| { |
| "epoch": 0.3068119932808174, |
| "global_step": 10000, |
| "max_steps": 10000, |
| "logging_steps": 20, |
| "eval_steps": 500, |
| "save_steps": 500, |
| "train_batch_size": 32, |
| "num_train_epochs": 1, |
| "num_input_tokens_seen": 0, |
| "total_flos": 4.748331528373666e+18, |
| "log_history": [ |
| { |
| "loss": 10.1429, |
| "grad_norm": 24.328838348388672, |
| "learning_rate": 1e-05, |
| "epoch": 0.0006136239865616347, |
| "step": 20 |
| }, |
| { |
| "loss": 2.0091, |
| "grad_norm": 9.998310089111328, |
| "learning_rate": 2e-05, |
| "epoch": 0.0012272479731232694, |
| "step": 40 |
| }, |
| { |
| "loss": 0.4768, |
| "grad_norm": 1.95796799659729, |
| "learning_rate": 3e-05, |
| "epoch": 0.001840871959684904, |
| "step": 60 |
| }, |
| { |
| "loss": 0.397, |
| "grad_norm": 0.9538208842277527, |
| "learning_rate": 4e-05, |
| "epoch": 0.0024544959462465387, |
| "step": 80 |
| }, |
| { |
| "loss": 0.3703, |
| "grad_norm": 3.8885602951049805, |
| "learning_rate": 5e-05, |
| "epoch": 0.0030681199328081734, |
| "step": 100 |
| }, |
| { |
| "loss": 0.377, |
| "grad_norm": 1.2746047973632812, |
| "learning_rate": 6e-05, |
| "epoch": 0.003681743919369808, |
| "step": 120 |
| }, |
| { |
| "loss": 0.3449, |
| "grad_norm": 0.5722544193267822, |
| "learning_rate": 7e-05, |
| "epoch": 0.004295367905931443, |
| "step": 140 |
| }, |
| { |
| "loss": 0.3432, |
| "grad_norm": 0.4266990125179291, |
| "learning_rate": 8e-05, |
| "epoch": 0.004908991892493077, |
| "step": 160 |
| }, |
| { |
| "loss": 0.3447, |
| "grad_norm": 0.940947949886322, |
| "learning_rate": 9e-05, |
| "epoch": 0.0055226158790547125, |
| "step": 180 |
| }, |
| { |
| "loss": 0.3363, |
| "grad_norm": 1.3794434070587158, |
| "learning_rate": 0.0001, |
| "epoch": 0.006136239865616347, |
| "step": 200 |
| }, |
| { |
| "loss": 0.3097, |
| "grad_norm": 0.6603377461433411, |
| "learning_rate": 9.99989723479183e-05, |
| "epoch": 0.006749863852177982, |
| "step": 220 |
| }, |
| { |
| "loss": 0.3234, |
| "grad_norm": 0.5309205055236816, |
| "learning_rate": 9.999588943391597e-05, |
| "epoch": 0.007363487838739616, |
| "step": 240 |
| }, |
| { |
| "loss": 0.322, |
| "grad_norm": 0.4918404817581177, |
| "learning_rate": 9.999075138471951e-05, |
| "epoch": 0.007977111825301251, |
| "step": 260 |
| }, |
| { |
| "loss": 0.3061, |
| "grad_norm": 0.23719413578510284, |
| "learning_rate": 9.9983558411534e-05, |
| "epoch": 0.008590735811862886, |
| "step": 280 |
| }, |
| { |
| "loss": 0.3141, |
| "grad_norm": 0.5652378797531128, |
| "learning_rate": 9.99743108100344e-05, |
| "epoch": 0.00920435979842452, |
| "step": 300 |
| }, |
| { |
| "loss": 0.3007, |
| "grad_norm": 0.3501427471637726, |
| "learning_rate": 9.996300896035339e-05, |
| "epoch": 0.009817983784986155, |
| "step": 320 |
| }, |
| { |
| "loss": 0.2783, |
| "grad_norm": 2.7963829040527344, |
| "learning_rate": 9.994965332706573e-05, |
| "epoch": 0.01043160777154779, |
| "step": 340 |
| }, |
| { |
| "loss": 0.2972, |
| "grad_norm": 0.4121073782444, |
| "learning_rate": 9.993424445916923e-05, |
| "epoch": 0.011045231758109425, |
| "step": 360 |
| }, |
| { |
| "loss": 0.2799, |
| "grad_norm": 0.3509398400783539, |
| "learning_rate": 9.991678299006205e-05, |
| "epoch": 0.011658855744671058, |
| "step": 380 |
| }, |
| { |
| "loss": 0.268, |
| "grad_norm": 0.25531092286109924, |
| "learning_rate": 9.989726963751682e-05, |
| "epoch": 0.012272479731232694, |
| "step": 400 |
| }, |
| { |
| "loss": 0.2601, |
| "grad_norm": 0.4110245108604431, |
| "learning_rate": 9.987570520365104e-05, |
| "epoch": 0.012886103717794329, |
| "step": 420 |
| }, |
| { |
| "loss": 0.2672, |
| "grad_norm": 0.6753450632095337, |
| "learning_rate": 9.98520905748941e-05, |
| "epoch": 0.013499727704355964, |
| "step": 440 |
| }, |
| { |
| "loss": 0.2618, |
| "grad_norm": 0.6639530062675476, |
| "learning_rate": 9.982642672195092e-05, |
| "epoch": 0.014113351690917597, |
| "step": 460 |
| }, |
| { |
| "loss": 0.2562, |
| "grad_norm": 1.2062132358551025, |
| "learning_rate": 9.979871469976196e-05, |
| "epoch": 0.014726975677479232, |
| "step": 480 |
| }, |
| { |
| "loss": 0.2404, |
| "grad_norm": 0.3097586929798126, |
| "learning_rate": 9.976895564745991e-05, |
| "epoch": 0.015340599664040867, |
| "step": 500 |
| }, |
| { |
| "eval_loss": 0.23828153312206268, |
| "eval_runtime": 2767.3744, |
| "eval_samples_per_second": 161.937, |
| "eval_steps_per_second": 2.531, |
| "epoch": 0.015340599664040867, |
| "step": 500 |
| }, |
| { |
| "loss": 0.2348, |
| "grad_norm": 0.3366382122039795, |
| "learning_rate": 9.973715078832288e-05, |
| "epoch": 0.015954223650602502, |
| "step": 520 |
| }, |
| { |
| "loss": 0.2475, |
| "grad_norm": 0.23378945887088776, |
| "learning_rate": 9.970330142972401e-05, |
| "epoch": 0.016567847637164138, |
| "step": 540 |
| }, |
| { |
| "loss": 0.2499, |
| "grad_norm": 0.7393000721931458, |
| "learning_rate": 9.966740896307791e-05, |
| "epoch": 0.017181471623725773, |
| "step": 560 |
| }, |
| { |
| "loss": 0.2479, |
| "grad_norm": 0.40257546305656433, |
| "learning_rate": 9.962947486378326e-05, |
| "epoch": 0.017795095610287408, |
| "step": 580 |
| }, |
| { |
| "loss": 0.227, |
| "grad_norm": 0.2757706046104431, |
| "learning_rate": 9.95895006911623e-05, |
| "epoch": 0.01840871959684904, |
| "step": 600 |
| }, |
| { |
| "loss": 0.2199, |
| "grad_norm": 0.32403552532196045, |
| "learning_rate": 9.954748808839674e-05, |
| "epoch": 0.019022343583410675, |
| "step": 620 |
| }, |
| { |
| "loss": 0.2315, |
| "grad_norm": 0.6505448818206787, |
| "learning_rate": 9.95034387824601e-05, |
| "epoch": 0.01963596756997231, |
| "step": 640 |
| }, |
| { |
| "loss": 0.2247, |
| "grad_norm": 0.4369623363018036, |
| "learning_rate": 9.945735458404681e-05, |
| "epoch": 0.020249591556533945, |
| "step": 660 |
| }, |
| { |
| "loss": 0.2286, |
| "grad_norm": 0.43412795662879944, |
| "learning_rate": 9.940923738749778e-05, |
| "epoch": 0.02086321554309558, |
| "step": 680 |
| }, |
| { |
| "loss": 0.2202, |
| "grad_norm": 0.5289716720581055, |
| "learning_rate": 9.935908917072252e-05, |
| "epoch": 0.021476839529657215, |
| "step": 700 |
| }, |
| { |
| "loss": 0.2124, |
| "grad_norm": 0.552556037902832, |
| "learning_rate": 9.930691199511775e-05, |
| "epoch": 0.02209046351621885, |
| "step": 720 |
| }, |
| { |
| "loss": 0.2174, |
| "grad_norm": 0.3651241958141327, |
| "learning_rate": 9.925270800548285e-05, |
| "epoch": 0.022704087502780485, |
| "step": 740 |
| }, |
| { |
| "loss": 0.2245, |
| "grad_norm": 0.46145933866500854, |
| "learning_rate": 9.919647942993148e-05, |
| "epoch": 0.023317711489342117, |
| "step": 760 |
| }, |
| { |
| "loss": 0.223, |
| "grad_norm": 0.37703219056129456, |
| "learning_rate": 9.91382285798002e-05, |
| "epoch": 0.023931335475903752, |
| "step": 780 |
| }, |
| { |
| "loss": 0.2117, |
| "grad_norm": 0.3304797112941742, |
| "learning_rate": 9.907795784955327e-05, |
| "epoch": 0.024544959462465387, |
| "step": 800 |
| }, |
| { |
| "loss": 0.2091, |
| "grad_norm": 0.32046860456466675, |
| "learning_rate": 9.901566971668437e-05, |
| "epoch": 0.025158583449027022, |
| "step": 820 |
| }, |
| { |
| "loss": 0.201, |
| "grad_norm": 0.516817569732666, |
| "learning_rate": 9.895136674161465e-05, |
| "epoch": 0.025772207435588657, |
| "step": 840 |
| }, |
| { |
| "loss": 0.2117, |
| "grad_norm": 0.35681387782096863, |
| "learning_rate": 9.888505156758759e-05, |
| "epoch": 0.026385831422150292, |
| "step": 860 |
| }, |
| { |
| "loss": 0.1972, |
| "grad_norm": 0.6307485103607178, |
| "learning_rate": 9.881672692056021e-05, |
| "epoch": 0.026999455408711927, |
| "step": 880 |
| }, |
| { |
| "loss": 0.2081, |
| "grad_norm": 0.5634998083114624, |
| "learning_rate": 9.874639560909117e-05, |
| "epoch": 0.027613079395273563, |
| "step": 900 |
| }, |
| { |
| "loss": 0.2022, |
| "grad_norm": 0.38356727361679077, |
| "learning_rate": 9.867406052422524e-05, |
| "epoch": 0.028226703381835194, |
| "step": 920 |
| }, |
| { |
| "loss": 0.1949, |
| "grad_norm": 0.6488237977027893, |
| "learning_rate": 9.859972463937441e-05, |
| "epoch": 0.02884032736839683, |
| "step": 940 |
| }, |
| { |
| "loss": 0.2012, |
| "grad_norm": 0.4440819025039673, |
| "learning_rate": 9.852339101019574e-05, |
| "epoch": 0.029453951354958464, |
| "step": 960 |
| }, |
| { |
| "loss": 0.2108, |
| "grad_norm": 0.6403115391731262, |
| "learning_rate": 9.844506277446577e-05, |
| "epoch": 0.0300675753415201, |
