3789905385365178e-05, | |
| "loss": 0.995, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.5500838542460741, | |
| "grad_norm": 0.024157658219337463, | |
| "learning_rate": 1.3729542284935478e-05, | |
| "loss": 1.0128, | |
| "step": 451 | |
| }, | |
| { | |
| "epoch": 0.5513035523707882, | |
| "grad_norm": 0.024094339460134506, | |
| "learning_rate": 1.3669199900232637e-05, | |
| "loss": 1.0122, | |
| "step": 452 | |
| }, | |
| { | |
| "epoch": 0.5525232504955023, | |
| "grad_norm": 0.025125445798039436, | |
| "learning_rate": 1.3608879215182644e-05, | |
| "loss": 0.9962, | |
| "step": 453 | |
| }, | |
| { | |
| "epoch": 0.5537429486202164, | |
| "grad_norm": 0.025818830356001854, | |
| "learning_rate": 1.3548581213357656e-05, | |
| "loss": 1.0373, | |
| "step": 454 | |
| }, | |
| { | |
| "epoch": 0.5549626467449307, | |
| "grad_norm": 0.025979438796639442, | |
| "learning_rate": 1.348830687795996e-05, | |
| "loss": 1.0053, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.5561823448696448, | |
| "grad_norm": 0.023782270029187202, | |
| "learning_rate": 1.342805719180596e-05, | |
| "loss": 0.9897, | |
| "step": 456 | |
| }, | |
| { | |
| "epoch": 0.5574020429943589, | |
| "grad_norm": 0.024968748912215233, | |
| "learning_rate": 1.336783313731012e-05, | |
| "loss": 1.0301, | |
| "step": 457 | |
| }, | |
| { | |
| "epoch": 0.558621741119073, | |
| "grad_norm": 0.024653363972902298, | |
| "learning_rate": 1.3307635696468957e-05, | |
| "loss": 0.9932, | |
| "step": 458 | |
| }, | |
| { | |
| "epoch": 0.5598414392437872, | |
| "grad_norm": 0.024653034284710884, | |
| "learning_rate": 1.3247465850845059e-05, | |
| "loss": 0.9959, | |
| "step": 459 | |
| }, | |
| { | |
| "epoch": 0.5610611373685013, | |
| "grad_norm": 0.026350928470492363, | |
| "learning_rate": 1.3187324581551021e-05, | |
| "loss": 0.9845, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.5622808354932154, | |
| "grad_norm": 0.026692520827054977, | |
| "learning_rate": 1.3127212869233515e-05, | |
| "loss": 1.0276, | |
| "step": 461 | |
| }, | |
| { | |
| "epoch": 0.5635005336179295, | |
| "grad_norm": 0.02479124441742897, | |
| "learning_rate": 1.3067131694057239e-05, | |
| "loss": 1.0034, | |
| "step": 462 | |
| }, | |
| { | |
| "epoch": 0.5647202317426437, | |
| "grad_norm": 0.024124760180711746, | |
| "learning_rate": 1.3007082035688963e-05, | |
| "loss": 1.0023, | |
| "step": 463 | |
| }, | |
| { | |
| "epoch": 0.5659399298673579, | |
| "grad_norm": 0.027967294678092003, | |
| "learning_rate": 1.2947064873281573e-05, | |
| "loss": 1.0066, | |
| "step": 464 | |
| }, | |
| { | |
| "epoch": 0.567159627992072, | |
| "grad_norm": 0.0263604074716568, | |
| "learning_rate": 1.2887081185458059e-05, | |
| "loss": 0.9904, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.5683793261167861, | |
| "grad_norm": 0.024409962818026543, | |
| "learning_rate": 1.2827131950295602e-05, | |
| "loss": 1.0121, | |
| "step": 466 | |
| }, | |
| { | |
| "epoch": 0.5695990242415002, | |
| "grad_norm": 0.025176391005516052, | |
| "learning_rate": 1.2767218145309605e-05, | |
| "loss": 1.0198, | |
| "step": 467 | |
| }, | |
| { | |
| "epoch": 0.5708187223662143, | |
| "grad_norm": 0.024721505120396614, | |
| "learning_rate": 1.2707340747437745e-05, | |
| "loss": 1.0128, | |
| "step": 468 | |
| }, | |
| { | |
| "epoch": 0.5720384204909285, | |
| "grad_norm": 0.0248569343239069, | |
| "learning_rate": 1.2647500733024071e-05, | |
| "loss": 0.9944, | |
| "step": 469 | |
| }, | |
| { | |
| "epoch": 0.5732581186156426, | |
| "grad_norm": 0.025843387469649315, | |
| "learning_rate": 1.2587699077803056e-05, | |
| "loss": 1.0012, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.5744778167403568, | |
| "grad_norm": 0.025511378422379494, | |
| "learning_rate": 1.2527936756883701e-05, | |
| "loss": 1.0057, | |
| "step": 471 | |
| }, | |
| { | |
| "epoch": 0.5756975148650709, | |
| "grad_norm": 0.025058850646018982, | |
| "learning_rate": 1.2468214744733641e-05, | |
| "loss": 1.0057, | |
| "step": 472 | |
| }, | |
| { | |
| "epoch": 0.576917212989785, | |
| "grad_norm": 0.024858184158802032, | |
| "learning_rate": 1.2408534015163231e-05, | |
| "loss": 1.0003, | |
| "step": 473 | |
| }, | |
| { | |
| "epoch": 0.5781369111144992, | |
| "grad_norm": 0.024318138137459755, | |
| "learning_rate": 1.2348895541309704e-05, | |
| "loss": 0.983, | |
| "step": 474 | |
| }, | |
| { | |
| "epoch": 0.5793566092392133, | |
| "grad_norm": 0.024911191314458847, | |
| "learning_rate": 1.2289300295621273e-05, | |
| "loss": 0.9848, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.5805763073639274, | |
| "grad_norm": 0.025810109451413155, | |
| "learning_rate": 1.2229749249841268e-05, | |
| "loss": 1.0019, | |
| "step": 476 | |
| }, | |
| { | |
| "epoch": 0.5817960054886415, | |
| "grad_norm": 0.026411840692162514, | |
| "learning_rate": 1.217024337499235e-05, | |
| "loss": 1.0023, | |
| "step": 477 | |
| }, | |
| { | |
| "epoch": 0.5830157036133556, | |
| "grad_norm": 0.02432168833911419, | |
| "learning_rate": 1.2110783641360587e-05, | |
| "loss": 0.9884, | |
| "step": 478 | |
| }, | |
| { | |
| "epoch": 0.5842354017380699, | |
| "grad_norm": 0.0252446997910738, | |
| "learning_rate": 1.2051371018479721e-05, | |
| "loss": 0.9922, | |
| "step": 479 | |
| }, | |
| { | |
| "epoch": 0.585455099862784, | |
| "grad_norm": 0.02487318031489849, | |
| "learning_rate": 1.1992006475115298e-05, | |
| "loss": 1.0194, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.5866747979874981, | |
| "grad_norm": 0.0248030424118042, | |
| "learning_rate": 1.1932690979248888e-05, | |
| "loss": 0.9774, | |
| "step": 481 | |
| }, | |
| { | |
| "epoch": 0.5878944961122122, | |
| "grad_norm": 0.02666288986802101, | |
| "learning_rate": 1.1873425498062326e-05, | |
| "loss": 1.0047, | |
| "step": 482 | |
| }, | |
| { | |
| "epoch": 0.5891141942369263, | |
| "grad_norm": 0.026000840589404106, | |
| "learning_rate": 1.1814210997921908e-05, | |
| "loss": 0.9892, | |
| "step": 483 | |
| }, | |
| { | |
| "epoch": 0.5903338923616405, | |
| "grad_norm": 0.024337952956557274, | |
| "learning_rate": 1.1755048444362647e-05, | |
| "loss": 0.9874, | |
| "step": 484 | |
| }, | |
| { | |
| "epoch": 0.5915535904863546, | |
| "grad_norm": 0.024667885154485703, | |
| "learning_rate": 1.1695938802072545e-05, | |
| "loss": 1.0012, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.5927732886110688, | |
| "grad_norm": 0.027820762246847153, | |
| "learning_rate": 1.1636883034876821e-05, | |
| "loss": 1.0141, | |
| "step": 486 | |
| }, | |
| { | |
| "epoch": 0.5939929867357829, | |
| "grad_norm": 0.025526076555252075, | |
| "learning_rate": 1.157788210572225e-05, | |
| "loss": 0.9838, | |
| "step": 487 | |
| }, | |
| { | |
| "epoch": 0.595212684860497, | |
| "grad_norm": 0.023680077865719795, | |
| "learning_rate": 1.1518936976661417e-05, | |
| "loss": 1.0049, | |
| "step": 488 | |
| }, | |
| { | |
| "epoch": 0.5964323829852112, | |
| "grad_norm": 0.02427358739078045, | |
| "learning_rate": 1.1460048608837038e-05, | |
| "loss": 1.0213, | |
| "step": 489 | |
| }, | |
| { | |
| "epoch": 0.5976520811099253, | |
| "grad_norm": 0.024500545114278793, | |
| "learning_rate": 1.1401217962466326e-05, | |
| "loss": 0.999, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.5988717792346394, | |
| "grad_norm": 0.023401236161589622, | |
| "learning_rate": 1.1342445996825269e-05, | |
| "loss": 0.9866, | |
| "step": 491 | |
| }, | |
| { | |
| "epoch": 0.6000914773593535, | |
| "grad_norm": 0.024014266207814217, | |
| "learning_rate": 1.1283733670233053e-05, | |
| "loss": 0.9879, | |
| "step": 492 | |
| }, | |
| { | |
| "epoch": 0.6013111754840677, | |
| "grad_norm": 0.02356668934226036, | |
| "learning_rate": 1.122508194003639e-05, | |
| "loss": 1.0054, | |
| "step": 493 | |
| }, | |
| { | |
| "epoch": 0.6025308736087819, | |
| "grad_norm": 0.024357525631785393, | |
| "learning_rate": 1.1166491762593923e-05, | |
| "loss": 0.9704, | |
| "step": 494 | |
| }, | |
| { | |
| "epoch": 0.603750571733496, | |
| "grad_norm": 0.023848682641983032, | |
| "learning_rate": 1.1107964093260646e-05, | |
| "loss": 0.9931, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.6049702698582101, | |
| "grad_norm": 0.023952610790729523, | |
| "learning_rate": 1.1049499886372299e-05, | |
| "loss": 0.9704, | |