| "step": 980 |
| }, |
| { |
| "loss": 0.1907, |
| "grad_norm": 0.41512686014175415, |
| "learning_rate": 9.836474315195147e-05, |
| "epoch": 0.030681199328081735, |
| "step": 1000 |
| }, |
| { |
| "eval_loss": 0.19328325986862183, |
| "eval_runtime": 2767.2068, |
| "eval_samples_per_second": 161.947, |
| "eval_steps_per_second": 2.531, |
| "epoch": 0.030681199328081735, |
| "step": 1000 |
| }, |
| { |
| "loss": 0.1993, |
| "grad_norm": 0.44694626331329346, |
| "learning_rate": 9.828243544427796e-05, |
| "epoch": 0.031294823314643366, |
| "step": 1020 |
| }, |
| { |
| "loss": 0.1997, |
| "grad_norm": 0.3413143754005432, |
| "learning_rate": 9.819814303479267e-05, |
| "epoch": 0.031908447301205005, |
| "step": 1040 |
| }, |
| { |
| "loss": 0.1823, |
| "grad_norm": 0.40824049711227417, |
| "learning_rate": 9.811186938842645e-05, |
| "epoch": 0.032522071287766637, |
| "step": 1060 |
| }, |
| { |
| "loss": 0.2114, |
| "grad_norm": 1.721802830696106, |
| "learning_rate": 9.802361805155097e-05, |
| "epoch": 0.033135695274328275, |
| "step": 1080 |
| }, |
| { |
| "loss": 0.2016, |
| "grad_norm": 0.4164980351924896, |
| "learning_rate": 9.793339265183303e-05, |
| "epoch": 0.03374931926088991, |
| "step": 1100 |
| }, |
| { |
| "loss": 0.1798, |
| "grad_norm": 1.188069224357605, |
| "learning_rate": 9.784119689808544e-05, |
| "epoch": 0.034362943247451545, |
| "step": 1120 |
| }, |
| { |
| "loss": 0.1893, |
| "grad_norm": 0.41302207112312317, |
| "learning_rate": 9.774703458011453e-05, |
| "epoch": 0.03497656723401318, |
| "step": 1140 |
| }, |
| { |
| "loss": 0.1864, |
| "grad_norm": 0.5009692907333374, |
| "learning_rate": 9.765090956856436e-05, |
| "epoch": 0.035590191220574816, |
| "step": 1160 |
| }, |
| { |
| "loss": 0.1823, |
| "grad_norm": 0.5319084525108337, |
| "learning_rate": 9.755282581475769e-05, |
| "epoch": 0.03620381520713645, |
| "step": 1180 |
| }, |
| { |
| "loss": 0.1885, |
| "grad_norm": 0.286171555519104, |
| "learning_rate": 9.745278735053343e-05, |
| "epoch": 0.03681743919369808, |
| "step": 1200 |
| }, |
| { |
| "loss": 0.1831, |
| "grad_norm": 0.38070616126060486, |
| "learning_rate": 9.735079828808107e-05, |
| "epoch": 0.03743106318025972, |
| "step": 1220 |
| }, |
| { |
| "loss": 0.1815, |
| "grad_norm": 0.37657469511032104, |
| "learning_rate": 9.724686281977146e-05, |
| "epoch": 0.03804468716682135, |
| "step": 1240 |
| }, |
| { |
| "loss": 0.174, |
| "grad_norm": 0.3888895511627197, |
| "learning_rate": 9.714098521798465e-05, |
| "epoch": 0.03865831115338299, |
| "step": 1260 |
| }, |
| { |
| "loss": 0.1789, |
| "grad_norm": 0.3658376932144165, |
| "learning_rate": 9.703316983493414e-05, |
| "epoch": 0.03927193513994462, |
| "step": 1280 |
| }, |
| { |
| "loss": 0.1801, |
| "grad_norm": 0.5093252062797546, |
| "learning_rate": 9.692342110248802e-05, |
| "epoch": 0.03988555912650626, |
| "step": 1300 |
| }, |
| { |
| "loss": 0.1847, |
| "grad_norm": 0.6297820806503296, |
| "learning_rate": 9.681174353198687e-05, |
| "epoch": 0.04049918311306789, |
| "step": 1320 |
| }, |
| { |
| "loss": 0.1721, |
| "grad_norm": 1.3242509365081787, |
| "learning_rate": 9.669814171405816e-05, |
| "epoch": 0.04111280709962952, |
| "step": 1340 |
| }, |
| { |
| "loss": 0.1799, |
| "grad_norm": 0.3201231360435486, |
| "learning_rate": 9.65826203184277e-05, |
| "epoch": 0.04172643108619116, |
| "step": 1360 |
| }, |
| { |
| "loss": 0.1776, |
| "grad_norm": 0.6388235688209534, |
| "learning_rate": 9.64651840937276e-05, |
| "epoch": 0.04234005507275279, |
| "step": 1380 |
| }, |
| { |
| "loss": 0.1883, |
| "grad_norm": 0.379491925239563, |
| "learning_rate": 9.63458378673011e-05, |
| "epoch": 0.04295367905931443, |
| "step": 1400 |
| }, |
| { |
| "loss": 0.1673, |
| "grad_norm": 0.29706379771232605, |
| "learning_rate": 9.622458654500409e-05, |
| "epoch": 0.04356730304587606, |
| "step": 1420 |
| }, |
| { |
| "loss": 0.1757, |
| "grad_norm": 0.31686845421791077, |
| "learning_rate": 9.610143511100354e-05, |
| "epoch": 0.0441809270324377, |
| "step": 1440 |
| }, |
| { |
| "loss": 0.175, |
| "grad_norm": 0.24471138417720795, |
| "learning_rate": 9.597638862757255e-05, |
| "epoch": 0.04479455101899933, |
| "step": 1460 |
| }, |
| { |
| "loss": 0.1823, |
| "grad_norm": 0.7488877177238464, |
| "learning_rate": 9.584945223488227e-05, |
| "epoch": 0.04540817500556097, |
| "step": 1480 |
| }, |
| { |
| "loss": 0.174, |
| "grad_norm": 0.3987618386745453, |
| "learning_rate": 9.572063115079063e-05, |
| "epoch": 0.0460217989921226, |
| "step": 1500 |
| }, |
| { |
| "eval_loss": 0.1716459095478058, |
| "eval_runtime": 2769.3725, |
| "eval_samples_per_second": 161.82, |
| "eval_steps_per_second": 2.529, |
| "epoch": 0.0460217989921226, |
| "step": 1500 |
| }, |
| { |
| "loss": 0.1752, |
| "grad_norm": 0.44050776958465576, |
| "learning_rate": 9.558993067062785e-05, |
| "epoch": 0.046635422978684234, |
| "step": 1520 |
| }, |
| { |
| "loss": 0.1733, |
| "grad_norm": 0.3778401017189026, |
| "learning_rate": 9.545735616697875e-05, |
| "epoch": 0.04724904696524587, |
| "step": 1540 |
| }, |
| { |
| "loss": 0.1643, |
| "grad_norm": 0.2891245484352112, |
| "learning_rate": 9.53229130894619e-05, |
| "epoch": 0.047862670951807504, |
| "step": 1560 |
| }, |
| { |
| "loss": 0.1734, |
| "grad_norm": 0.9185692667961121, |
| "learning_rate": 9.518660696450568e-05, |
| "epoch": 0.04847629493836914, |
| "step": 1580 |
| }, |
| { |
| "loss": 0.1786, |
| "grad_norm": 0.4238841235637665, |
| "learning_rate": 9.504844339512095e-05, |
| "epoch": 0.049089918924930774, |
| "step": 1600 |
| }, |
| { |
| "loss": 0.1752, |
| "grad_norm": 0.24845579266548157, |
| "learning_rate": 9.490842806067095e-05, |
| "epoch": 0.04970354291149241, |
| "step": 1620 |
| }, |
| { |
| "loss": 0.1817, |
| "grad_norm": 0.3146928548812866, |
| "learning_rate": 9.476656671663765e-05, |
| "epoch": 0.050317166898054044, |
| "step": 1640 |
| }, |
| { |
| "loss": 0.1647, |
| "grad_norm": 0.26675552129745483, |
| "learning_rate": 9.46228651943853e-05, |
| "epoch": 0.05093079088461568, |
| "step": 1660 |
| }, |
| { |
| "loss": 0.169, |
| "grad_norm": 0.47638726234436035, |
| "learning_rate": 9.44773294009206e-05, |
| "epoch": 0.051544414871177315, |
| "step": 1680 |
| }, |
| { |
| "loss": 0.1652, |
| "grad_norm": 0.5169867873191833, |
| "learning_rate": 9.432996531865002e-05, |
| "epoch": 0.052158038857738946, |
| "step": 1700 |
| }, |
| { |
| "loss": 0.1633, |
| "grad_norm": 0.44044116139411926, |
| "learning_rate": 9.418077900513377e-05, |
| "epoch": 0.052771662844300585, |
| "step": 1720 |
| }, |
| { |
| "loss": 0.1622, |
| "grad_norm": 0.24241219460964203, |
| "learning_rate": 9.40297765928369e-05, |
| "epoch": 0.053385286830862216, |
| "step": 1740 |
| }, |
| { |
| "loss": 0.1655, |
| "grad_norm": 0.21895462274551392, |
| "learning_rate": 9.387696428887716e-05, |
| "epoch": 0.053998910817423855, |
| "step": 1760 |
| }, |
| { |
| "loss": 0.1658, |
| "grad_norm": 0.2668932378292084, |
| "learning_rate": 9.372234837476978e-05, |
| "epoch": 0.05461253480398549, |
| "step": 1780 |
| }, |
| { |
| "loss": 0.1708, |
| "grad_norm": 0.2590928375720978, |
| "learning_rate": 9.356593520616948e-05, |
| "epoch": 0.055226158790547125, |
| "step": 1800 |
| }, |
| { |
| "loss": 0.1692, |
| "grad_norm": 1.094123125076294, |
| "learning_rate": 9.340773121260893e-05, |
| "epoch": 0.05583978277710876, |
| "step": 1820 |
| }, |
| { |
| "loss": 0.1609, |
| "grad_norm": 0.5052621364593506, |
| "learning_rate": 9.324774289723468e-05, |
| "epoch": 0.05645340676367039, |
| "step": 1840 |
| }, |
| { |
| "loss": 0.1665, |
| "grad_norm": 0.35525742173194885, |
| "learning_rate": 9.308597683653975e-05, |
| "epoch": 0.05706703075023203, |
| "step": 1860 |
| }, |
| { |
| "loss": 0.1644, |
| "grad_norm": 0.5005172491073608, |
| "learning_rate": 9.292243968009331e-05, |
| "epoch": 0.05768065473679366, |
| "step": 1880 |
| }, |
| { |
| "loss": 0.1697, |
| "grad_norm": 0.23915395140647888, |
| "learning_rate": 9.275713815026731e-05, |
| "epoch": 0.0582942787233553, |
| "step": 1900 |
| }, |
| { |
| "loss": 0.1616, |
| "grad_norm": 0.23820538818836212, |
| "learning_rate": 9.259007904196023e-05, |
| "epoch": 0.05890790270991693, |
| "step": 1920 |
| }, |
| { |
| "loss": 0.1562, |
| "grad_norm": 0.19122985005378723, |
| "learning_rate": 9.242126922231763e-05, |
| "epoch": 0.05952152669647857, |
| "step": 1940 |
| }, |
| { |
| "loss": 0.1568, |
| "grad_norm": 0.24955634772777557, |
| "learning_rate": 9.225071563045007e-05, |
| "epoch": 0.0601351506830402, |
| "step": 1960 |
| }, |
| { |
| "loss": 0.1559, |
| "grad_norm": 0.2041475921869278, |
| "learning_rate": 9.207842527714767e-05, |
| "epoch": 0.06074877466960184, |
| "step": 1980 |
| }, |
| { |
| "loss": 0.1611, |
| "grad_norm": 0.38136520981788635, |
| "learning_rate": 9.190440524459203e-05, |
| "epoch": 0.06136239865616347, |
| "step": 2000 |
| }, |
| { |
| "eval_loss": 0.15973800420761108, |
| "eval_runtime": 2770.9129, |
| "eval_samples_per_second": 161.73, |
| "eval_steps_per_second": 2.527, |
| "epoch": 0.06136239865616347, |
| "step": 2000 |
| }, |
| { |
| "loss": 0.1515, |
| "grad_norm": 0.27136150002479553, |
| "learning_rate": 9.172866268606513e-05, |
| "epoch": 0.0619760226427251, |
| "step": 2020 |
| }, |
| { |
| "loss": 0.1589, |
| "grad_norm": 0.2267642617225647, |
| "learning_rate": 9.155120482565521e-05, |
| "epoch": 0.06258964662928673, |
| "step": 2040 |
| }, |
| { |
| "loss": 0.1527, |
| "grad_norm": 0.23289571702480316, |
| "learning_rate": 9.137203895795983e-05, |
| "epoch": 0.06320327061584838, |
| "step": 2060 |
| }, |
| { |
| "loss": 0.1636, |
| "grad_norm": 0.7123669385910034, |
| "learning_rate": 9.119117244778607e-05, |
| "epoch": 0.06381689460241001, |
| "step": 2080 |
| }, |
| { |
| "loss": 0.1559, |
| "grad_norm": 0.28865331411361694, |
| "learning_rate": 9.10086127298478e-05, |
| "epoch": 0.06443051858897164, |
| "step": 2100 |
| }, |
| { |
| "loss": 0.1614, |
| "grad_norm": 0.35813504457473755, |
| "learning_rate": 9.082436730845993e-05, |
| "epoch": 0.06504414257553327, |
| "step": 2120 |
| }, |
| { |
| "loss": 0.1567, |
| "grad_norm": 0.3127194344997406, |
| "learning_rate": 9.063844375723014e-05, |
| "epoch": 0.06565776656209492, |
| "step": 2140 |
| }, |
| { |
| "loss": 0.1613, |
| "grad_norm": 0.4829489588737488, |
| "learning_rate": 9.045084971874738e-05, |
| "epoch": 0.06627139054865655, |
| "step": 2160 |
| }, |
| { |
| "loss": 0.158, |
| "grad_norm": 0.5470940470695496, |
| "learning_rate": 9.02615929042678e-05, |
| "epoch": 0.06688501453521818, |
| "step": 2180 |
| }, |
| { |
| "loss": 0.1536, |
| "grad_norm": 0.2029639482498169, |
| "learning_rate": 9.007068109339784e-05, |
| "epoch": 0.06749863852177981, |
| "step": 2200 |
| }, |
| { |
| "loss": 0.159, |
| "grad_norm": 0.45802491903305054, |
| "learning_rate": 8.987812213377424e-05, |
| "epoch": 0.06811226250834145, |
| "step": 2220 |
| }, |
| { |
| "loss": 0.1571, |
| "grad_norm": 0.3373512923717499, |
| "learning_rate": 8.968392394074164e-05, |
| "epoch": 0.06872588649490309, |
| "step": 2240 |
| }, |
| { |
| "loss": 0.1547, |
| "grad_norm": 0.2957380712032318, |
| "learning_rate": 8.948809449702711e-05, |
| "epoch": 0.06933951048146472, |
| "step": 2260 |
| }, |
| { |
| "loss": 0.1566, |
| "grad_norm": 0.45103269815444946, |
| "learning_rate": 8.929064185241213e-05, |
| "epoch": 0.06995313446802635, |
| "step": 2280 |
| }, |
| { |
| "loss": 0.1544, |
| "grad_norm": 0.21954606473445892, |
| "learning_rate": 8.90915741234015e-05, |
| "epoch": 0.07056675845458799, |
| "step": 2300 |
| }, |
| { |
| "loss": 0.1539, |
| "grad_norm": 0.3216409385204315, |
| "learning_rate": 8.889089949288986e-05, |
| "epoch": 0.07118038244114963, |
| "step": 2320 |
| }, |
| { |
| "loss": 0.1515, |
| "grad_norm": 0.32575568556785583, |
| "learning_rate": 8.868862620982534e-05, |
| "epoch": 0.07179400642771126, |
| "step": 2340 |
| }, |
| { |
| "loss": 0.1519, |
| "grad_norm": 0.3038860559463501, |
| "learning_rate": 8.848476258887031e-05, |
| "epoch": 0.0724076304142729, |
| "step": 2360 |
| }, |
| { |
| "loss": 0.1529, |
| "grad_norm": 0.3296835720539093, |
| "learning_rate": 8.827931701005974e-05, |
| "epoch": 0.07302125440083453, |
| "step": 2380 |
| }, |
| { |
| "loss": 0.1491, |
| "grad_norm": 0.2646135687828064, |
| "learning_rate": 8.807229791845673e-05, |
| "epoch": 0.07363487838739616, |
| "step": 2400 |
| }, |
| { |
| "loss": 0.1587, |
| "grad_norm": 0.303659588098526, |
| "learning_rate": 8.786371382380528e-05, |
| "epoch": 0.0742485023739578, |
| "step": 2420 |
| }, |
| { |
| "loss": 0.1554, |
| "grad_norm": 0.5423962473869324, |
| "learning_rate": 8.765357330018056e-05, |
| "epoch": 0.07486212636051943, |
| "step": 2440 |
| }, |
| { |
| "loss": 0.1531, |
| "grad_norm": 0.22406215965747833, |
| "learning_rate": 8.744188498563641e-05, |
| "epoch": 0.07547575034708107, |
| "step": 2460 |
| }, |
| { |
| "loss": 0.1556, |
| "grad_norm": 0.32763445377349854, |
| "learning_rate": 8.722865758185035e-05, |
| "epoch": 0.0760893743336427, |
| "step": 2480 |
| }, |
| { |
| "loss": 0.1547, |
| "grad_norm": 0.23374123871326447, |
| "learning_rate": 8.701389985376578e-05, |
| "epoch": 0.07670299832020434, |
| "step": 2500 |
| }, |
| { |
| "eval_loss": 0.1528489738702774, |
| "eval_runtime": 2771.0889, |
| "eval_samples_per_second": 161.72, |
| "eval_steps_per_second": 2.527, |
| "epoch": 0.07670299832020434, |
| "step": 2500 |
| }, |
| { |
| "loss": 0.1534, |
| "grad_norm": 0.3874337673187256, |
| "learning_rate": 8.679762062923175e-05, |
| "epoch": 0.07731662230676598, |
| "step": 2520 |
| }, |
| { |
| "loss": 0.1567, |
| "grad_norm": 0.18201296031475067, |
| "learning_rate": 8.657982879864007e-05, |
| "epoch": 0.0779302462933276, |
| "step": 2540 |
| }, |
| { |
| "loss": 0.1485, |
| "grad_norm": 0.30263030529022217, |
| "learning_rate": 8.636053331455987e-05, |
| "epoch": 0.07854387027988924, |
| "step": 2560 |
| }, |
| { |
| "loss": 0.1497, |
| "grad_norm": 0.47069185972213745, |
| "learning_rate": 8.613974319136958e-05, |
| "epoch": 0.07915749426645087, |
| "step": 2580 |
| }, |
| { |
| "loss": 0.1583, |
| "grad_norm": 0.24262110888957977, |
| "learning_rate": 8.591746750488639e-05, |
| "epoch": 0.07977111825301252, |
| "step": 2600 |
| }, |
| { |
| "loss": 0.1562, |
| "grad_norm": 0.22001530230045319, |
| "learning_rate": 8.569371539199316e-05, |
| "epoch": 0.08038474223957415, |
| "step": 2620 |
| }, |
| { |
| "loss": 0.1628, |
| "grad_norm": 0.21581719815731049, |
| "learning_rate": 8.54684960502629e-05, |
| "epoch": 0.08099836622613578, |
| "step": 2640 |
| }, |
| { |
| "loss": 0.1597, |
| "grad_norm": 0.34623074531555176, |
| "learning_rate": 8.524181873758059e-05, |
| "epoch": 0.08161199021269741, |
| "step": 2660 |
| }, |
| { |
| "loss": 0.153, |
| "grad_norm": 0.17073972523212433, |
| "learning_rate": 8.501369277176276e-05, |
| "epoch": 0.08222561419925904, |
| "step": 2680 |
| }, |
| { |
| "loss": 0.1584, |
| "grad_norm": 0.41088077425956726, |
| "learning_rate": 8.478412753017433e-05, |
| "epoch": 0.08283923818582069, |
| "step": 2700 |
| }, |
| { |
| "loss": 0.1635, |
| "grad_norm": 0.22260364890098572, |
| "learning_rate": 8.455313244934324e-05, |
| "epoch": 0.08345286217238232, |
| "step": 2720 |
| }, |
| { |
| "loss": 0.156, |
| "grad_norm": 0.3336716592311859, |
| "learning_rate": 8.432071702457252e-05, |
| "epoch": 0.08406648615894395, |
| "step": 2740 |
| }, |
| { |
| "loss": 0.1558, |
| "grad_norm": 0.4109443426132202, |
| "learning_rate": 8.408689080954998e-05, |
| "epoch": 0.08468011014550558, |
| "step": 2760 |
| }, |
| { |
| "loss": 0.1552, |
| "grad_norm": 0.3245985805988312, |
| "learning_rate": 8.385166341595548e-05, |
| "epoch": 0.08529373413206723, |
| "step": 2780 |
| }, |
| { |
| "loss": 0.1524, |
| "grad_norm": 0.2673623859882355, |
| "learning_rate": 8.361504451306585e-05, |
| "epoch": 0.08590735811862886, |
| "step": 2800 |
| }, |
| { |
| "loss": 0.1495, |
| "grad_norm": 1.2059513330459595, |
| "learning_rate": 8.33770438273574e-05, |
| "epoch": 0.08652098210519049, |
| "step": 2820 |
| }, |
| { |
| "loss": 0.1564, |
| "grad_norm": 0.24513472616672516, |
| "learning_rate": 8.313767114210615e-05, |
| "epoch": 0.08713460609175212, |
| "step": 2840 |
| }, |
| { |
| "loss": 0.1546, |
| "grad_norm": 0.2619301378726959, |
| "learning_rate": 8.289693629698564e-05, |
| "epoch": 0.08774823007831375, |
| "step": 2860 |
| }, |
| { |
| "loss": 0.1564, |
| "grad_norm": 0.16868187487125397, |
| "learning_rate": 8.265484918766243e-05, |
| "epoch": 0.0883618540648754, |
| "step": 2880 |
| }, |
| { |
| "loss": 0.1564, |
| "grad_norm": 0.2369966059923172, |
| "learning_rate": 8.241141976538943e-05, |
| "epoch": 0.08897547805143703, |
| "step": 2900 |
| }, |
| { |
| "loss": 0.1511, |