| "step": 496 | |
| }, | |
| { | |
| "epoch": 0.6061899679829242, | |
| "grad_norm": 0.02463214099407196, | |
| "learning_rate": 1.099110009522983e-05, | |
| "loss": 0.9725, | |
| "step": 497 | |
| }, | |
| { | |
| "epoch": 0.6074096661076384, | |
| "grad_norm": 0.025442618876695633, | |
| "learning_rate": 1.0932765672083839e-05, | |
| "loss": 0.9876, | |
| "step": 498 | |
| }, | |
| { | |
| "epoch": 0.6086293642323525, | |
| "grad_norm": 0.02486078254878521, | |
| "learning_rate": 1.0874497568119056e-05, | |
| "loss": 0.9885, | |
| "step": 499 | |
| }, | |
| { | |
| "epoch": 0.6098490623570666, | |
| "grad_norm": 0.025211885571479797, | |
| "learning_rate": 1.0816296733438825e-05, | |
| "loss": 1.0071, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.6110687604817807, | |
| "grad_norm": 0.0240833330899477, | |
| "learning_rate": 1.0758164117049624e-05, | |
| "loss": 1.0035, | |
| "step": 501 | |
| }, | |
| { | |
| "epoch": 0.612288458606495, | |
| "grad_norm": 0.024268394336104393, | |
| "learning_rate": 1.0700100666845568e-05, | |
| "loss": 0.9872, | |
| "step": 502 | |
| }, | |
| { | |
| "epoch": 0.6135081567312091, | |
| "grad_norm": 0.024747120216488838, | |
| "learning_rate": 1.0642107329592983e-05, | |
| "loss": 0.9964, | |
| "step": 503 | |
| }, | |
| { | |
| "epoch": 0.6147278548559232, | |
| "grad_norm": 0.02338460274040699, | |
| "learning_rate": 1.0584185050914942e-05, | |
| "loss": 0.9885, | |
| "step": 504 | |
| }, | |
| { | |
| "epoch": 0.6159475529806373, | |
| "grad_norm": 0.02610083669424057, | |
| "learning_rate": 1.0526334775275871e-05, | |
| "loss": 1.0097, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.6171672511053514, | |
| "grad_norm": 0.02438538335263729, | |
| "learning_rate": 1.0468557445966115e-05, | |
| "loss": 0.9762, | |
| "step": 506 | |
| }, | |
| { | |
| "epoch": 0.6183869492300655, | |
| "grad_norm": 0.026639766991138458, | |
| "learning_rate": 1.0410854005086583e-05, | |
| "loss": 0.9783, | |
| "step": 507 | |
| }, | |
| { | |
| "epoch": 0.6196066473547797, | |
| "grad_norm": 0.024155141785740852, | |
| "learning_rate": 1.0353225393533399e-05, | |
| "loss": 0.9824, | |
| "step": 508 | |
| }, | |
| { | |
| "epoch": 0.6208263454794938, | |
| "grad_norm": 0.025130869820713997, | |
| "learning_rate": 1.0295672550982519e-05, | |
| "loss": 0.9952, | |
| "step": 509 | |
| }, | |
| { | |
| "epoch": 0.622046043604208, | |
| "grad_norm": 0.02368522435426712, | |
| "learning_rate": 1.0238196415874426e-05, | |
| "loss": 0.9794, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.6232657417289221, | |
| "grad_norm": 0.02479284070432186, | |
| "learning_rate": 1.0180797925398856e-05, | |
| "loss": 1.0137, | |
| "step": 511 | |
| }, | |
| { | |
| "epoch": 0.6244854398536362, | |
| "grad_norm": 0.02335909940302372, | |
| "learning_rate": 1.0123478015479482e-05, | |
| "loss": 0.998, | |
| "step": 512 | |
| }, | |
| { | |
| "epoch": 0.6257051379783504, | |
| "grad_norm": 0.02379526011645794, | |
| "learning_rate": 1.0066237620758663e-05, | |
| "loss": 0.9846, | |
| "step": 513 | |
| }, | |
| { | |
| "epoch": 0.6269248361030645, | |
| "grad_norm": 0.026183098554611206, | |
| "learning_rate": 1.0009077674582207e-05, | |
| "loss": 0.9987, | |
| "step": 514 | |
| }, | |
| { | |
| "epoch": 0.6281445342277786, | |
| "grad_norm": 0.024180760607123375, | |
| "learning_rate": 9.951999108984146e-06, | |
| "loss": 0.9703, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.6293642323524927, | |
| "grad_norm": 0.024601083248853683, | |
| "learning_rate": 9.895002854671556e-06, | |
| "loss": 0.9746, | |
| "step": 516 | |
| }, | |
| { | |
| "epoch": 0.6305839304772068, | |
| "grad_norm": 0.025474026799201965, | |
| "learning_rate": 9.838089841009348e-06, | |
| "loss": 0.9842, | |
| "step": 517 | |
| }, | |
| { | |
| "epoch": 0.6318036286019211, | |
| "grad_norm": 0.029366258531808853, | |
| "learning_rate": 9.781260996005163e-06, | |
| "loss": 0.9744, | |
| "step": 518 | |
| }, | |
| { | |
| "epoch": 0.6330233267266352, | |
| "grad_norm": 0.025403836742043495, | |
| "learning_rate": 9.724517246294184e-06, | |
| "loss": 0.9819, | |
| "step": 519 | |
| }, | |
| { | |
| "epoch": 0.6342430248513493, | |
| "grad_norm": 0.024061335250735283, | |
| "learning_rate": 9.667859517124059e-06, | |
| "loss": 0.9836, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.6354627229760634, | |
| "grad_norm": 0.02342204749584198, | |
| "learning_rate": 9.61128873233983e-06, | |
| "loss": 0.979, | |
| "step": 521 | |
| }, | |
| { | |
| "epoch": 0.6366824211007776, | |
| "grad_norm": 0.023378275334835052, | |
| "learning_rate": 9.554805814368825e-06, | |
| "loss": 0.978, | |
| "step": 522 | |
| }, | |
| { | |
| "epoch": 0.6379021192254917, | |
| "grad_norm": 0.02484005317091942, | |
| "learning_rate": 9.498411684205646e-06, | |
| "loss": 0.9558, | |
| "step": 523 | |
| }, | |
| { | |
| "epoch": 0.6391218173502058, | |
| "grad_norm": 0.025706559419631958, | |
| "learning_rate": 9.442107261397154e-06, | |
| "loss": 0.9858, | |
| "step": 524 | |
| }, | |
| { | |
| "epoch": 0.6403415154749199, | |
| "grad_norm": 0.025092389434576035, | |
| "learning_rate": 9.385893464027456e-06, | |
| "loss": 0.9506, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.6415612135996341, | |
| "grad_norm": 0.025388307869434357, | |
| "learning_rate": 9.329771208702961e-06, | |
| "loss": 0.9999, | |
| "step": 526 | |
| }, | |
| { | |
| "epoch": 0.6427809117243483, | |
| "grad_norm": 0.02457476034760475, | |
| "learning_rate": 9.273741410537402e-06, | |
| "loss": 1.0019, | |
| "step": 527 | |
| }, | |
| { | |
| "epoch": 0.6440006098490624, | |
| "grad_norm": 0.027972573414444923, | |
| "learning_rate": 9.21780498313694e-06, | |
| "loss": 0.9428, | |
| "step": 528 | |
| }, | |
| { | |
| "epoch": 0.6452203079737765, | |
| "grad_norm": 0.02474704198539257, | |
| "learning_rate": 9.161962838585258e-06, | |
| "loss": 0.9697, | |
| "step": 529 | |
| }, | |
| { | |
| "epoch": 0.6464400060984906, | |
| "grad_norm": 0.024982508271932602, | |
| "learning_rate": 9.106215887428694e-06, | |
| "loss": 0.9677, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.6476597042232047, | |
| "grad_norm": 0.02674170210957527, | |
| "learning_rate": 9.05056503866138e-06, | |
| "loss": 0.9855, | |
| "step": 531 | |
| }, | |
| { | |
| "epoch": 0.6488794023479189, | |
| "grad_norm": 0.025713030248880386, | |
| "learning_rate": 8.995011199710438e-06, | |
| "loss": 1.0238, | |
| "step": 532 | |
| }, | |
| { | |
| "epoch": 0.650099100472633, | |
| "grad_norm": 0.02563433349132538, | |
| "learning_rate": 8.93955527642116e-06, | |
| "loss": 0.9963, | |
| "step": 533 | |
| }, | |
| { | |
| "epoch": 0.6513187985973472, | |
| "grad_norm": 0.02553797885775566, | |
| "learning_rate": 8.884198173042278e-06, | |
| "loss": 0.9846, | |
| "step": 534 | |
| }, | |
| { | |
| "epoch": 0.6525384967220613, | |
| "grad_norm": 0.025448612868785858, | |
| "learning_rate": 8.828940792211175e-06, | |
| "loss": 0.9915, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.6537581948467754, | |
| "grad_norm": 0.026187589392066002, | |
| "learning_rate": 8.773784034939183e-06, | |
| "loss": 0.9928, | |
| "step": 536 | |
| }, | |
| { | |
| "epoch": 0.6549778929714896, | |
| "grad_norm": 0.024752166122198105, | |
| "learning_rate": 8.718728800596912e-06, | |
| "loss": 0.997, | |
| "step": 537 | |
| }, | |
| { | |
| "epoch": 0.6561975910962037, | |
| "grad_norm": 0.02544168196618557, | |
| "learning_rate": 8.663775986899552e-06, | |
| "loss": 0.9956, | |
| "step": 538 | |
| }, | |
| { | |
| "epoch": 0.6574172892209178, | |
| "grad_norm": 0.024699505418539047, | |
| "learning_rate": 8.608926489892267e-06, | |
| "loss": 0.9714, | |
| "step": 539 | |
| }, | |
| { | |
| "epoch": 0.6586369873456319, | |
| "grad_norm": 0.024431807920336723, | |
| "learning_rate": 8.55418120393555e-06, | |
| "loss": 0.99, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.659856685470346, | |
| "grad_norm": 0.024489274248480797, | |
| "learning_rate": 8.499541021690663e-06, | |
| "loss": 0.9994, | |
| "step": 541 | |
| }, | |
| { | |
| "epoch": 0.6610763835950603, | |
| "grad_norm": 0.024301953613758087, | |
| "learning_rate": 8.445006834105088e-06, | |
| "loss": 0.984, | |
| "step": 542 | |
| }, | |
| { | |
| "epoch": 0.6622960817197744, | |
| "grad_norm": 0.02470739744603634, | |