| "grad_norm": 0.23205088078975677, |
| "learning_rate": 8.216665803659671e-05, |
| "epoch": 0.08958910203799866, |
| "step": 2920 |
| }, |
| { |
| "loss": 0.157, |
| "grad_norm": 0.19588901102542877, |
| "learning_rate": 8.192057406248028e-05, |
| "epoch": 0.0902027260245603, |
| "step": 2940 |
| }, |
| { |
| "loss": 0.1496, |
| "grad_norm": 0.3407858908176422, |
| "learning_rate": 8.167317795858851e-05, |
| "epoch": 0.09081635001112194, |
| "step": 2960 |
| }, |
| { |
| "loss": 0.151, |
| "grad_norm": 0.2402908205986023, |
| "learning_rate": 8.142447989440618e-05, |
| "epoch": 0.09142997399768357, |
| "step": 2980 |
| }, |
| { |
| "loss": 0.1567, |
| "grad_norm": 0.23841014504432678, |
| "learning_rate": 8.117449009293668e-05, |
| "epoch": 0.0920435979842452, |
| "step": 3000 |
| }, |
| { |
| "eval_loss": 0.150652676820755, |
| "eval_runtime": 2770.5122, |
| "eval_samples_per_second": 161.753, |
| "eval_steps_per_second": 2.528, |
| "epoch": 0.0920435979842452, |
| "step": 3000 |
| }, |
| { |
| "loss": 0.1551, |
| "grad_norm": 0.20382452011108398, |
| "learning_rate": 8.092321883028158e-05, |
| "epoch": 0.09265722197080684, |
| "step": 3020 |
| }, |
| { |
| "loss": 0.1511, |
| "grad_norm": 0.28016752004623413, |
| "learning_rate": 8.067067643521834e-05, |
| "epoch": 0.09327084595736847, |
| "step": 3040 |
| }, |
| { |
| "loss": 0.1571, |
| "grad_norm": 0.24572430551052094, |
| "learning_rate": 8.041687328877567e-05, |
| "epoch": 0.09388446994393011, |
| "step": 3060 |
| }, |
| { |
| "loss": 0.1475, |
| "grad_norm": 0.31108465790748596, |
| "learning_rate": 8.016181982380682e-05, |
| "epoch": 0.09449809393049174, |
| "step": 3080 |
| }, |
| { |
| "loss": 0.151, |
| "grad_norm": 0.1946084052324295, |
| "learning_rate": 7.990552652456081e-05, |
| "epoch": 0.09511171791705338, |
| "step": 3100 |
| }, |
| { |
| "loss": 0.156, |
| "grad_norm": 0.2956799864768982, |
| "learning_rate": 7.964800392625129e-05, |
| "epoch": 0.09572534190361501, |
| "step": 3120 |
| }, |
| { |
| "loss": 0.1505, |
| "grad_norm": 0.6577198505401611, |
| "learning_rate": 7.938926261462366e-05, |
| "epoch": 0.09633896589017665, |
| "step": 3140 |
| }, |
| { |
| "loss": 0.1542, |
| "grad_norm": 0.16438567638397217, |
| "learning_rate": 7.91293132255198e-05, |
| "epoch": 0.09695258987673828, |
| "step": 3160 |
| }, |
| { |
| "loss": 0.1474, |
| "grad_norm": 0.19348329305648804, |
| "learning_rate": 7.886816644444098e-05, |
| "epoch": 0.09756621386329992, |
| "step": 3180 |
| }, |
| { |
| "loss": 0.1559, |
| "grad_norm": 0.3346814811229706, |
| "learning_rate": 7.860583300610849e-05, |
| "epoch": 0.09817983784986155, |
| "step": 3200 |
| }, |
| { |
| "loss": 0.1471, |
| "grad_norm": 0.1847156286239624, |
| "learning_rate": 7.83423236940225e-05, |
| "epoch": 0.09879346183642318, |
| "step": 3220 |
| }, |
| { |
| "loss": 0.1443, |
| "grad_norm": 0.5336312055587769, |
| "learning_rate": 7.807764934001874e-05, |
| "epoch": 0.09940708582298483, |
| "step": 3240 |
| }, |
| { |
| "loss": 0.1499, |
| "grad_norm": 0.23365505039691925, |
| "learning_rate": 7.781182082382325e-05, |
| "epoch": 0.10002070980954646, |
| "step": 3260 |
| }, |
| { |
| "loss": 0.1552, |
| "grad_norm": 0.20067407190799713, |
| "learning_rate": 7.754484907260513e-05, |
| "epoch": 0.10063433379610809, |
| "step": 3280 |
| }, |
| { |
| "loss": 0.1485, |
| "grad_norm": 0.18101651966571808, |
| "learning_rate": 7.727674506052743e-05, |
| "epoch": 0.10124795778266972, |
| "step": 3300 |
| }, |
| { |
| "loss": 0.1482, |
| "grad_norm": 0.20169106125831604, |
| "learning_rate": 7.700751980829602e-05, |
| "epoch": 0.10186158176923137, |
| "step": 3320 |
| }, |
| { |
| "loss": 0.1426, |
| "grad_norm": 0.15832331776618958, |
| "learning_rate": 7.673718438270648e-05, |
| "epoch": 0.102475205755793, |
| "step": 3340 |
| }, |
| { |
| "loss": 0.1467, |
| "grad_norm": 0.2241671085357666, |
| "learning_rate": 7.646574989618938e-05, |
| "epoch": 0.10308882974235463, |
| "step": 3360 |
| }, |
| { |
| "loss": 0.1436, |
| "grad_norm": 0.14387348294258118, |
| "learning_rate": 7.619322750635327e-05, |
| "epoch": 0.10370245372891626, |
| "step": 3380 |
| }, |
| { |
| "loss": 0.1472, |
| "grad_norm": 0.3439478576183319, |
| "learning_rate": 7.591962841552627e-05, |
| "epoch": 0.10431607771547789, |
| "step": 3400 |
| }, |
| { |
| "loss": 0.1451, |
| "grad_norm": 0.2688335180282593, |
| "learning_rate": 7.564496387029532e-05, |
| "epoch": 0.10492970170203954, |
| "step": 3420 |
| }, |
| { |
| "loss": 0.1529, |
| "grad_norm": 0.5446487069129944, |
| "learning_rate": 7.536924516104411e-05, |
| "epoch": 0.10554332568860117, |
| "step": 3440 |
| }, |
| { |
| "loss": 0.1548, |
| "grad_norm": 0.35945549607276917, |
| "learning_rate": 7.509248362148889e-05, |
| "epoch": 0.1061569496751628, |
| "step": 3460 |
| }, |
| { |
| "loss": 0.1439, |
| "grad_norm": 0.4427342116832733, |
| "learning_rate": 7.481469062821252e-05, |
| "epoch": 0.10677057366172443, |
| "step": 3480 |
| }, |
| { |
| "loss": 0.1514, |
| "grad_norm": 0.2129109799861908, |
| "learning_rate": 7.45358776001969e-05, |
| "epoch": 0.10738419764828608, |
| "step": 3500 |
| }, |
| { |
| "eval_loss": 0.1471174657344818, |
| "eval_runtime": 2771.0363, |
| "eval_samples_per_second": 161.723, |
| "eval_steps_per_second": 2.527, |
| "epoch": 0.10738419764828608, |
| "step": 3500 |
| }, |
| { |
| "loss": 0.1463, |
| "grad_norm": 0.32425713539123535, |
| "learning_rate": 7.425605599835361e-05, |
| "epoch": 0.10799782163484771, |
| "step": 3520 |
| }, |
| { |
| "loss": 0.1424, |
| "grad_norm": 0.39084091782569885, |
| "learning_rate": 7.39752373250527e-05, |
| "epoch": 0.10861144562140934, |
| "step": 3540 |
| }, |
| { |
| "loss": 0.1437, |
| "grad_norm": 0.5381523370742798, |
| "learning_rate": 7.369343312364993e-05, |
| "epoch": 0.10922506960797097, |
| "step": 3560 |
| }, |
| { |
| "loss": 0.1447, |
| "grad_norm": 0.12256123870611191, |
| "learning_rate": 7.34106549780123e-05, |
| "epoch": 0.1098386935945326, |
| "step": 3580 |
| }, |
| { |
| "loss": 0.1415, |
| "grad_norm": 0.16140857338905334, |
| "learning_rate": 7.312691451204178e-05, |
| "epoch": 0.11045231758109425, |
| "step": 3600 |
| }, |
| { |
| "loss": 0.15, |
| "grad_norm": 0.1626661866903305, |
| "learning_rate": 7.284222338919758e-05, |
| "epoch": 0.11106594156765588, |
| "step": 3620 |
| }, |
| { |
| "loss": 0.1508, |
| "grad_norm": 0.17792241275310516, |
| "learning_rate": 7.255659331201673e-05, |
| "epoch": 0.11167956555421751, |
| "step": 3640 |
| }, |
| { |
| "loss": 0.1468, |
| "grad_norm": 0.24779079854488373, |
| "learning_rate": 7.227003602163295e-05, |
| "epoch": 0.11229318954077915, |
| "step": 3660 |
| }, |
| { |
| "loss": 0.1495, |
| "grad_norm": 0.1464058756828308, |
| "learning_rate": 7.198256329729412e-05, |
| "epoch": 0.11290681352734078, |
| "step": 3680 |
| }, |
| { |
| "loss": 0.1465, |
| "grad_norm": 0.2891758680343628, |
| "learning_rate": 7.169418695587791e-05, |
| "epoch": 0.11352043751390242, |
| "step": 3700 |
| }, |
| { |
| "loss": 0.1433, |
| "grad_norm": 0.12943851947784424, |
| "learning_rate": 7.14049188514063e-05, |
| "epoch": 0.11413406150046405, |
| "step": 3720 |
| }, |
| { |
| "loss": 0.1425, |
| "grad_norm": 0.24431823194026947, |
| "learning_rate": 7.1114770874558e-05, |
| "epoch": 0.11474768548702569, |
| "step": 3740 |
| }, |
| { |
| "loss": 0.1478, |
| "grad_norm": 0.19931627810001373, |
| "learning_rate": 7.082375495217995e-05, |
| "epoch": 0.11536130947358732, |
| "step": 3760 |
| }, |
| { |
| "loss": 0.1447, |
| "grad_norm": 0.2738446295261383, |
| "learning_rate": 7.05318830467969e-05, |
| "epoch": 0.11597493346014896, |
| "step": 3780 |
| }, |
| { |
| "loss": 0.147, |
| "grad_norm": 0.18363946676254272, |
| "learning_rate": 7.023916715611969e-05, |
| "epoch": 0.1165885574467106, |
| "step": 3800 |
| }, |
| { |
| "loss": 0.1484, |
| "grad_norm": 0.21691597998142242, |
| "learning_rate": 6.99456193125521e-05, |
| "epoch": 0.11720218143327223, |
| "step": 3820 |
| }, |
| { |
| "loss": 0.144, |
| "grad_norm": 0.29352739453315735, |
| "learning_rate": 6.965125158269619e-05, |
| "epoch": 0.11781580541983386, |
| "step": 3840 |
| }, |
| { |
| "loss": 0.1482, |
| "grad_norm": 0.15962694585323334, |
| "learning_rate": 6.935607606685642e-05, |
| "epoch": 0.11842942940639549, |
| "step": 3860 |
| }, |
| { |
| "loss": 0.1476, |
| "grad_norm": 0.2411005049943924, |
| "learning_rate": 6.906010489854209e-05, |
| "epoch": 0.11904305339295713, |
| "step": 3880 |
| }, |
| { |
| "loss": 0.1461, |
| "grad_norm": 0.3792065382003784, |
| "learning_rate": 6.876335024396872e-05, |
| "epoch": 0.11965667737951877, |
| "step": 3900 |
| }, |
| { |
| "loss": 0.1456, |
| "grad_norm": 0.21769222617149353, |
| "learning_rate": 6.846582430155783e-05, |
| "epoch": 0.1202703013660804, |
| "step": 3920 |
| }, |
| { |
| "loss": 0.1407, |
| "grad_norm": 0.12780731916427612, |
| "learning_rate": 6.816753930143558e-05, |
| "epoch": 0.12088392535264203, |
| "step": 3940 |
| }, |
| { |
| "loss": 0.143, |
| "grad_norm": 0.18131625652313232, |
| "learning_rate": 6.786850750493006e-05, |
| "epoch": 0.12149754933920368, |
| "step": 3960 |
| }, |
| { |
| "loss": 0.1451, |
| "grad_norm": 0.15587596595287323, |
| "learning_rate": 6.756874120406714e-05, |
| "epoch": 0.12211117332576531, |
| "step": 3980 |
| }, |
| { |
| "loss": 0.1441, |
| "grad_norm": 0.2455044686794281, |
| "learning_rate": 6.726825272106538e-05, |
| "epoch": 0.12272479731232694, |
| "step": 4000 |
| }, |
| { |
| "eval_loss": 0.1430002897977829, |
| "eval_runtime": 2770.4767, |
| "eval_samples_per_second": 161.756, |
| "eval_steps_per_second": 2.528, |
| "epoch": 0.12272479731232694, |
| "step": 4000 |
| }, |
| { |
| "loss": 0.1459, |
| "grad_norm": 0.24546217918395996, |
| "learning_rate": 6.696705440782938e-05, |
| "epoch": 0.12333842129888857, |
| "step": 4020 |
| }, |
| { |
| "loss": 0.142, |
| "grad_norm": 0.11568096280097961, |
| "learning_rate": 6.666515864544209e-05, |
| "epoch": 0.1239520452854502, |
| "step": 4040 |
| }, |
| { |
| "loss": 0.1564, |
| "grad_norm": 0.7489081621170044, |
| "learning_rate": 6.636257784365584e-05, |
| "epoch": 0.12456566927201185, |
| "step": 4060 |
| }, |
| { |
| "loss": 0.1428, |
| "grad_norm": 0.24135908484458923, |
| "learning_rate": 6.605932444038229e-05, |
| "epoch": 0.12517929325857347, |
| "step": 4080 |
| }, |
| { |
| "loss": 0.1484, |
| "grad_norm": 0.22589971125125885, |
| "learning_rate": 6.575541090118105e-05, |
| "epoch": 0.1257929172451351, |
| "step": 4100 |
| }, |
| { |
| "loss": 0.1473, |
| "grad_norm": 0.17859506607055664, |
| "learning_rate": 6.545084971874738e-05, |
| "epoch": 0.12640654123169676, |
| "step": 4120 |
| }, |
| { |
| "loss": 0.1465, |
| "grad_norm": 0.17700891196727753, |
| "learning_rate": 6.514565341239861e-05, |
| "epoch": 0.12702016521825837, |
| "step": 4140 |
| }, |
| { |
| "loss": 0.1434, |
| "grad_norm": 0.21767599880695343, |
| "learning_rate": 6.483983452755953e-05, |
| "epoch": 0.12763378920482002, |
| "step": 4160 |
| }, |
| { |
| "loss": 0.1423, |
| "grad_norm": 0.1528337001800537, |
| "learning_rate": 6.453340563524669e-05, |
| "epoch": 0.12824741319138164, |
| "step": 4180 |
| }, |
| { |
| "loss": 0.1407, |
| "grad_norm": 0.2475786954164505, |
| "learning_rate": 6.422637933155162e-05, |
| "epoch": 0.12886103717794328, |
| "step": 4200 |
| }, |
| { |
| "loss": 0.142, |
| "grad_norm": 0.2731213867664337, |
| "learning_rate": 6.391876823712317e-05, |
| "epoch": 0.12947466116450493, |
| "step": 4220 |
| }, |
| { |
| "loss": 0.1439, |
| "grad_norm": 0.4397956132888794, |
| "learning_rate": 6.361058499664856e-05, |
| "epoch": 0.13008828515106655, |
| "step": 4240 |
| }, |
| { |
| "loss": 0.1428, |
| "grad_norm": 0.20277808606624603, |
| "learning_rate": 6.330184227833376e-05, |
| "epoch": 0.1307019091376282, |
| "step": 4260 |
| }, |
| { |
| "loss": 0.1394, |
| "grad_norm": 0.1790492981672287, |
| "learning_rate": 6.299255277338265e-05, |
| "epoch": 0.13131553312418984, |
| "step": 4280 |
| }, |
| { |
| "loss": 0.1386, |
| "grad_norm": 0.181828111410141, |
| "learning_rate": 6.268272919547537e-05, |
| "epoch": 0.13192915711075145, |
| "step": 4300 |
| }, |
| { |
| "loss": 0.1459, |
| "grad_norm": 0.1925831288099289, |
| "learning_rate": 6.237238428024572e-05, |
| "epoch": 0.1325427810973131, |
| "step": 4320 |
| }, |
| { |
| "loss": 0.1424, |
| "grad_norm": 0.6288266181945801, |
| "learning_rate": 6.206153078475763e-05, |
| "epoch": 0.13315640508387472, |
| "step": 4340 |
| }, |
| { |
| "loss": 0.1454, |
| "grad_norm": 0.15917836129665375, |
| "learning_rate": 6.175018148698077e-05, |
| "epoch": 0.13377002907043636, |
| "step": 4360 |
| }, |
| { |
| "loss": 0.1437, |
| "grad_norm": 0.21755409240722656, |
| "learning_rate": 6.143834918526527e-05, |
| "epoch": 0.134383653056998, |
| "step": 4380 |
| }, |
| { |
| "loss": 0.1447, |
| "grad_norm": 0.14852923154830933, |
| "learning_rate": 6.112604669781572e-05, |
| "epoch": 0.13499727704355963, |
| "step": 4400 |
| }, |
| { |
| "loss": 0.1428, |
| "grad_norm": 0.20071613788604736, |
| "learning_rate": 6.081328686216418e-05, |
| "epoch": 0.13561090103012127, |
| "step": 4420 |
| }, |
| { |
| "loss": 0.1418, |
| "grad_norm": 0.13478811085224152, |
| "learning_rate": 6.0500082534642464e-05, |
| "epoch": 0.1362245250166829, |
| "step": 4440 |
| }, |
| { |
| "loss": 0.1513, |
| "grad_norm": 0.11774014681577682, |
| "learning_rate": 6.0186446589853784e-05, |
| "epoch": 0.13683814900324454, |
| "step": 4460 |
| }, |
| { |
| "loss": 0.1455, |
| "grad_norm": 0.3604123890399933, |
| "learning_rate": 5.987239192014336e-05, |
| "epoch": 0.13745177298980618, |
| "step": 4480 |
| }, |
| { |
| "loss": 0.1311, |
| "grad_norm": 0.19379453361034393, |
| "learning_rate": 5.955793143506863e-05, |
| "epoch": 0.1380653969763678, |
| "step": 4500 |
| }, |
| { |
| "eval_loss": 0.14158321917057037, |
| "eval_runtime": 2770.8316, |
| "eval_samples_per_second": 161.735, |
| "eval_steps_per_second": 2.527, |
| "epoch": 0.1380653969763678, |
| "step": 4500 |
| }, |
| { |
| "loss": 0.145, |
| "grad_norm": 0.1940959244966507, |
| "learning_rate": 5.924307806086844e-05, |
| "epoch": 0.13867902096292944, |
| "step": 4520 |
| }, |
| { |
| "loss": 0.1376, |
| "grad_norm": 0.17542797327041626, |
| "learning_rate": 5.8927844739931834e-05, |
| "epoch": 0.13929264494949106, |
| "step": 4540 |
| }, |
| { |
| "loss": 0.1369, |
| "grad_norm": 0.13041825592517853, |
| "learning_rate": 5.861224443026595e-05, |
| "epoch": 0.1399062689360527, |
| "step": 4560 |
| }, |
| { |
| "loss": 0.1406, |
| "grad_norm": 0.19473542273044586, |
| "learning_rate": 5.82962901049634e-05, |
| "epoch": 0.14051989292261435, |
| "step": 4580 |
| }, |
| { |
| "loss": 0.1442, |
| "grad_norm": 0.22222813963890076, |
| "learning_rate": 5.7979994751668964e-05, |
| "epoch": 0.14113351690917597, |
| "step": 4600 |
| }, |
| { |
| "loss": 0.14, |
| "grad_norm": 0.21635234355926514, |
| "learning_rate": 5.766337137204579e-05, |
| "epoch": 0.14174714089573762, |
| "step": 4620 |
| }, |
| { |
| "loss": 0.1359, |
| "grad_norm": 0.20376254618167877, |
| "learning_rate": 5.7346432981240904e-05, |
| "epoch": 0.14236076488229926, |
| "step": 4640 |
| }, |
| { |
| "loss": 0.146, |
| "grad_norm": 0.18470820784568787, |
| "learning_rate": 5.7029192607350146e-05, |
| "epoch": 0.14297438886886088, |
| "step": 4660 |
| }, |
| { |
| "loss": 0.1405, |
| "grad_norm": 0.14597751200199127, |
| "learning_rate": 5.6711663290882776e-05, |
| "epoch": 0.14358801285542253, |
| "step": 4680 |
| }, |
| { |
| "loss": 0.1347, |
| "grad_norm": 0.10637301951646805, |
| "learning_rate": 5.6393858084225305e-05, |
| "epoch": 0.14420163684198414, |
| "step": 4700 |
| }, |
| { |
| "loss": 0.1504, |
| "grad_norm": 0.1544061303138733, |
| "learning_rate": 5.6075790051105023e-05, |
| "epoch": 0.1448152608285458, |
| "step": 4720 |
| }, |
| { |
| "loss": 0.1426, |
| "grad_norm": 0.15634353458881378, |
| "learning_rate": 5.575747226605298e-05, |
| "epoch": 0.14542888481510743, |
| "step": 4740 |
| }, |
| { |
| "loss": 0.1389, |
| "grad_norm": 0.5837845802307129, |
| "learning_rate": 5.5438917813866554e-05, |
| "epoch": 0.14604250880166905, |
| "step": 4760 |
| }, |
| { |
| "loss": 0.1454, |
| "grad_norm": 0.43749943375587463, |
| "learning_rate": 5.512013978907157e-05, |
| "epoch": 0.1466561327882307, |
| "step": 4780 |
| }, |
| { |
| "loss": 0.1514, |
| "grad_norm": 0.18316301703453064, |
| "learning_rate": 5.480115129538409e-05, |
| "epoch": 0.14726975677479232, |
| "step": 4800 |
| }, |
| { |
| "loss": 0.1422, |
| "grad_norm": 0.17283789813518524, |
| "learning_rate": 5.448196544517168e-05, |
| "epoch": 0.14788338076135396, |
| "step": 4820 |
| }, |
| { |
| "loss": 0.1498, |
| "grad_norm": 0.2187805473804474, |
| "learning_rate": 5.416259535891447e-05, |
| "epoch": 0.1484970047479156, |