| "learning_rate": 8.390579530397969e-06, | |
| "loss": 0.9929, | |
| "step": 543 | |
| }, | |
| { | |
| "epoch": 0.6635157798444885, | |
| "grad_norm": 0.024048753082752228, | |
| "learning_rate": 8.336259998045655e-06, | |
| "loss": 0.9713, | |
| "step": 544 | |
| }, | |
| { | |
| "epoch": 0.6647354779692026, | |
| "grad_norm": 0.023202264681458473, | |
| "learning_rate": 8.282049122767182e-06, | |
| "loss": 0.9662, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.6659551760939167, | |
| "grad_norm": 0.0271754153072834, | |
| "learning_rate": 8.227947788509862e-06, | |
| "loss": 0.9717, | |
| "step": 546 | |
| }, | |
| { | |
| "epoch": 0.6671748742186309, | |
| "grad_norm": 0.024151325225830078, | |
| "learning_rate": 8.173956877434872e-06, | |
| "loss": 0.9723, | |
| "step": 547 | |
| }, | |
| { | |
| "epoch": 0.668394572343345, | |
| "grad_norm": 0.025481166318058968, | |
| "learning_rate": 8.120077269902845e-06, | |
| "loss": 0.9712, | |
| "step": 548 | |
| }, | |
| { | |
| "epoch": 0.6696142704680591, | |
| "grad_norm": 0.024747828021645546, | |
| "learning_rate": 8.066309844459543e-06, | |
| "loss": 1.0042, | |
| "step": 549 | |
| }, | |
| { | |
| "epoch": 0.6708339685927733, | |
| "grad_norm": 0.023438094183802605, | |
| "learning_rate": 8.012655477821513e-06, | |
| "loss": 0.9782, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.6720536667174875, | |
| "grad_norm": 0.025299984961748123, | |
| "learning_rate": 7.959115044861783e-06, | |
| "loss": 0.9735, | |
| "step": 551 | |
| }, | |
| { | |
| "epoch": 0.6732733648422016, | |
| "grad_norm": 0.02624981664121151, | |
| "learning_rate": 7.905689418595637e-06, | |
| "loss": 0.9906, | |
| "step": 552 | |
| }, | |
| { | |
| "epoch": 0.6744930629669157, | |
| "grad_norm": 0.023212846368551254, | |
| "learning_rate": 7.852379470166334e-06, | |
| "loss": 0.9792, | |
| "step": 553 | |
| }, | |
| { | |
| "epoch": 0.6757127610916298, | |
| "grad_norm": 0.023791294544935226, | |
| "learning_rate": 7.799186068830924e-06, | |
| "loss": 0.9637, | |
| "step": 554 | |
| }, | |
| { | |
| "epoch": 0.6769324592163439, | |
| "grad_norm": 0.02353626862168312, | |
| "learning_rate": 7.746110081946085e-06, | |
| "loss": 0.9567, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.678152157341058, | |
| "grad_norm": 0.025470618158578873, | |
| "learning_rate": 7.69315237495395e-06, | |
| "loss": 0.9693, | |
| "step": 556 | |
| }, | |
| { | |
| "epoch": 0.6793718554657722, | |
| "grad_norm": 0.024698182940483093, | |
| "learning_rate": 7.640313811368036e-06, | |
| "loss": 0.9811, | |
| "step": 557 | |
| }, | |
| { | |
| "epoch": 0.6805915535904864, | |
| "grad_norm": 0.023996494710445404, | |
| "learning_rate": 7.5875952527591195e-06, | |
| "loss": 0.9875, | |
| "step": 558 | |
| }, | |
| { | |
| "epoch": 0.6818112517152005, | |
| "grad_norm": 0.023118799552321434, | |
| "learning_rate": 7.5349975587412145e-06, | |
| "loss": 0.9706, | |
| "step": 559 | |
| }, | |
| { | |
| "epoch": 0.6830309498399146, | |
| "grad_norm": 0.025125648826360703, | |
| "learning_rate": 7.482521586957563e-06, | |
| "loss": 0.9843, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.6842506479646288, | |
| "grad_norm": 0.02443476766347885, | |
| "learning_rate": 7.430168193066622e-06, | |
| "loss": 0.9894, | |
| "step": 561 | |
| }, | |
| { | |
| "epoch": 0.6854703460893429, | |
| "grad_norm": 0.02420630119740963, | |
| "learning_rate": 7.377938230728148e-06, | |
| "loss": 0.9863, | |
| "step": 562 | |
| }, | |
| { | |
| "epoch": 0.686690044214057, | |
| "grad_norm": 0.02491050213575363, | |
| "learning_rate": 7.3258325515892286e-06, | |
| "loss": 0.9797, | |
| "step": 563 | |
| }, | |
| { | |
| "epoch": 0.6879097423387711, | |
| "grad_norm": 0.024198705330491066, | |
| "learning_rate": 7.273852005270457e-06, | |
| "loss": 0.9573, | |
| "step": 564 | |
| }, | |
| { | |
| "epoch": 0.6891294404634852, | |
| "grad_norm": 0.023970816284418106, | |
| "learning_rate": 7.221997439352015e-06, | |
| "loss": 0.9783, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.6903491385881995, | |
| "grad_norm": 0.025408467277884483, | |
| "learning_rate": 7.170269699359913e-06, | |
| "loss": 0.9822, | |
| "step": 566 | |
| }, | |
| { | |
| "epoch": 0.6915688367129136, | |
| "grad_norm": 0.023949354887008667, | |
| "learning_rate": 7.118669628752144e-06, | |
| "loss": 0.9942, | |
| "step": 567 | |
| }, | |
| { | |
| "epoch": 0.6927885348376277, | |
| "grad_norm": 0.023415151983499527, | |
| "learning_rate": 7.067198068904985e-06, | |
| "loss": 0.974, | |
| "step": 568 | |
| }, | |
| { | |
| "epoch": 0.6940082329623418, | |
| "grad_norm": 0.023221898823976517, | |
| "learning_rate": 7.015855859099233e-06, | |
| "loss": 0.9866, | |
| "step": 569 | |
| }, | |
| { | |
| "epoch": 0.6952279310870559, | |
| "grad_norm": 0.02316485531628132, | |
| "learning_rate": 6.96464383650656e-06, | |
| "loss": 0.9891, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.6964476292117701, | |
| "grad_norm": 0.02432584948837757, | |
| "learning_rate": 6.9135628361758225e-06, | |
| "loss": 0.9636, | |
| "step": 571 | |
| }, | |
| { | |
| "epoch": 0.6976673273364842, | |
| "grad_norm": 0.02315198816359043, | |
| "learning_rate": 6.862613691019467e-06, | |
| "loss": 0.96, | |
| "step": 572 | |
| }, | |
| { | |
| "epoch": 0.6988870254611984, | |
| "grad_norm": 0.023875737562775612, | |
| "learning_rate": 6.811797231799967e-06, | |
| "loss": 0.9659, | |
| "step": 573 | |
| }, | |
| { | |
| "epoch": 0.7001067235859125, | |
| "grad_norm": 0.024141857400536537, | |
| "learning_rate": 6.761114287116226e-06, | |
| "loss": 0.9706, | |
| "step": 574 | |
| }, | |
| { | |
| "epoch": 0.7013264217106266, | |
| "grad_norm": 0.023318905383348465, | |
| "learning_rate": 6.710565683390124e-06, | |
| "loss": 0.9763, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.7025461198353408, | |
| "grad_norm": 0.023288151249289513, | |
| "learning_rate": 6.6601522448530005e-06, | |
| "loss": 0.9556, | |
| "step": 576 | |
| }, | |
| { | |
| "epoch": 0.7037658179600549, | |
| "grad_norm": 0.024233417585492134, | |
| "learning_rate": 6.609874793532224e-06, | |
| "loss": 1.0007, | |
| "step": 577 | |
| }, | |
| { | |
| "epoch": 0.704985516084769, | |
| "grad_norm": 0.023677371442317963, | |
| "learning_rate": 6.559734149237816e-06, | |
| "loss": 0.9621, | |
| "step": 578 | |
| }, | |
| { | |
| "epoch": 0.7062052142094831, | |
| "grad_norm": 0.02412758395075798, | |
| "learning_rate": 6.509731129549038e-06, | |
| "loss": 0.9875, | |
| "step": 579 | |
| }, | |
| { | |
| "epoch": 0.7074249123341972, | |
| "grad_norm": 0.02296406216919422, | |
| "learning_rate": 6.45986654980109e-06, | |
| "loss": 0.9773, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.7086446104589115, | |
| "grad_norm": 0.023887090384960175, | |
| "learning_rate": 6.410141223071821e-06, | |
| "loss": 0.9817, | |
| "step": 581 | |
| }, | |
| { | |
| "epoch": 0.7098643085836256, | |
| "grad_norm": 0.02324073202908039, | |
| "learning_rate": 6.3605559601684385e-06, | |
| "loss": 0.9907, | |
| "step": 582 | |
| }, | |
| { | |
| "epoch": 0.7110840067083397, | |
| "grad_norm": 0.0239879097789526, | |
| "learning_rate": 6.311111569614332e-06, | |
| "loss": 0.994, | |
| "step": 583 | |
| }, | |
| { | |
| "epoch": 0.7123037048330538, | |
| "grad_norm": 0.02468857169151306, | |
| "learning_rate": 6.261808857635847e-06, | |
| "loss": 0.9679, | |
| "step": 584 | |
| }, | |
| { | |
| "epoch": 0.713523402957768, | |
| "grad_norm": 0.023504672572016716, | |
| "learning_rate": 6.21264862814916e-06, | |
| "loss": 0.9614, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.7147431010824821, | |
| "grad_norm": 0.025298135355114937, | |
| "learning_rate": 6.163631682747186e-06, | |
| "loss": 0.9968, | |
| "step": 586 | |
| }, | |
| { | |
| "epoch": 0.7159627992071962, | |
| "grad_norm": 0.022650722414255142, | |
| "learning_rate": 6.114758820686462e-06, | |
| "loss": 0.9614, | |
| "step": 587 | |
| }, | |
| { | |
| "epoch": 0.7171824973319103, | |
| "grad_norm": 0.02556091547012329, | |
| "learning_rate": 6.066030838874172e-06, | |
| "loss": 1.0024, | |
| "step": 588 | |
| }, | |
| { | |
| "epoch": 0.7184021954566245, | |
| "grad_norm": 0.024834884330630302, | |
| "learning_rate": 6.017448531855101e-06, | |
| "loss": 0.9597, | |
| "step": 589 | |