| "step": 4840 |
| }, |
| { |
| "loss": 0.1311, |
| "grad_norm": 0.160814106464386, |
| "learning_rate": 5.384305416466584e-05, |
| "epoch": 0.14911062873447722, |
| "step": 4860 |
| }, |
| { |
| "loss": 0.138, |
| "grad_norm": 0.5161686539649963, |
| "learning_rate": 5.35233549975127e-05, |
| "epoch": 0.14972425272103887, |
| "step": 4880 |
| }, |
| { |
| "loss": 0.1401, |
| "grad_norm": 0.11132976412773132, |
| "learning_rate": 5.320351099903565e-05, |
| "epoch": 0.1503378767076005, |
| "step": 4900 |
| }, |
| { |
| "loss": 0.1424, |
| "grad_norm": 0.13249003887176514, |
| "learning_rate": 5.288353531676873e-05, |
| "epoch": 0.15095150069416213, |
| "step": 4920 |
| }, |
| { |
| "loss": 0.142, |
| "grad_norm": 0.13185490667819977, |
| "learning_rate": 5.256344110365896e-05, |
| "epoch": 0.15156512468072378, |
| "step": 4940 |
| }, |
| { |
| "loss": 0.1423, |
| "grad_norm": 0.3014187216758728, |
| "learning_rate": 5.2243241517525754e-05, |
| "epoch": 0.1521787486672854, |
| "step": 4960 |
| }, |
| { |
| "loss": 0.1387, |
| "grad_norm": 0.09553083777427673, |
| "learning_rate": 5.192294972051992e-05, |
| "epoch": 0.15279237265384704, |
| "step": 4980 |
| }, |
| { |
| "loss": 0.1413, |
| "grad_norm": 0.37995925545692444, |
| "learning_rate": 5.1602578878582776e-05, |
| "epoch": 0.1534059966404087, |
| "step": 5000 |
| }, |
| { |
| "eval_loss": 0.1410340666770935, |
| "eval_runtime": 2772.7302, |
| "eval_samples_per_second": 161.624, |
| "eval_steps_per_second": 2.526, |
| "epoch": 0.1534059966404087, |
| "step": 5000 |
| }, |
| { |
| "loss": 0.1426, |
| "grad_norm": 0.18281251192092896, |
| "learning_rate": 5.128214216090478e-05, |
| "epoch": 0.1540196206269703, |
| "step": 5020 |
| }, |
| { |
| "loss": 0.144, |
| "grad_norm": 0.37768101692199707, |
| "learning_rate": 5.0961652739384356e-05, |
| "epoch": 0.15463324461353195, |
| "step": 5040 |
| }, |
| { |
| "loss": 0.1385, |
| "grad_norm": 0.25653406977653503, |
| "learning_rate": 5.064112378808637e-05, |
| "epoch": 0.15524686860009357, |
| "step": 5060 |
| }, |
| { |
| "loss": 0.1438, |
| "grad_norm": 0.16125090420246124, |
| "learning_rate": 5.0320568482700556e-05, |
| "epoch": 0.1558604925866552, |
| "step": 5080 |
| }, |
| { |
| "loss": 0.1365, |
| "grad_norm": 0.13656456768512726, |
| "learning_rate": 5e-05, |
| "epoch": 0.15647411657321686, |
| "step": 5100 |
| }, |
| { |
| "loss": 0.1407, |
| "grad_norm": 0.19003720581531525, |
| "learning_rate": 4.967943151729945e-05, |
| "epoch": 0.15708774055977848, |
| "step": 5120 |
| }, |
| { |
| "loss": 0.1443, |
| "grad_norm": 0.12324690073728561, |
| "learning_rate": 4.935887621191364e-05, |
| "epoch": 0.15770136454634012, |
| "step": 5140 |
| }, |
| { |
| "loss": 0.1407, |
| "grad_norm": 0.4486618936061859, |
| "learning_rate": 4.903834726061565e-05, |
| "epoch": 0.15831498853290174, |
| "step": 5160 |
| }, |
| { |
| "loss": 0.1459, |
| "grad_norm": 0.2653239965438843, |
| "learning_rate": 4.871785783909523e-05, |
| "epoch": 0.15892861251946339, |
| "step": 5180 |
| }, |
| { |
| "loss": 0.1406, |
| "grad_norm": 0.1883934885263443, |
| "learning_rate": 4.839742112141724e-05, |
| "epoch": 0.15954223650602503, |
| "step": 5200 |
| }, |
| { |
| "loss": 0.1477, |
| "grad_norm": 0.23059062659740448, |
| "learning_rate": 4.807705027948008e-05, |
| "epoch": 0.16015586049258665, |
| "step": 5220 |
| }, |
| { |
| "loss": 0.138, |
| "grad_norm": 0.10817737877368927, |
| "learning_rate": 4.775675848247427e-05, |
| "epoch": 0.1607694844791483, |
| "step": 5240 |
| }, |
| { |
| "loss": 0.1388, |
| "grad_norm": 0.23007145524024963, |
| "learning_rate": 4.743655889634105e-05, |
| "epoch": 0.1613831084657099, |
| "step": 5260 |
| }, |
| { |
| "loss": 0.14, |
| "grad_norm": 0.12711915373802185, |
| "learning_rate": 4.711646468323129e-05, |
| "epoch": 0.16199673245227156, |
| "step": 5280 |
| }, |
| { |
| "loss": 0.1368, |
| "grad_norm": 0.11912371963262558, |
| "learning_rate": 4.679648900096436e-05, |
| "epoch": 0.1626103564388332, |
| "step": 5300 |
| }, |
| { |
| "loss": 0.1374, |
| "grad_norm": 0.14275406301021576, |
| "learning_rate": 4.64766450024873e-05, |
| "epoch": 0.16322398042539482, |
| "step": 5320 |
| }, |
| { |
| "loss": 0.1347, |
| "grad_norm": 0.3042926490306854, |
| "learning_rate": 4.6156945835334184e-05, |
| "epoch": 0.16383760441195647, |
| "step": 5340 |
| }, |
| { |
| "loss": 0.139, |
| "grad_norm": 0.16544413566589355, |
| "learning_rate": 4.583740464108554e-05, |
| "epoch": 0.16445122839851808, |
| "step": 5360 |
| }, |
| { |
| "loss": 0.1424, |
| "grad_norm": 0.4930761754512787, |
| "learning_rate": 4.551803455482833e-05, |
| "epoch": 0.16506485238507973, |
| "step": 5380 |
| }, |
| { |
| "loss": 0.1412, |
| "grad_norm": 0.4234517216682434, |
| "learning_rate": 4.5198848704615914e-05, |
| "epoch": 0.16567847637164138, |
| "step": 5400 |
| }, |
| { |
| "loss": 0.1415, |
| "grad_norm": 0.15803220868110657, |
| "learning_rate": 4.487986021092844e-05, |
| "epoch": 0.166292100358203, |
| "step": 5420 |
| }, |
| { |
| "loss": 0.1413, |
| "grad_norm": 0.18349704146385193, |
| "learning_rate": 4.4561082186133464e-05, |
| "epoch": 0.16690572434476464, |
| "step": 5440 |
| }, |
| { |
| "loss": 0.141, |
| "grad_norm": 0.14680899679660797, |
| "learning_rate": 4.424252773394704e-05, |
| "epoch": 0.16751934833132628, |
| "step": 5460 |
| }, |
| { |
| "loss": 0.1462, |
| "grad_norm": 0.18841388821601868, |
| "learning_rate": 4.392420994889498e-05, |
| "epoch": 0.1681329723178879, |
| "step": 5480 |
| }, |
| { |
| "loss": 0.1403, |
| "grad_norm": 0.3068675398826599, |
| "learning_rate": 4.3606141915774693e-05, |
| "epoch": 0.16874659630444955, |
| "step": 5500 |
| }, |
| { |
| "eval_loss": 0.1398838609457016, |
| "eval_runtime": 2772.4795, |
| "eval_samples_per_second": 161.639, |
| "eval_steps_per_second": 2.526, |
| "epoch": 0.16874659630444955, |
| "step": 5500 |
| }, |
| { |
| "loss": 0.1369, |
| "grad_norm": 0.13292258977890015, |
| "learning_rate": 4.328833670911724e-05, |
| "epoch": 0.16936022029101117, |
| "step": 5520 |
| }, |
| { |
| "loss": 0.147, |
| "grad_norm": 0.11828584223985672, |
| "learning_rate": 4.297080739264987e-05, |
| "epoch": 0.1699738442775728, |
| "step": 5540 |
| }, |
| { |
| "loss": 0.1405, |
| "grad_norm": 0.15388241410255432, |
| "learning_rate": 4.265356701875911e-05, |
| "epoch": 0.17058746826413446, |
| "step": 5560 |
| }, |
| { |
| "loss": 0.137, |
| "grad_norm": 0.20380249619483948, |
| "learning_rate": 4.23366286279542e-05, |
| "epoch": 0.17120109225069607, |
| "step": 5580 |
| }, |
| { |
| "loss": 0.1299, |
| "grad_norm": 0.18336381018161774, |
| "learning_rate": 4.2020005248331054e-05, |
| "epoch": 0.17181471623725772, |
| "step": 5600 |
| }, |
| { |
| "loss": 0.138, |
| "grad_norm": 0.16339316964149475, |
| "learning_rate": 4.1703709895036625e-05, |
| "epoch": 0.17242834022381934, |
| "step": 5620 |
| }, |
| { |
| "loss": 0.1383, |
| "grad_norm": 0.14355255663394928, |
| "learning_rate": 4.138775556973406e-05, |
| "epoch": 0.17304196421038098, |
| "step": 5640 |
| }, |
| { |
| "loss": 0.141, |
| "grad_norm": 0.19084544479846954, |
| "learning_rate": 4.107215526006817e-05, |
| "epoch": 0.17365558819694263, |
| "step": 5660 |
| }, |
| { |
| "loss": 0.1304, |
| "grad_norm": 0.11718132346868515, |
| "learning_rate": 4.0756921939131565e-05, |
| "epoch": 0.17426921218350425, |
| "step": 5680 |
| }, |
| { |
| "loss": 0.1371, |
| "grad_norm": 0.13921724259853363, |
| "learning_rate": 4.04420685649314e-05, |
| "epoch": 0.1748828361700659, |
| "step": 5700 |
| }, |
| { |
| "loss": 0.1376, |
| "grad_norm": 0.14769333600997925, |
| "learning_rate": 4.012760807985665e-05, |
| "epoch": 0.1754964601566275, |
| "step": 5720 |
| }, |
| { |
| "loss": 0.1423, |
| "grad_norm": 0.11135128140449524, |
| "learning_rate": 3.981355341014623e-05, |
| "epoch": 0.17611008414318916, |
| "step": 5740 |
| }, |
| { |
| "loss": 0.1363, |
| "grad_norm": 0.15607145428657532, |
| "learning_rate": 3.9499917465357534e-05, |
| "epoch": 0.1767237081297508, |
| "step": 5760 |
| }, |
| { |
| "loss": 0.1367, |
| "grad_norm": 0.2499862164258957, |
| "learning_rate": 3.9186713137835826e-05, |
| "epoch": 0.17733733211631242, |
| "step": 5780 |
| }, |
| { |
| "loss": 0.1372, |
| "grad_norm": 0.31999796628952026, |
| "learning_rate": 3.887395330218429e-05, |
| "epoch": 0.17795095610287406, |
| "step": 5800 |
| }, |
| { |
| "loss": 0.1412, |
| "grad_norm": 0.2039441466331482, |
| "learning_rate": 3.856165081473474e-05, |
| "epoch": 0.1785645800894357, |
| "step": 5820 |
| }, |
| { |
| "loss": 0.1349, |
| "grad_norm": 0.1782849133014679, |
| "learning_rate": 3.8249818513019244e-05, |
| "epoch": 0.17917820407599733, |
| "step": 5840 |
| }, |
| { |
| "loss": 0.1337, |
| "grad_norm": 0.14570361375808716, |
| "learning_rate": 3.793846921524237e-05, |
| "epoch": 0.17979182806255897, |
| "step": 5860 |
| }, |
| { |
| "loss": 0.1416, |
| "grad_norm": 0.199429452419281, |
| "learning_rate": 3.762761571975429e-05, |
| "epoch": 0.1804054520491206, |
| "step": 5880 |
| }, |
| { |
| "loss": 0.1459, |
| "grad_norm": 0.18289752304553986, |
| "learning_rate": 3.731727080452464e-05, |
| "epoch": 0.18101907603568224, |
| "step": 5900 |
| }, |
| { |
| "loss": 0.1359, |
| "grad_norm": 0.13586929440498352, |
| "learning_rate": 3.7007447226617366e-05, |
| "epoch": 0.18163270002224388, |
| "step": 5920 |
| }, |
| { |
| "loss": 0.1394, |
| "grad_norm": 0.19824199378490448, |
| "learning_rate": 3.6698157721666246e-05, |
| "epoch": 0.1822463240088055, |
| "step": 5940 |
| }, |
| { |
| "loss": 0.1383, |
| "grad_norm": 0.2370585948228836, |
| "learning_rate": 3.638941500335145e-05, |
| "epoch": 0.18285994799536714, |
| "step": 5960 |
| }, |
| { |
| "loss": 0.138, |
| "grad_norm": 0.14067555963993073, |
| "learning_rate": 3.608123176287685e-05, |
| "epoch": 0.18347357198192876, |
| "step": 5980 |
| }, |
| { |
| "loss": 0.1415, |
| "grad_norm": 0.21730934083461761, |
| "learning_rate": 3.5773620668448384e-05, |
| "epoch": 0.1840871959684904, |
| "step": 6000 |
| }, |
| { |
| "eval_loss": 0.1372094452381134, |
| "eval_runtime": 2771.4148, |
| "eval_samples_per_second": 161.701, |
| "eval_steps_per_second": 2.527, |
| "epoch": 0.1840871959684904, |
| "step": 6000 |
| }, |
| { |
| "loss": 0.1384, |
| "grad_norm": 0.12257638573646545, |
| "learning_rate": 3.5466594364753326e-05, |
| "epoch": 0.18470081995505205, |
| "step": 6020 |
| }, |
| { |
| "loss": 0.1381, |
| "grad_norm": 0.13667023181915283, |
| "learning_rate": 3.5160165472440473e-05, |
| "epoch": 0.18531444394161367, |
| "step": 6040 |
| }, |
| { |
| "loss": 0.1373, |
| "grad_norm": 0.20999397337436676, |
| "learning_rate": 3.48543465876014e-05, |
| "epoch": 0.18592806792817532, |
| "step": 6060 |
| }, |
| { |
| "loss": 0.1388, |
| "grad_norm": 0.1176946684718132, |
| "learning_rate": 3.4549150281252636e-05, |
| "epoch": 0.18654169191473693, |
| "step": 6080 |
| }, |
| { |
| "loss": 0.1406, |
| "grad_norm": 0.12363361567258835, |
| "learning_rate": 3.424458909881897e-05, |
| "epoch": 0.18715531590129858, |
| "step": 6100 |
| }, |
| { |
| "loss": 0.1346, |
| "grad_norm": 0.1650696098804474, |
| "learning_rate": 3.3940675559617724e-05, |
| "epoch": 0.18776893988786023, |
| "step": 6120 |
| }, |
| { |
| "loss": 0.1411, |
| "grad_norm": 0.13882821798324585, |
| "learning_rate": 3.363742215634415e-05, |
| "epoch": 0.18838256387442184, |
| "step": 6140 |
| }, |
| { |
| "loss": 0.135, |
| "grad_norm": 0.18730175495147705, |
| "learning_rate": 3.333484135455792e-05, |
| "epoch": 0.1889961878609835, |
| "step": 6160 |
| }, |
| { |
| "loss": 0.1375, |
| "grad_norm": 0.15268158912658691, |
| "learning_rate": 3.303294559217063e-05, |
| "epoch": 0.18960981184754513, |
| "step": 6180 |
| }, |
| { |
| "loss": 0.1309, |
| "grad_norm": 0.6368291974067688, |
| "learning_rate": 3.273174727893463e-05, |
| "epoch": 0.19022343583410675, |
| "step": 6200 |
| }, |
| { |
| "loss": 0.1431, |
| "grad_norm": 0.19259753823280334, |
| "learning_rate": 3.243125879593286e-05, |
| "epoch": 0.1908370598206684, |
| "step": 6220 |
| }, |
| { |
| "loss": 0.134, |
| "grad_norm": 0.20992296934127808, |
| "learning_rate": 3.213149249506997e-05, |
| "epoch": 0.19145068380723002, |
| "step": 6240 |
| }, |
| { |
| "loss": 0.134, |
| "grad_norm": 0.1862725019454956, |
| "learning_rate": 3.183246069856443e-05, |
| "epoch": 0.19206430779379166, |
| "step": 6260 |
| }, |
| { |
| "loss": 0.1339, |
| "grad_norm": 0.10517208278179169, |
| "learning_rate": 3.153417569844219e-05, |
| "epoch": 0.1926779317803533, |
| "step": 6280 |
| }, |
| { |
| "loss": 0.1377, |
| "grad_norm": 0.16911420226097107, |
| "learning_rate": 3.12366497560313e-05, |
| "epoch": 0.19329155576691492, |
| "step": 6300 |
| }, |
| { |
| "loss": 0.1402, |
| "grad_norm": 0.1685948520898819, |
| "learning_rate": 3.0939895101457916e-05, |
| "epoch": 0.19390517975347657, |
| "step": 6320 |
| }, |
| { |
| "loss": 0.138, |
| "grad_norm": 0.2124929577112198, |
| "learning_rate": 3.06439239331436e-05, |
| "epoch": 0.1945188037400382, |
| "step": 6340 |
| }, |
| { |
| "loss": 0.1332, |
| "grad_norm": 0.24099479615688324, |
| "learning_rate": 3.0348748417303823e-05, |
| "epoch": 0.19513242772659983, |
| "step": 6360 |
| }, |
| { |
| "loss": 0.1347, |
| "grad_norm": 0.1686478555202484, |
| "learning_rate": 3.005438068744792e-05, |
| "epoch": 0.19574605171316148, |
| "step": 6380 |
| }, |
| { |
| "loss": 0.1353, |
| "grad_norm": 0.11981747299432755, |
| "learning_rate": 2.976083284388031e-05, |
| "epoch": 0.1963596756997231, |
| "step": 6400 |
| }, |
| { |
| "loss": 0.1425, |
| "grad_norm": 0.10391107201576233, |
| "learning_rate": 2.9468116953203107e-05, |
| "epoch": 0.19697329968628474, |
| "step": 6420 |
| }, |
| { |
| "loss": 0.1354, |
| "grad_norm": 0.1283193975687027, |
| "learning_rate": 2.917624504782006e-05, |
| "epoch": 0.19758692367284636, |
| "step": 6440 |
| }, |
| { |
| "loss": 0.1372, |
| "grad_norm": 0.1691513955593109, |
| "learning_rate": 2.888522912544202e-05, |
| "epoch": 0.198200547659408, |
| "step": 6460 |
| }, |
| { |
| "loss": 0.1278, |
| "grad_norm": 0.14271607995033264, |
| "learning_rate": 2.8595081148593738e-05, |
| "epoch": 0.19881417164596965, |
| "step": 6480 |
| }, |
| { |
| "loss": 0.1315, |
| "grad_norm": 0.1568264663219452, |
| "learning_rate": 2.8305813044122097e-05, |
| "epoch": 0.19942779563253127, |
| "step": 6500 |
| }, |
| { |
| "eval_loss": 0.13607633113861084, |
| "eval_runtime": 2773.0124, |
| "eval_samples_per_second": 161.608, |
| "eval_steps_per_second": 2.525, |
| "epoch": 0.19942779563253127, |
| "step": 6500 |
| }, |
| { |
| "loss": 0.1348, |
| "grad_norm": 0.2550095319747925, |
| "learning_rate": 2.80174367027059e-05, |
| "epoch": 0.20004141961909291, |
| "step": 6520 |
| }, |
| { |
| "loss": 0.137, |
| "grad_norm": 0.14450562000274658, |
| "learning_rate": 2.772996397836704e-05, |
| "epoch": 0.20065504360565453, |
| "step": 6540 |
| }, |
| { |
| "loss": 0.1348, |
| "grad_norm": 0.13089318573474884, |
| "learning_rate": 2.7443406687983265e-05, |
| "epoch": 0.20126866759221618, |
| "step": 6560 |
| }, |
| { |
| "loss": 0.1406, |
| "grad_norm": 0.10475198924541473, |
| "learning_rate": 2.7157776610802415e-05, |
| "epoch": 0.20188229157877782, |
| "step": 6580 |
| }, |
| { |
| "loss": 0.134, |
| "grad_norm": 0.16278155148029327, |
| "learning_rate": 2.687308548795825e-05, |
| "epoch": 0.20249591556533944, |
| "step": 6600 |
| }, |
| { |
| "loss": 0.1342, |
| "grad_norm": 0.19838862121105194, |
| "learning_rate": 2.658934502198772e-05, |
| "epoch": 0.20310953955190109, |
| "step": 6620 |
| }, |
| { |
| "loss": 0.1339, |
| "grad_norm": 0.3514244854450226, |
| "learning_rate": 2.630656687635007e-05, |
| "epoch": 0.20372316353846273, |
| "step": 6640 |
| }, |
| { |
| "loss": 0.1421, |
| "grad_norm": 0.21309345960617065, |
| "learning_rate": 2.6024762674947313e-05, |
| "epoch": 0.20433678752502435, |
| "step": 6660 |
| }, |
| { |
| "loss": 0.1308, |
| "grad_norm": 0.17511409521102905, |
| "learning_rate": 2.574394400164639e-05, |
| "epoch": 0.204950411511586, |
| "step": 6680 |
| }, |
| { |
| "loss": 0.1329, |
| "grad_norm": 0.1690303385257721, |
| "learning_rate": 2.5464122399803125e-05, |
| "epoch": 0.2055640354981476, |
| "step": 6700 |
| }, |
| { |
| "loss": 0.1373, |
| "grad_norm": 0.1409410983324051, |
| "learning_rate": 2.5185309371787513e-05, |
| "epoch": 0.20617765948470926, |
| "step": 6720 |
| }, |
| { |
| "loss": 0.1414, |
| "grad_norm": 0.19619417190551758, |
| "learning_rate": 2.4907516378511135e-05, |
| "epoch": 0.2067912834712709, |
| "step": 6740 |
| }, |
| { |
| "loss": 0.1369, |
| "grad_norm": 0.2113848179578781, |
| "learning_rate": 2.46307548389559e-05, |
| "epoch": 0.20740490745783252, |
| "step": 6760 |
| }, |
| { |
| "loss": 0.14, |
| "grad_norm": 0.1930362433195114, |