| }, | |
| { | |
| "epoch": 0.7196218935813387, | |
| "grad_norm": 0.0242439117282629, | |
| "learning_rate": 5.969012691798704e-06, | |
| "loss": 0.9672, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.7208415917060528, | |
| "grad_norm": 0.024517491459846497, | |
| "learning_rate": 5.9207241084862025e-06, | |
| "loss": 0.9869, | |
| "step": 591 | |
| }, | |
| { | |
| "epoch": 0.7220612898307669, | |
| "grad_norm": 0.024099048227071762, | |
| "learning_rate": 5.872583569297673e-06, | |
| "loss": 0.9836, | |
| "step": 592 | |
| }, | |
| { | |
| "epoch": 0.723280987955481, | |
| "grad_norm": 0.024225030094385147, | |
| "learning_rate": 5.82459185919924e-06, | |
| "loss": 0.9873, | |
| "step": 593 | |
| }, | |
| { | |
| "epoch": 0.7245006860801951, | |
| "grad_norm": 0.02390759438276291, | |
| "learning_rate": 5.776749760730249e-06, | |
| "loss": 0.9843, | |
| "step": 594 | |
| }, | |
| { | |
| "epoch": 0.7257203842049093, | |
| "grad_norm": 0.02401016838848591, | |
| "learning_rate": 5.729058053990535e-06, | |
| "loss": 0.9551, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.7269400823296234, | |
| "grad_norm": 0.024140631780028343, | |
| "learning_rate": 5.681517516627672e-06, | |
| "loss": 0.9566, | |
| "step": 596 | |
| }, | |
| { | |
| "epoch": 0.7281597804543376, | |
| "grad_norm": 0.023510301485657692, | |
| "learning_rate": 5.634128923824327e-06, | |
| "loss": 0.9592, | |
| "step": 597 | |
| }, | |
| { | |
| "epoch": 0.7293794785790517, | |
| "grad_norm": 0.02437392808496952, | |
| "learning_rate": 5.586893048285581e-06, | |
| "loss": 0.9718, | |
| "step": 598 | |
| }, | |
| { | |
| "epoch": 0.7305991767037658, | |
| "grad_norm": 0.026473166421055794, | |
| "learning_rate": 5.5398106602263696e-06, | |
| "loss": 0.9895, | |
| "step": 599 | |
| }, | |
| { | |
| "epoch": 0.73181887482848, | |
| "grad_norm": 0.0227658674120903, | |
| "learning_rate": 5.492882527358886e-06, | |
| "loss": 0.9831, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.7330385729531941, | |
| "grad_norm": 0.02404717542231083, | |
| "learning_rate": 5.446109414880103e-06, | |
| "loss": 0.9599, | |
| "step": 601 | |
| }, | |
| { | |
| "epoch": 0.7342582710779082, | |
| "grad_norm": 0.024451611563563347, | |
| "learning_rate": 5.399492085459256e-06, | |
| "loss": 0.9792, | |
| "step": 602 | |
| }, | |
| { | |
| "epoch": 0.7354779692026223, | |
| "grad_norm": 0.023207319900393486, | |
| "learning_rate": 5.353031299225426e-06, | |
| "loss": 0.9867, | |
| "step": 603 | |
| }, | |
| { | |
| "epoch": 0.7366976673273364, | |
| "grad_norm": 0.023120377212762833, | |
| "learning_rate": 5.306727813755164e-06, | |
| "loss": 0.9792, | |
| "step": 604 | |
| }, | |
| { | |
| "epoch": 0.7379173654520507, | |
| "grad_norm": 0.02580856904387474, | |
| "learning_rate": 5.260582384060099e-06, | |
| "loss": 0.9901, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.7391370635767648, | |
| "grad_norm": 0.023656006902456284, | |
| "learning_rate": 5.214595762574649e-06, | |
| "loss": 0.9698, | |
| "step": 606 | |
| }, | |
| { | |
| "epoch": 0.7403567617014789, | |
| "grad_norm": 0.0238718930631876, | |
| "learning_rate": 5.168768699143764e-06, | |
| "loss": 0.9869, | |
| "step": 607 | |
| }, | |
| { | |
| "epoch": 0.741576459826193, | |
| "grad_norm": 0.02434396743774414, | |
| "learning_rate": 5.123101941010669e-06, | |
| "loss": 0.9874, | |
| "step": 608 | |
| }, | |
| { | |
| "epoch": 0.7427961579509071, | |
| "grad_norm": 0.024040423333644867, | |
| "learning_rate": 5.077596232804707e-06, | |
| "loss": 0.9949, | |
| "step": 609 | |
| }, | |
| { | |
| "epoch": 0.7440158560756213, | |
| "grad_norm": 0.02326955646276474, | |
| "learning_rate": 5.0322523165291825e-06, | |
| "loss": 0.975, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.7452355542003354, | |
| "grad_norm": 0.023894594982266426, | |
| "learning_rate": 4.987070931549257e-06, | |
| "loss": 0.9729, | |
| "step": 611 | |
| }, | |
| { | |
| "epoch": 0.7464552523250495, | |
| "grad_norm": 0.024369854480028152, | |
| "learning_rate": 4.942052814579922e-06, | |
| "loss": 0.9693, | |
| "step": 612 | |
| }, | |
| { | |
| "epoch": 0.7476749504497637, | |
| "grad_norm": 0.024013178423047066, | |
| "learning_rate": 4.897198699673945e-06, | |
| "loss": 0.9742, | |
| "step": 613 | |
| }, | |
| { | |
| "epoch": 0.7488946485744779, | |
| "grad_norm": 0.022922024130821228, | |
| "learning_rate": 4.852509318209944e-06, | |
| "loss": 0.986, | |
| "step": 614 | |
| }, | |
| { | |
| "epoch": 0.750114346699192, | |
| "grad_norm": 0.025753559544682503, | |
| "learning_rate": 4.80798539888042e-06, | |
| "loss": 0.9812, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.7513340448239061, | |
| "grad_norm": 0.024801813066005707, | |
| "learning_rate": 4.7636276676799e-06, | |
| "loss": 0.9509, | |
| "step": 616 | |
| }, | |
| { | |
| "epoch": 0.7525537429486202, | |
| "grad_norm": 0.023664768785238266, | |
| "learning_rate": 4.719436847893106e-06, | |
| "loss": 0.9829, | |
| "step": 617 | |
| }, | |
| { | |
| "epoch": 0.7537734410733343, | |
| "grad_norm": 0.02687576785683632, | |
| "learning_rate": 4.675413660083137e-06, | |
| "loss": 0.9839, | |
| "step": 618 | |
| }, | |
| { | |
| "epoch": 0.7549931391980484, | |
| "grad_norm": 0.026260821148753166, | |
| "learning_rate": 4.631558822079728e-06, | |
| "loss": 0.9813, | |
| "step": 619 | |
| }, | |
| { | |
| "epoch": 0.7562128373227626, | |
| "grad_norm": 0.02375507913529873, | |
| "learning_rate": 4.587873048967568e-06, | |
| "loss": 0.9854, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.7574325354474768, | |
| "grad_norm": 0.02434990555047989, | |
| "learning_rate": 4.544357053074601e-06, | |
| "loss": 0.9674, | |
| "step": 621 | |
| }, | |
| { | |
| "epoch": 0.7586522335721909, | |
| "grad_norm": 0.029402360320091248, | |
| "learning_rate": 4.501011543960448e-06, | |
| "loss": 0.9848, | |
| "step": 622 | |
| }, | |
| { | |
| "epoch": 0.759871931696905, | |
| "grad_norm": 0.026814134791493416, | |
| "learning_rate": 4.4578372284048045e-06, | |
| "loss": 0.9624, | |
| "step": 623 | |
| }, | |
| { | |
| "epoch": 0.7610916298216192, | |
| "grad_norm": 0.02349013090133667, | |
| "learning_rate": 4.414834810395948e-06, | |
| "loss": 0.9454, | |
| "step": 624 | |
| }, | |
| { | |
| "epoch": 0.7623113279463333, | |
| "grad_norm": 0.026365788653492928, | |
| "learning_rate": 4.37200499111923e-06, | |
| "loss": 0.961, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.7635310260710474, | |
| "grad_norm": 0.024836424738168716, | |
| "learning_rate": 4.329348468945663e-06, | |
| "loss": 0.9641, | |
| "step": 626 | |
| }, | |
| { | |
| "epoch": 0.7647507241957615, | |
| "grad_norm": 0.023668797686696053, | |
| "learning_rate": 4.28686593942052e-06, | |
| "loss": 0.9814, | |
| "step": 627 | |
| }, | |
| { | |
| "epoch": 0.7659704223204756, | |
| "grad_norm": 0.02429603785276413, | |
| "learning_rate": 4.244558095252008e-06, | |
| "loss": 0.9671, | |
| "step": 628 | |
| }, | |
| { | |
| "epoch": 0.7671901204451899, | |
| "grad_norm": 0.024740032851696014, | |
| "learning_rate": 4.20242562629995e-06, | |
| "loss": 0.9923, | |
| "step": 629 | |
| }, | |
| { | |
| "epoch": 0.768409818569904, | |
| "grad_norm": 0.022463154047727585, | |
| "learning_rate": 4.160469219564564e-06, | |
| "loss": 0.957, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.7696295166946181, | |
| "grad_norm": 0.02261923998594284, | |
| "learning_rate": 4.118689559175234e-06, | |
| "loss": 0.9855, | |
| "step": 631 | |
| }, | |
| { | |
| "epoch": 0.7708492148193322, | |
| "grad_norm": 0.02499699778854847, | |
| "learning_rate": 4.077087326379382e-06, | |
| "loss": 0.965, | |
| "step": 632 | |
| }, | |
| { | |
| "epoch": 0.7720689129440463, | |
| "grad_norm": 0.022720512002706528, | |
| "learning_rate": 4.035663199531333e-06, | |
| "loss": 0.9677, | |
| "step": 633 | |
| }, | |
| { | |
| "epoch": 0.7732886110687605, | |
| "grad_norm": 0.02426447533071041, | |
| "learning_rate": 3.994417854081264e-06, | |
| "loss": 0.9693, | |
| "step": 634 | |
| }, | |
| { | |
| "epoch": 0.7745083091934746, | |
| "grad_norm": 0.02303161658346653, | |
| "learning_rate": 3.953351962564213e-06, | |
| "loss": 0.9566, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.7757280073181887, | |
| "grad_norm": 0.02424640767276287, | |
| "learning_rate": 3.912466194589072e-06, | |
| "loss": 1.002, | |
| "step": 636 | |
| }, | |
| { | |
| "epoch": 0.7769477054429029, | |
| "grad_norm": 0.023175561800599098, | |
| "learning_rate": 3.871761216827692e-06, | |
| "loss": 0.98, | |
| "step": 637 | |
| }, | |
| { | |
| "epoch": 0.778167403567617, | |
| "grad_norm": 0.02275039441883564, | |
| "learning_rate": 3.8312376930040235e-06, | |
| "loss": 0.9637, | |
| "step": 638 | |
| }, | |
| { | |
| "epoch": 0.7793871016923312, | |
| "grad_norm": 0.023998327553272247, | |
| "learning_rate": 3.7908962838832607e-06, | |
| "loss": 0.9722, | |
| "step": 639 | |
| }, | |
| { | |
| "epoch": 0.7806067998170453, | |
| "grad_norm": 0.02376868575811386, | |
| "learning_rate": 3.750737647261105e-06, | |
| "loss": 0.947, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.7818264979417594, | |
| "grad_norm": 0.022538412362337112, | |
| "learning_rate": 3.7107624379530087e-06, | |
| "loss": 0.965, | |
| "step": 641 | |
| }, | |
| { | |
| "epoch": 0.7830461960664735, | |
| "grad_norm": 0.02420453354716301, | |
| "learning_rate": 3.670971307783506e-06, | |
| "loss": 0.9687, | |
| "step": 642 | |
| }, | |
| { | |
| "epoch": 0.7842658941911876, | |
| "grad_norm": 0.023562395945191383, | |
| "learning_rate": 3.6313649055756077e-06, | |
| "loss": 0.9643, | |
| "step": 643 | |
| }, | |
| { | |
| "epoch": 0.7854855923159018, | |
| "grad_norm": 0.02423342689871788, | |
| "learning_rate": 3.591943877140178e-06, | |
| "loss": 0.9901, | |
| "step": 644 | |
| }, | |
| { | |
| "epoch": 0.786705290440616, | |
| "grad_norm": 0.023145092651247978, | |
| "learning_rate": 3.5527088652654497e-06, | |
| "loss": 0.9874, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.7879249885653301, | |
| "grad_norm": 0.022533966228365898, | |
| "learning_rate": 3.5136605097065107e-06, | |
| "loss": 0.9735, | |
| "step": 646 | |
| }, | |
| { | |
| "epoch": 0.7891446866900442, | |
| "grad_norm": 0.02232767455279827, | |
| "learning_rate": 3.4747994471748784e-06, | |
| "loss": 0.9641, | |
| "step": 647 | |
| }, | |
| { | |
| "epoch": 0.7903643848147583, | |
| "grad_norm": 0.022890832275152206, | |
| "learning_rate": 3.43612631132814e-06, | |
| "loss": 0.9617, | |
| "step": 648 | |
| }, | |
| { | |
| "epoch": 0.7915840829394725, | |
| "grad_norm": 0.02412954904139042, | |
| "learning_rate": 3.3976417327595845e-06, | |
| "loss": 0.9896, | |
| "step": 649 | |
| }, | |
| { | |
| "epoch": 0.7928037810641866, | |
| "grad_norm": 0.021985746920108795, | |
| "learning_rate": 3.359346338987947e-06, | |
| "loss": 0.9547, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.7940234791889007, | |
| "grad_norm": 0.023143811151385307, | |
| "learning_rate": 3.321240754447174e-06, | |
| "loss": 0.946, | |
| "step": 651 | |
| }, | |
| { | |
| "epoch": 0.7952431773136149, | |
| "grad_norm": 0.023708276450634003, | |
| "learning_rate": 3.2833256004762206e-06, | |
| "loss": 0.9863, | |
| "step": 652 | |
| }, | |
| { | |
| "epoch": 0.796462875438329, | |
| "grad_norm": 0.023042501881718636, | |
| "learning_rate": 3.2456014953089565e-06, | |
| "loss": 0.9706, | |
| "step": 653 | |
| }, | |
| { | |
| "epoch": 0.7976825735630432, | |
| "grad_norm": 0.022853119298815727, | |
| "learning_rate": 3.2080690540640382e-06, | |
| "loss": 0.9818, | |
| "step": 654 | |
| }, | |
| { | |
| "epoch": 0.7989022716877573, | |
| "grad_norm": 0.022742360830307007, | |
| "learning_rate": 3.170728888734926e-06, | |
| "loss": 0.9726, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.8001219698124714, | |
| "grad_norm": 0.023535635322332382, | |
| "learning_rate": 3.1335816081798667e-06, | |
| "loss": 0.9635, | |
| "step": 656 | |
| }, | |
| { | |
| "epoch": 0.8013416679371855, | |
| "grad_norm": 0.023480627685785294, | |
| "learning_rate": 3.096627818111997e-06, | |
| "loss": 0.9751, | |
| "step": 657 | |
| }, | |
| { | |
| "epoch": 0.8025613660618997, | |
| "grad_norm": 0.022936956956982613, | |
| "learning_rate": 3.0598681210894357e-06, | |
| "loss": 0.9479, | |
| "step": 658 | |
| }, | |
| { | |
| "epoch": 0.8037810641866138, | |
| "grad_norm": 0.02312362752854824, | |
| "learning_rate": 3.023303116505491e-06, | |
| "loss": 0.9766, | |
| "step": 659 | |
| }, | |
| { | |
| "epoch": 0.805000762311328, | |
| "grad_norm": 0.022837257012724876, | |
| "learning_rate": 2.986933400578856e-06, | |
| "loss": 0.9901, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.8062204604360421, | |
| "grad_norm": 0.02275831811130047, | |
| "learning_rate": 2.950759566343917e-06, | |
| "loss": 0.9898, | |
| "step": 661 | |
| }, | |
| { | |
| "epoch": 0.8074401585607562, | |
| "grad_norm": 0.022969406098127365, | |
| "learning_rate": 2.914782203641056e-06, | |
| "loss": 0.962, | |
| "step": 662 | |
| }, | |
| { | |
| "epoch": 0.8086598566854704, | |
| "grad_norm": 0.022439919412136078, | |
| "learning_rate": 2.8790018991070443e-06, | |
| "loss": 0.9439, | |
| "step": 663 | |
| }, | |
| { | |
| "epoch": 0.8098795548101845, | |
| "grad_norm": 0.023537321016192436, | |
| "learning_rate": 2.8434192361654946e-06, | |
| "loss": 0.9677, | |
| "step": 664 | |
| }, | |
| { | |
| "epoch": 0.8110992529348986, | |
| "grad_norm": 0.02396860532462597, | |
| "learning_rate": 2.8080347950173097e-06, | |
| "loss": 0.9667, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.8123189510596127, | |
| "grad_norm": 0.02312847599387169, | |
| "learning_rate": 2.772849152631263e-06, | |
| "loss": 0.9838, | |
| "step": 666 | |
| }, | |
| { | |
| "epoch": 0.8135386491843268, | |
| "grad_norm": 0.02397603914141655, | |
| "learning_rate": 2.737862882734558e-06, | |
| "loss": 0.971, | |
| "step": 667 | |
| }, | |
| { | |
| "epoch": 0.8147583473090411, | |
| "grad_norm": 0.02314850687980652, | |
| "learning_rate": 2.7030765558034855e-06, | |
| "loss": 0.9646, | |
| "step": 668 | |
| }, | |
| { | |
| "epoch": 0.8159780454337552, | |
| "grad_norm": 0.021661944687366486, | |
| "learning_rate": 2.668490739054136e-06, | |
| "loss": 0.9862, | |
| "step": 669 | |
| }, | |
| { | |
| "epoch": 0.8171977435584693, | |
| "grad_norm": 0.022165222093462944, | |
| "learning_rate": 2.6341059964331206e-06, | |
| "loss": 0.968, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.8184174416831834, | |
| "grad_norm": 0.022893980145454407, | |
| "learning_rate": 2.599922888608408e-06, | |
| "loss": 0.9612, | |
| "step": 671 | |
| }, | |
| { | |
| "epoch": 0.8196371398078975, | |
| "grad_norm": 0.024064816534519196, | |
| "learning_rate": 2.5659419729601543e-06, | |
| "loss": 0.9594, | |
| "step": 672 | |
| }, | |
| { | |
| "epoch": 0.8208568379326117, | |
| "grad_norm": 0.02254064381122589, | |
| "learning_rate": 2.5321638035716276e-06, | |
| "loss": 0.9551, | |
| "step": 673 | |
| }, | |
| { | |
| "epoch": 0.8220765360573258, | |
| "grad_norm": 0.022333014756441116, | |
| "learning_rate": 2.4985889312201827e-06, | |
| "loss": 0.9619, | |
| "step": 674 | |
| }, | |
| { | |
| "epoch": 0.8232962341820399, | |
| "grad_norm": 0.023305319249629974, | |
| "learning_rate": 2.465217903368261e-06, | |
| "loss": 0.9776, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.8245159323067541, | |
| "grad_norm": 0.0232695359736681, | |
| "learning_rate": 2.4320512641544702e-06, | |
| "loss": 1.0044, | |
| "step": 676 | |
| }, | |
| { | |
| "epoch": 0.8257356304314682, | |
| "grad_norm": 0.022830158472061157, | |
| "learning_rate": 2.3990895543847254e-06, | |
| "loss": 0.9875, | |
| "step": 677 | |
| }, | |
| { | |
| "epoch": 0.8269553285561824, | |
| "grad_norm": 0.023292941972613335, | |
| "learning_rate": 2.3663333115234057e-06, | |
| "loss": 0.9661, | |
| "step": 678 | |
| }, | |
| { | |
| "epoch": 0.8281750266808965, | |
| "grad_norm": 0.02342226542532444, | |
| "learning_rate": 2.33378306968462e-06, | |
| "loss": 0.9689, | |
| "step": 679 | |
| }, | |
| { | |
| "epoch": 0.8293947248056106, | |
| "grad_norm": 0.022832544520497322, | |
| "learning_rate": 2.301439359623473e-06, | |
| "loss": 0.98, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.8306144229303247, | |
| "grad_norm": 0.022983156144618988, | |
| "learning_rate": 2.2693027087274142e-06, | |
| "loss": 0.9665, | |
| "step": 681 | |
| }, | |
| { | |
| "epoch": 0.8318341210550388, | |
| "grad_norm": 0.02287602238357067, | |
| "learning_rate": 2.237373641007659e-06, | |
| "loss": 0.9686, | |
| "step": 682 | |
| }, | |