| "learning_rate": 2.43550361297047e-05, |
| "epoch": 0.20801853144439417, |
| "step": 6780 |
| }, |
| { |
| "loss": 0.1296, |
| "grad_norm": 0.12248807400465012, |
| "learning_rate": 2.4080371584473748e-05, |
| "epoch": 0.20863215543095578, |
| "step": 6800 |
| }, |
| { |
| "loss": 0.1382, |
| "grad_norm": 0.20456106960773468, |
| "learning_rate": 2.3806772493646723e-05, |
| "epoch": 0.20924577941751743, |
| "step": 6820 |
| }, |
| { |
| "loss": 0.1374, |
| "grad_norm": 0.1549805998802185, |
| "learning_rate": 2.353425010381063e-05, |
| "epoch": 0.20985940340407908, |
| "step": 6840 |
| }, |
| { |
| "loss": 0.1355, |
| "grad_norm": 0.14484341442584991, |
| "learning_rate": 2.3262815617293517e-05, |
| "epoch": 0.2104730273906407, |
| "step": 6860 |
| }, |
| { |
| "loss": 0.1389, |
| "grad_norm": 0.12407292425632477, |
| "learning_rate": 2.2992480191704002e-05, |
| "epoch": 0.21108665137720234, |
| "step": 6880 |
| }, |
| { |
| "loss": 0.1359, |
| "grad_norm": 0.20337499678134918, |
| "learning_rate": 2.272325493947257e-05, |
| "epoch": 0.21170027536376396, |
| "step": 6900 |
| }, |
| { |
| "loss": 0.1408, |
| "grad_norm": 0.1620754599571228, |
| "learning_rate": 2.245515092739488e-05, |
| "epoch": 0.2123138993503256, |
| "step": 6920 |
| }, |
| { |
| "loss": 0.1406, |
| "grad_norm": 0.16787594556808472, |
| "learning_rate": 2.2188179176176766e-05, |
| "epoch": 0.21292752333688725, |
| "step": 6940 |
| }, |
| { |
| "loss": 0.1322, |
| "grad_norm": 0.1271612048149109, |
| "learning_rate": 2.192235065998126e-05, |
| "epoch": 0.21354114732344887, |
| "step": 6960 |
| }, |
| { |
| "loss": 0.1396, |
| "grad_norm": 0.1334231197834015, |
| "learning_rate": 2.165767630597752e-05, |
| "epoch": 0.2141547713100105, |
| "step": 6980 |
| }, |
| { |
| "loss": 0.1246, |
| "grad_norm": 0.1719616800546646, |
| "learning_rate": 2.139416699389153e-05, |
| "epoch": 0.21476839529657216, |
| "step": 7000 |
| }, |
| { |
| "eval_loss": 0.1357450932264328, |
| "eval_runtime": 2775.089, |
| "eval_samples_per_second": 161.487, |
| "eval_steps_per_second": 2.524, |
| "epoch": 0.21476839529657216, |
| "step": 7000 |
| }, |
| { |
| "loss": 0.1331, |
| "grad_norm": 0.1086181178689003, |
| "learning_rate": 2.1131833555559037e-05, |
| "epoch": 0.21538201928313377, |
| "step": 7020 |
| }, |
| { |
| "loss": 0.1359, |
| "grad_norm": 0.13283145427703857, |
| "learning_rate": 2.0870686774480196e-05, |
| "epoch": 0.21599564326969542, |
| "step": 7040 |
| }, |
| { |
| "loss": 0.1333, |
| "grad_norm": 0.17758311331272125, |
| "learning_rate": 2.061073738537635e-05, |
| "epoch": 0.21660926725625704, |
| "step": 7060 |
| }, |
| { |
| "loss": 0.1292, |
| "grad_norm": 0.18912430107593536, |
| "learning_rate": 2.0351996073748713e-05, |
| "epoch": 0.21722289124281868, |
| "step": 7080 |
| }, |
| { |
| "loss": 0.1283, |
| "grad_norm": 0.16375404596328735, |
| "learning_rate": 2.0094473475439202e-05, |
| "epoch": 0.21783651522938033, |
| "step": 7100 |
| }, |
| { |
| "loss": 0.1357, |
| "grad_norm": 0.1428246796131134, |
| "learning_rate": 1.9838180176193178e-05, |
| "epoch": 0.21845013921594195, |
| "step": 7120 |
| }, |
| { |
| "loss": 0.1299, |
| "grad_norm": 0.1967194378376007, |
| "learning_rate": 1.9583126711224343e-05, |
| "epoch": 0.2190637632025036, |
| "step": 7140 |
| }, |
| { |
| "loss": 0.1307, |
| "grad_norm": 0.1559692770242691, |
| "learning_rate": 1.9329323564781682e-05, |
| "epoch": 0.2196773871890652, |
| "step": 7160 |
| }, |
| { |
| "loss": 0.1293, |
| "grad_norm": 0.11402944475412369, |
| "learning_rate": 1.9076781169718428e-05, |
| "epoch": 0.22029101117562686, |
| "step": 7180 |
| }, |
| { |
| "loss": 0.1321, |
| "grad_norm": 0.13555893301963806, |
| "learning_rate": 1.8825509907063327e-05, |
| "epoch": 0.2209046351621885, |
| "step": 7200 |
| }, |
| { |
| "loss": 0.1334, |
| "grad_norm": 0.13484172523021698, |
| "learning_rate": 1.8575520105593817e-05, |
| "epoch": 0.22151825914875012, |
| "step": 7220 |
| }, |
| { |
| "loss": 0.1363, |
| "grad_norm": 0.17900648713111877, |
| "learning_rate": 1.8326822041411524e-05, |
| "epoch": 0.22213188313531176, |
| "step": 7240 |
| }, |
| { |
| "loss": 0.1367, |
| "grad_norm": 0.12890270352363586, |
| "learning_rate": 1.807942593751973e-05, |
| "epoch": 0.22274550712187338, |
| "step": 7260 |
| }, |
| { |
| "loss": 0.1407, |
| "grad_norm": 0.156462624669075, |
| "learning_rate": 1.783334196340331e-05, |
| "epoch": 0.22335913110843503, |
| "step": 7280 |
| }, |
| { |
| "loss": 0.1352, |
| "grad_norm": 0.13109950721263885, |
| "learning_rate": 1.758858023461059e-05, |
| "epoch": 0.22397275509499667, |
| "step": 7300 |
| }, |
| { |
| "loss": 0.1325, |
| "grad_norm": 0.1458718329668045, |
| "learning_rate": 1.7345150812337564e-05, |
| "epoch": 0.2245863790815583, |
| "step": 7320 |
| }, |
| { |
| "loss": 0.1332, |
| "grad_norm": 0.16703669726848602, |
| "learning_rate": 1.7103063703014372e-05, |
| "epoch": 0.22520000306811994, |
| "step": 7340 |
| }, |
| { |
| "loss": 0.1342, |
| "grad_norm": 0.15283693373203278, |
| "learning_rate": 1.6862328857893854e-05, |
| "epoch": 0.22581362705468155, |
| "step": 7360 |
| }, |
| { |
| "loss": 0.1308, |
| "grad_norm": 0.12556903064250946, |
| "learning_rate": 1.66229561726426e-05, |
| "epoch": 0.2264272510412432, |
| "step": 7380 |
| }, |
| { |
| "loss": 0.1289, |
| "grad_norm": 0.11614284664392471, |
| "learning_rate": 1.6384955486934156e-05, |
| "epoch": 0.22704087502780484, |
| "step": 7400 |
| }, |
| { |
| "loss": 0.1372, |
| "grad_norm": 0.10490608215332031, |
| "learning_rate": 1.614833658404454e-05, |
| "epoch": 0.22765449901436646, |
| "step": 7420 |
| }, |
| { |
| "loss": 0.1313, |
| "grad_norm": 0.13296884298324585, |
| "learning_rate": 1.5913109190450032e-05, |
| "epoch": 0.2282681230009281, |
| "step": 7440 |
| }, |
| { |
| "loss": 0.1352, |
| "grad_norm": 0.1308923065662384, |
| "learning_rate": 1.567928297542749e-05, |
| "epoch": 0.22888174698748975, |
| "step": 7460 |
| }, |
| { |
| "loss": 0.1279, |
| "grad_norm": 0.1364806592464447, |
| "learning_rate": 1.544686755065677e-05, |
| "epoch": 0.22949537097405137, |
| "step": 7480 |
| }, |
| { |
| "loss": 0.1401, |
| "grad_norm": 0.19729793071746826, |
| "learning_rate": 1.5215872469825682e-05, |
| "epoch": 0.23010899496061302, |
| "step": 7500 |
| }, |
| { |
| "eval_loss": 0.13517485558986664, |
| "eval_runtime": 2776.5225, |
| "eval_samples_per_second": 161.403, |
| "eval_steps_per_second": 2.522, |
| "epoch": 0.23010899496061302, |
| "step": 7500 |
| }, |
| { |
| "loss": 0.1342, |
| "grad_norm": 0.14020280539989471, |
| "learning_rate": 1.4986307228237268e-05, |
| "epoch": 0.23072261894717463, |
| "step": 7520 |
| }, |
| { |
| "loss": 0.1401, |
| "grad_norm": 0.11004228889942169, |
| "learning_rate": 1.4758181262419423e-05, |
| "epoch": 0.23133624293373628, |
| "step": 7540 |
| }, |
| { |
| "loss": 0.1382, |
| "grad_norm": 0.10248783230781555, |
| "learning_rate": 1.4531503949737108e-05, |
| "epoch": 0.23194986692029793, |
| "step": 7560 |
| }, |
| { |
| "loss": 0.127, |
| "grad_norm": 0.09407743066549301, |
| "learning_rate": 1.4306284608006836e-05, |
| "epoch": 0.23256349090685954, |
| "step": 7580 |
| }, |
| { |
| "loss": 0.1371, |
| "grad_norm": 0.15860360860824585, |
| "learning_rate": 1.4082532495113626e-05, |
| "epoch": 0.2331771148934212, |
| "step": 7600 |
| }, |
| { |
| "loss": 0.1312, |
| "grad_norm": 0.12674517929553986, |
| "learning_rate": 1.3860256808630428e-05, |
| "epoch": 0.2337907388799828, |
| "step": 7620 |
| }, |
| { |
| "loss": 0.1374, |
| "grad_norm": 0.15133555233478546, |
| "learning_rate": 1.3639466685440132e-05, |
| "epoch": 0.23440436286654445, |
| "step": 7640 |
| }, |
| { |
| "loss": 0.1358, |
| "grad_norm": 0.09516581892967224, |
| "learning_rate": 1.3420171201359933e-05, |
| "epoch": 0.2350179868531061, |
| "step": 7660 |
| }, |
| { |
| "loss": 0.1337, |
| "grad_norm": 0.1541604995727539, |
| "learning_rate": 1.3202379370768252e-05, |
| "epoch": 0.23563161083966772, |
| "step": 7680 |
| }, |
| { |
| "loss": 0.1295, |
| "grad_norm": 0.1433165818452835, |
| "learning_rate": 1.2986100146234232e-05, |
| "epoch": 0.23624523482622936, |
| "step": 7700 |
| }, |
| { |
| "loss": 0.1349, |
| "grad_norm": 0.16317042708396912, |
| "learning_rate": 1.2771342418149657e-05, |
| "epoch": 0.23685885881279098, |
| "step": 7720 |
| }, |
| { |
| "loss": 0.1269, |
| "grad_norm": 0.14434