| { | |
| "epoch": 0.833053819179753, | |
| "grad_norm": 0.02267058938741684, | |
| "learning_rate": 2.2056526770906186e-06, | |
| "loss": 0.9659, | |
| "step": 683 | |
| }, | |
| { | |
| "epoch": 0.8342735173044672, | |
| "grad_norm": 0.022243564948439598, | |
| "learning_rate": 2.174140334209432e-06, | |
| "loss": 0.9647, | |
| "step": 684 | |
| }, | |
| { | |
| "epoch": 0.8354932154291813, | |
| "grad_norm": 0.022451631724834442, | |
| "learning_rate": 2.142837126195507e-06, | |
| "loss": 0.9536, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.8367129135538954, | |
| "grad_norm": 0.02310284599661827, | |
| "learning_rate": 2.1117435634701746e-06, | |
| "loss": 0.9639, | |
| "step": 686 | |
| }, | |
| { | |
| "epoch": 0.8379326116786096, | |
| "grad_norm": 0.022542696446180344, | |
| "learning_rate": 2.080860153036333e-06, | |
| "loss": 0.99, | |
| "step": 687 | |
| }, | |
| { | |
| "epoch": 0.8391523098033237, | |
| "grad_norm": 0.02292606607079506, | |
| "learning_rate": 2.0501873984702105e-06, | |
| "loss": 0.9752, | |
| "step": 688 | |
| }, | |
| { | |
| "epoch": 0.8403720079280378, | |
| "grad_norm": 0.022897856310009956, | |
| "learning_rate": 2.0197257999131265e-06, | |
| "loss": 0.9783, | |
| "step": 689 | |
| }, | |
| { | |
| "epoch": 0.8415917060527519, | |
| "grad_norm": 0.022700980305671692, | |
| "learning_rate": 1.9894758540633607e-06, | |
| "loss": 0.9615, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.842811404177466, | |
| "grad_norm": 0.022575560957193375, | |
| "learning_rate": 1.9594380541680336e-06, | |
| "loss": 0.9748, | |
| "step": 691 | |
| }, | |
| { | |
| "epoch": 0.8440311023021803, | |
| "grad_norm": 0.02273186482489109, | |
| "learning_rate": 1.929612890015083e-06, | |
| "loss": 0.9696, | |
| "step": 692 | |
| }, | |
| { | |
| "epoch": 0.8452508004268944, | |
| "grad_norm": 0.02176203578710556, | |
| "learning_rate": 1.9000008479252607e-06, | |
| "loss": 0.9719, | |
| "step": 693 | |
| }, | |
| { | |
| "epoch": 0.8464704985516085, | |
| "grad_norm": 0.02173522114753723, | |
| "learning_rate": 1.8706024107442105e-06, | |
| "loss": 0.9383, | |
| "step": 694 | |
| }, | |
| { | |
| "epoch": 0.8476901966763226, | |
| "grad_norm": 0.023877400904893875, | |
| "learning_rate": 1.8414180578346006e-06, | |
| "loss": 1.0017, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.8489098948010367, | |
| "grad_norm": 0.023012325167655945, | |
| "learning_rate": 1.8124482650682934e-06, | |
| "loss": 0.9334, | |
| "step": 696 | |
| }, | |
| { | |
| "epoch": 0.8501295929257509, | |
| "grad_norm": 0.022058291360735893, | |
| "learning_rate": 1.7836935048186003e-06, | |
| "loss": 0.9685, | |
| "step": 697 | |
| }, | |
| { | |
| "epoch": 0.851349291050465, | |
| "grad_norm": 0.022319795563817024, | |
| "learning_rate": 1.755154245952569e-06, | |
| "loss": 0.9591, | |
| "step": 698 | |
| }, | |
| { | |
| "epoch": 0.8525689891751791, | |
| "grad_norm": 0.0229625403881073, | |
| "learning_rate": 1.7268309538233396e-06, | |
| "loss": 0.9603, | |
| "step": 699 | |
| }, | |
| { | |
| "epoch": 0.8537886872998933, | |
| "grad_norm": 0.02283914014697075, | |
| "learning_rate": 1.6987240902625689e-06, | |
| "loss": 0.9679, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.8550083854246074, | |
| "grad_norm": 0.02273203618824482, | |
| "learning_rate": 1.6708341135728794e-06, | |
| "loss": 0.966, | |
| "step": 701 | |
| }, | |
| { | |
| "epoch": 0.8562280835493216, | |
| "grad_norm": 0.022080956026911736, | |
| "learning_rate": 1.6431614785204025e-06, | |
| "loss": 0.9377, | |
| "step": 702 | |
| }, | |
| { | |
| "epoch": 0.8574477816740357, | |
| "grad_norm": 0.023274419829249382, | |
| "learning_rate": 1.6157066363273594e-06, | |
| "loss": 0.9784, | |
| "step": 703 | |
| }, | |
| { | |
| "epoch": 0.8586674797987498, | |
| "grad_norm": 0.0224025696516037, | |
| "learning_rate": 1.5884700346646969e-06, | |
| "loss": 0.9783, | |
| "step": 704 | |
| }, | |
| { | |
| "epoch": 0.8598871779234639, | |
| "grad_norm": 0.02262830175459385, | |
| "learning_rate": 1.5614521176447988e-06, | |
| "loss": 0.9756, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.861106876048178, | |
| "grad_norm": 0.023085977882146835, | |
| "learning_rate": 1.5346533258142342e-06, | |
| "loss": 0.9811, | |
| "step": 706 | |
| }, | |
| { | |
| "epoch": 0.8623265741728922, | |
| "grad_norm": 0.02333192341029644, | |
| "learning_rate": 1.508074096146579e-06, | |
| "loss": 0.9683, | |
| "step": 707 | |
| }, | |
| { | |
| "epoch": 0.8635462722976064, | |
| "grad_norm": 0.022308291867375374, | |
| "learning_rate": 1.4817148620352921e-06, | |
| "loss": 0.9931, | |
| "step": 708 | |
| }, | |
| { | |
| "epoch": 0.8647659704223205, | |
| "grad_norm": 0.022744152694940567, | |
| "learning_rate": 1.455576053286643e-06, | |
| "loss": 0.9632, | |
| "step": 709 | |
| }, | |
| { | |
| "epoch": 0.8659856685470346, | |
| "grad_norm": 0.022630611434578896, | |
| "learning_rate": 1.4296580961127154e-06, | |
| "loss": 0.968, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.8672053666717487, | |
| "grad_norm": 0.022587062790989876, | |
| "learning_rate": 1.4039614131244388e-06, | |
| "loss": 0.9735, | |
| "step": 711 | |
| }, | |
| { | |
| "epoch": 0.8684250647964629, | |
| "grad_norm": 0.02243993617594242, | |
| "learning_rate": 1.3784864233247107e-06, | |
| "loss": 0.9409, | |
| "step": 712 | |
| }, | |
| { | |
| "epoch": 0.869644762921177, | |
| "grad_norm": 0.023805683478713036, | |
| "learning_rate": 1.3532335421015685e-06, | |
| "loss": 0.9486, | |
| "step": 713 | |
| }, | |
| { | |
| "epoch": 0.8708644610458911, | |
| "grad_norm": 0.0226603876799345, | |
| "learning_rate": 1.328203181221398e-06, | |
| "loss": 0.986, | |
| "step": 714 | |
| }, | |
| { | |
| "epoch": 0.8720841591706052, | |
| "grad_norm": 0.022799277678132057, | |
| "learning_rate": 1.3033957488222425e-06, | |
| "loss": 0.9834, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.8733038572953195, | |
| "grad_norm": 0.023162079975008965, | |
| "learning_rate": 1.2788116494071333e-06, | |
| "loss": 0.9924, | |
| "step": 716 | |
| }, | |
| { | |
| "epoch": 0.8745235554200336, | |
| "grad_norm": 0.02272317185997963, | |
| "learning_rate": 1.254451283837486e-06, | |
| "loss": 0.9528, | |
| "step": 717 | |
| }, | |
| { | |
| "epoch": 0.8757432535447477, | |
| "grad_norm": 0.02237888239324093, | |
| "learning_rate": 1.230315049326593e-06, | |
| "loss": 0.9819, | |
| "step": 718 | |
| }, | |
| { | |
| "epoch": 0.8769629516694618, | |
| "grad_norm": 0.022658847272396088, | |
| "learning_rate": 1.2064033394331114e-06, | |
| "loss": 0.9863, | |
| "step": 719 | |
| }, | |
| { | |
| "epoch": 0.8781826497941759, | |
| "grad_norm": 0.022705573588609695, | |
| "learning_rate": 1.1827165440546705e-06, | |
| "loss": 0.9815, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.87940234791889, | |
| "grad_norm": 0.022633442655205727, | |
| "learning_rate": 1.1592550494215098e-06, | |
| "loss": 0.9921, | |
| "step": 721 | |
| }, | |
| { | |
| "epoch": 0.8806220460436042, | |
| "grad_norm": 0.022194115445017815, | |
| "learning_rate": 1.1360192380901723e-06, | |
| "loss": 0.9712, | |
| "step": 722 | |
| }, | |
| { | |
| "epoch": 0.8818417441683183, | |
| "grad_norm": 0.022446295246481895, | |
| "learning_rate": 1.1130094889372783e-06, | |
| "loss": 0.9865, | |
| "step": 723 | |
| }, | |
| { | |
| "epoch": 0.8830614422930325, | |
| "grad_norm": 0.02220536395907402, | |
| "learning_rate": 1.0902261771533345e-06, | |
| "loss": 0.9821, | |
| "step": 724 | |
| }, | |
| { | |
| "epoch": 0.8842811404177466, | |
| "grad_norm": 0.022694041952490807, | |
| "learning_rate": 1.0676696742366266e-06, | |
| "loss": 1.0006, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.8855008385424608, | |
| "grad_norm": 0.02287505939602852, | |
| "learning_rate": 1.045340347987165e-06, | |
| "loss": 0.9708, | |
| "step": 726 | |
| }, | |
| { | |
| "epoch": 0.8867205366671749, | |
| "grad_norm": 0.022034669294953346, | |
| "learning_rate": 1.0232385625006663e-06, | |
| "loss": 0.9773, | |
| "step": 727 | |
| }, | |
| { | |
| "epoch": 0.887940234791889, | |
| "grad_norm": 0.02193499729037285, | |
| "learning_rate": 1.001364678162649e-06, | |
| "loss": 0.9735, | |
| "step": 728 | |
| }, | |
| { | |
| "epoch": 0.8891599329166031, | |
| "grad_norm": 0.02207253687083721, | |
| "learning_rate": 9.797190516425269e-07, | |
| "loss": 0.9686, | |
| "step": 729 | |
| }, | |
| { | |
| "epoch": 0.8903796310413172, | |
| "grad_norm": 0.022487325593829155, | |
| "learning_rate": 9.583020358878081e-07, | |
| "loss": 0.9669, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.8915993291660315, | |
| "grad_norm": 0.021718062460422516, | |
| "learning_rate": 9.371139801183437e-07, | |
| "loss": 0.9377, | |
| "step": 731 | |
| }, | |
| { | |
| "epoch": 0.8928190272907456, | |
| "grad_norm": 0.022089695557951927, | |
| "learning_rate": 9.161552298206199e-07, | |
| "loss": 0.9682, | |
| "step": 732 | |
| }, | |
| { | |
| "epoch": 0.8940387254154597, | |
| "grad_norm": 0.022582467645406723, | |
| "learning_rate": 8.954261267421338e-07, | |
| "loss": 0.9669, | |
| "step": 733 | |
| }, | |
| { | |
| "epoch": 0.8952584235401738, | |
| "grad_norm": 0.022160224616527557, | |
| "learning_rate": 8.749270088858258e-07, | |
| "loss": 0.9728, | |
| "step": 734 | |
| }, | |
| { | |
| "epoch": 0.8964781216648879, | |
| "grad_norm": 0.021837202832102776, | |
| "learning_rate": 8.546582105045547e-07, | |
| "loss": 0.9801, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.8976978197896021, | |
| "grad_norm": 0.022184019908308983, | |
| "learning_rate": 8.346200620956585e-07, | |
| "loss": 0.9392, | |
| "step": 736 | |
| }, | |
| { | |
| "epoch": 0.8989175179143162, | |
| "grad_norm": 0.022868379950523376, | |
| "learning_rate": 8.148128903955582e-07, | |
| "loss": 0.9371, | |
| "step": 737 | |
| }, | |
| { | |
| "epoch": 0.9001372160390303, | |
| "grad_norm": 0.02238878607749939, | |
| "learning_rate": 7.952370183744346e-07, | |
| "loss": 0.9602, | |
| "step": 738 | |
| }, | |
| { | |
| "epoch": 0.9013569141637445, | |
| "grad_norm": 0.023037489503622055, | |
| "learning_rate": 7.758927652309667e-07, | |
| "loss": 0.9547, | |
| "step": 739 | |
| }, | |
| { | |
| "epoch": 0.9025766122884586, | |
| "grad_norm": 0.02285122685134411, | |
| "learning_rate": 7.56780446387108e-07, | |
| "loss": 0.9884, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.9037963104131728, | |
| "grad_norm": 0.02257378399372101, | |
| "learning_rate": 7.379003734829726e-07, | |
| "loss": 0.9759, | |
| "step": 741 | |
| }, | |
| { | |
| "epoch": 0.9050160085378869, | |
| "grad_norm": 0.021989768370985985, | |
| "learning_rate": 7.192528543717248e-07, | |
| "loss": 0.9539, | |
| "step": 742 | |
| }, | |
| { | |
| "epoch": 0.906235706662601, | |
| "grad_norm": 0.021790876984596252, | |
| "learning_rate": 7.008381931145758e-07, | |
| "loss": 0.9661, | |
| "step": 743 | |
| }, | |
| { | |
| "epoch": 0.9074554047873151, | |
| "grad_norm": 0.023033056408166885, | |
| "learning_rate": 6.826566899758257e-07, | |
| "loss": 0.9919, | |
| "step": 744 | |
| }, | |
| { | |
| "epoch": 0.9086751029120292, | |
| "grad_norm": 0.022525478154420853, | |
| "learning_rate": 6.647086414179593e-07, | |
| "loss": 0.9605, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.9098948010367434, | |
| "grad_norm": 0.021785186603665352, | |
| "learning_rate": 6.469943400968176e-07, | |
| "loss": 0.947, | |
| "step": 746 | |
| }, | |
| { | |
| "epoch": 0.9111144991614576, | |
| "grad_norm": 0.022115524858236313, | |
| "learning_rate": 6.295140748568257e-07, | |
| "loss": 0.9702, | |
| "step": 747 | |
| }, | |
| { | |
| "epoch": 0.9123341972861717, | |
| "grad_norm": 0.022994324564933777, | |
| "learning_rate": 6.122681307262812e-07, | |
| "loss": 0.9767, | |
| "step": 748 | |
| }, | |
| { | |
| "epoch": 0.9135538954108858, | |
| "grad_norm": 0.022172952070832253, | |
| "learning_rate": 5.952567889127098e-07, | |
| "loss": 0.9511, | |
| "step": 749 | |
| }, | |
| { | |
| "epoch": 0.9147735935356, | |
| "grad_norm": 0.021728791296482086, | |
| "learning_rate": 5.784803267982752e-07, | |
| "loss": 0.9725, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.9159932916603141, | |
| "grad_norm": 0.021961642429232597, | |
| "learning_rate": 5.619390179352552e-07, | |
| "loss": 0.9503, | |
| "step": 751 | |
| }, | |
| { | |
| "epoch": 0.9172129897850282, | |
| "grad_norm": 0.022464880719780922, | |
| "learning_rate": 5.456331320415941e-07, | |
| "loss": 0.9567, | |
| "step": 752 | |
| }, | |
| { | |
| "epoch": 0.9184326879097423, | |
| "grad_norm": 0.022449767217040062, | |
| "learning_rate": 5.295629349964859e-07, | |
| "loss": 0.9734, | |
| "step": 753 | |
| }, | |
| { | |
| "epoch": 0.9196523860344564, | |
| "grad_norm": 0.02229849062860012, | |
| "learning_rate": 5.137286888360548e-07, | |
| "loss": 0.969, | |
| "step": 754 | |
| }, | |
| { | |
| "epoch": 0.9208720841591707, | |
| "grad_norm": 0.02238483913242817, | |
| "learning_rate": 4.981306517490747e-07, | |
| "loss": 0.9554, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.9220917822838848, | |
| "grad_norm": 0.02232952043414116, | |
| "learning_rate": 4.827690780727551e-07, | |
| "loss": 0.9351, | |
| "step": 756 | |
| }, | |
| { | |
| "epoch": 0.9233114804085989, | |
| "grad_norm": 0.022212905809283257, | |
| "learning_rate": 4.676442182886098e-07, | |
| "loss": 0.9547, | |
| "step": 757 | |
| }, | |
| { | |
| "epoch": 0.924531178533313, | |
| "grad_norm": 0.022673193365335464, | |
| "learning_rate": 4.52756319018352e-07, | |
| "loss": 0.9661, | |
| "step": 758 | |
| }, | |
| { | |
| "epoch": 0.9257508766580271, | |
| "grad_norm": 0.02219342067837715, | |
| "learning_rate": 4.3810562301989045e-07, | |
| "loss": 0.9774, | |
| "step": 759 | |
| }, | |
| { | |
| "epoch": 0.9269705747827413, | |
| "grad_norm": 0.02298573963344097, | |
| "learning_rate": 4.2369236918336106e-07, | |
| "loss": 0.9562, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.9281902729074554, | |
| "grad_norm": 0.02159619890153408, | |
| "learning_rate": 4.0951679252723194e-07, | |
| "loss": 0.9411, | |
| "step": 761 | |
| }, | |
| { | |
| "epoch": 0.9294099710321695, | |
| "grad_norm": 0.022104347124695778, | |
| "learning_rate": 3.9557912419448284e-07, | |
| "loss": 0.9716, | |
| "step": 762 | |
| }, | |
| { | |
| "epoch": 0.9306296691568837, | |
| "grad_norm": 0.02196996659040451, | |
| "learning_rate": 3.818795914488166e-07, | |
| "loss": 0.9613, | |
| "step": 763 | |
| }, | |
| { | |
| "epoch": 0.9318493672815978, | |
| "grad_norm": 0.02267381362617016, | |
| "learning_rate": 3.684184176709671e-07, | |
| "loss": 0.9851, | |
| "step": 764 | |
| }, | |
| { | |
| "epoch": 0.933069065406312, | |
| "grad_norm": 0.022056207060813904, | |
| "learning_rate": 3.551958223550589e-07, | |
| "loss": 0.9752, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.9342887635310261, | |
| "grad_norm": 0.02254798822104931, | |
| "learning_rate": 3.422120211050167e-07, | |
| "loss": 0.969, | |
| "step": 766 | |
| }, | |
| { | |
| "epoch": 0.9355084616557402, | |
| "grad_norm": 0.023022811859846115, | |
| "learning_rate": 3.2946722563106325e-07, | |
| "loss": 0.9666, | |
| "step": 767 | |
| }, | |
| { | |
| "epoch": 0.9367281597804543, | |
| "grad_norm": 0.02215251512825489, | |
| "learning_rate": 3.1696164374625025e-07, | |
| "loss": 0.9675, | |
| "step": 768 | |
| }, | |
| { | |
| "epoch": 0.9379478579051684, | |
| "grad_norm": 0.022114263847470284, | |
| "learning_rate": 3.0469547936308796e-07, | |
| "loss": 0.962, | |
| "step": 769 | |
| }, | |
| { | |
| "epoch": 0.9391675560298826, | |
| "grad_norm": 0.022293198853731155, | |
| "learning_rate": 2.92668932490211e-07, | |
| "loss": 0.9712, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.9403872541545968, | |
| "grad_norm": 0.02169954776763916, | |
| "learning_rate": 2.808821992291127e-07, | |
| "loss": 0.9723, | |
| "step": 771 | |
| }, | |
| { | |
| "epoch": 0.9416069522793109, | |
| "grad_norm": 0.022570161148905754, | |
| "learning_rate": 2.6933547177096117e-07, | |
| "loss": 0.9547, | |
| "step": 772 | |
| }, | |
| { | |
| "epoch": 0.942826650404025, | |
| "grad_norm": 0.021904993802309036, | |
| "learning_rate": 2.580289383934481e-07, | |
| "loss": 0.9693, | |
| "step": 773 | |
| }, | |
| { | |
| "epoch": 0.9440463485287391, | |
| "grad_norm": 0.021927563473582268, | |
| "learning_rate": 2.4696278345773816e-07, | |
| "loss": 0.9452, | |
| "step": 774 | |
| }, | |
| { | |
| "epoch": 0.9452660466534533, | |
| "grad_norm": 0.022093456238508224, | |
| "learning_rate": 2.3613718740544466e-07, | |
| "loss": 0.9738, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.9464857447781674, | |
| "grad_norm": 0.021996039897203445, | |
| "learning_rate": 2.2555232675569693e-07, | |
| "loss": 0.9625, | |
| "step": 776 | |
| }, | |
| { | |
| "epoch": 0.9477054429028815, | |
| "grad_norm": 0.022086620330810547, | |
| "learning_rate": 2.1520837410226256e-07, | |
| "loss": 0.9559, | |
| "step": 777 | |
| }, | |
| { | |
| "epoch": 0.9489251410275956, | |
| "grad_norm": 0.02196883223950863, | |
| "learning_rate": 2.0510549811073142e-07, | |
| "loss": 0.9634, | |
| "step": 778 | |
| }, | |
| { | |
| "epoch": 0.9501448391523099, | |
| "grad_norm": 0.022001352161169052, | |
| "learning_rate": 1.952438635157594e-07, | |
| "loss": 0.959, | |
| "step": 779 | |
| }, | |
| { | |
| "epoch": 0.951364537277024, | |
| "grad_norm": 0.02179436944425106, | |
| "learning_rate": 1.8562363111839565e-07, | |
| "loss": 0.9807, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.9525842354017381, | |
| "grad_norm": 0.022099655121564865, | |
| "learning_rate": 1.7624495778344796e-07, | |
| "loss": 0.9724, | |
| "step": 781 | |
| }, | |
| { | |
| "epoch": 0.9538039335264522, | |
| "grad_norm": 0.02246328443288803, | |
| "learning_rate": 1.671079964369282e-07, | |
| "loss": 0.9742, | |
| "step": 782 | |
| }, | |
| { | |
| "epoch": 0.9550236316511663, | |
| "grad_norm": 0.02213803492486477, | |
| "learning_rate": 1.5821289606356758e-07, | |
| "loss": 0.9759, | |
| "step": 783 | |
| }, | |
| { | |
| "epoch": 0.9562433297758804, | |
| "grad_norm": 0.022068796679377556, | |
| "learning_rate": 1.49559801704372e-07, | |
| "loss": 0.9662, | |
| "step": 784 | |
| }, | |
| { | |
| "epoch": 0.9574630279005946, | |
| "grad_norm": 0.02178051881492138, | |
| "learning_rate": 1.4114885445427218e-07, | |
| "loss": 0.9772, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.9586827260253087, | |
| "grad_norm": 0.02281547151505947, | |
| "learning_rate": 1.3298019145981223e-07, | |
| "loss": 0.9716, | |
| "step": 786 | |
| }, | |
| { | |
| "epoch": 0.9599024241500229, | |
| "grad_norm": 0.021949436515569687, | |
| "learning_rate": 1.2505394591691477e-07, | |
| "loss": 0.9413, | |
| "step": 787 | |
| }, | |
| { | |
| "epoch": 0.961122122274737, | |
| "grad_norm": 0.02194429375231266, | |
| "learning_rate": 1.1737024706871769e-07, | |
| "loss": 0.9228, | |
| "step": 788 | |
| }, | |
| { | |
| "epoch": 0.9623418203994512, | |
| "grad_norm": 0.02197294682264328, | |
| "learning_rate": 1.0992922020345742e-07, | |
| "loss": 0.94, | |
| "step": 789 | |
| }, | |
| { | |
| "epoch": 0.9635615185241653, | |
| "grad_norm": 0.02181713841855526, | |
| "learning_rate": 1.0273098665242731e-07, | |
| "loss": 0.9773, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.9647812166488794, | |
| "grad_norm": 0.02181742526590824, | |
| "learning_rate": 9.577566378800585e-08, | |
| "loss": 0.955, | |
| "step": 791 | |
| }, | |
| { | |
| "epoch": 0.9660009147735935, | |
| "grad_norm": 0.02232934720814228, | |
| "learning_rate": 8.906336502173151e-08, | |
| "loss": 0.9687, | |
| "step": 792 | |
| }, | |
| { | |
| "epoch": 0.9672206128983076, | |
| "grad_norm": 0.02299078367650509, | |
| "learning_rate": 8.259419980246596e-08, | |
| "loss": 0.9765, | |
| "step": 793 | |
| }, | |
| { | |
| "epoch": 0.9684403110230217, | |
| "grad_norm": 0.021996477618813515, | |
| "learning_rate": 7.636827361460207e-08, | |
| "loss": 0.9644, | |
| "step": 794 | |
| }, | |
| { | |
| "epoch": 0.969660009147736, | |
| "grad_norm": 0.02329733408987522, | |
| "learning_rate": 7.038568797634204e-08, | |
| "loss": 0.9787, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.9708797072724501, | |
| "grad_norm": 0.022560199722647667, | |
| "learning_rate": 6.46465404380503e-08, | |
| "loss": 0.9733, | |
| "step": 796 | |
| }, | |
| { | |
| "epoch": 0.9720994053971642, | |
| "grad_norm": 0.02261877991259098, | |
| "learning_rate": 5.915092458065319e-08, | |
| "loss": 0.9784, | |
| "step": 797 | |
| }, | |
| { | |
| "epoch": 0.9733191035218783, | |
| "grad_norm": 0.021751757711172104, | |
| "learning_rate": 5.389893001412183e-08, | |
| "loss": 0.9665, | |
| "step": 798 | |
| }, | |
| { | |
| "epoch": 0.9745388016465925, | |
| "grad_norm": 0.021826745942234993, | |
| "learning_rate": 4.8890642376004936e-08, | |
| "loss": 0.9437, | |
| "step": 799 | |
| }, | |
| { | |
| "epoch": 0.9757584997713066, | |
| "grad_norm": 0.022791143506765366, | |
| "learning_rate": 4.412614333003495e-08, | |
| "loss": 0.9505, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.9769781978960207, | |
| "grad_norm": 0.021623898297548294, | |
| "learning_rate": 3.960551056479578e-08, | |
| "loss": 0.9789, | |
| "step": 801 | |
| }, | |
| { | |
| "epoch": 0.9781978960207348, | |
| "grad_norm": 0.021686095744371414, | |
| "learning_rate": 3.532881779245545e-08, | |
| "loss": 0.9702, | |
| "step": 802 | |
| }, | |
| { | |
| "epoch": 0.979417594145449, | |
| "grad_norm": 0.022227788344025612, | |
| "learning_rate": 3.129613474756709e-08, | |
| "loss": 0.9719, | |
| "step": 803 | |
| }, | |
| { | |
| "epoch": 0.9806372922701632, | |
| "grad_norm": 0.02224772423505783, | |
| "learning_rate": 2.750752718592653e-08, | |
| "loss": 0.9661, | |
| "step": 804 | |
| }, | |
| { | |
| "epoch": 0.9818569903948773, | |
| "grad_norm": 0.022183561697602272, | |
| "learning_rate": 2.3963056883508084e-08, | |
| "loss": 0.9732, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.9830766885195914, | |
| "grad_norm": 0.022314051166176796, | |
| "learning_rate": 2.0662781635447102e-08, | |
| "loss": 0.9544, | |
| "step": 806 | |
| }, | |
| { | |
| "epoch": 0.9842963866443055, | |
| "grad_norm": 0.02197330817580223, | |
| "learning_rate": 1.760675525510569e-08, | |
| "loss": 0.9818, | |
| "step": 807 | |
| }, | |
| { | |
| "epoch": 0.9855160847690196, | |
| "grad_norm": 0.02160177007317543, | |
| "learning_rate": 1.4795027573193397e-08, | |
| "loss": 0.9606, | |
| "step": 808 | |
| }, | |
| { | |
| "epoch": 0.9867357828937338, | |
| "grad_norm": 0.02266581542789936, | |
| "learning_rate": 1.222764443694957e-08, | |
| "loss": 0.9805, | |
| "step": 809 | |
| }, | |
| { | |
| "epoch": 0.987955481018448, | |
| "grad_norm": 0.02238692156970501, | |
| "learning_rate": 9.904647709407244e-09, | |
| "loss": 0.9695, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.9891751791431621, | |
| "grad_norm": 0.022915730252861977, | |
| "learning_rate": 7.826075268695387e-09, | |
| "loss": 0.9664, | |
| "step": 811 | |
| }, | |
| { | |
| "epoch": 0.9903948772678762, | |
| "grad_norm": 0.022579343989491463, | |
| "learning_rate": 5.991961007434377e-09, | |
| "loss": 0.982, | |
| "step": 812 | |
| }, | |
| { | |
| "epoch": 0.9916145753925903, | |
| "grad_norm": 0.02270609140396118, | |
| "learning_rate": 4.402334832174782e-09, | |
| "loss": 0.9717, | |
| "step": 813 | |
| }, | |
| { | |
| "epoch": 0.9928342735173045, | |
| "grad_norm": 0.02186567708849907, | |
| "learning_rate": 3.057222662917747e-09, | |
| "loss": 0.9574, | |
| "step": 814 | |
| }, | |
| { | |
| "epoch": 0.9940539716420186, | |
| "grad_norm": 0.022045252844691277, | |
| "learning_rate": 1.9566464326820077e-09, | |
| "loss": 0.9369, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.9952736697667327, | |
| "grad_norm": 0.02271999791264534, | |
| "learning_rate": 1.1006240871558326e-09, | |
| "loss": 0.9674, | |
| "step": 816 | |
| }, | |
| { | |
| "epoch": 0.9964933678914468, | |
| "grad_norm": 0.022212740033864975, | |
| "learning_rate": 4.89169584400595e-10, | |
| "loss": 0.9698, | |
| "step": 817 | |
| }, | |
| { | |
| "epoch": 0.997713066016161, | |
| "grad_norm": 0.02206205390393734, | |
| "learning_rate": 1.2229289461762604e-10, | |
| "loss": 0.9636, | |
| "step": 818 | |
| }, | |
| { | |
| "epoch": 0.9989327641408752, | |
| "grad_norm": 0.021807022392749786, | |
| "learning_rate": 0.0, | |
| "loss": 0.942, | |
| "step": 819 | |
| }, | |
| { | |
| "epoch": 0.9989327641408752, | |
| "step": 819, | |
| "total_flos": 5766189284327424.0, | |
| "train_loss": 1.0495533346402048, | |
| "train_runtime": 174070.54, | |
| "train_samples_per_second": 0.603, | |
| "train_steps_per_second": 0.005 | |
| } | |
| ], | |
| "logging_steps": 1, | |
| "max_steps": 819, | |
| "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": 5766189284327424.0, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |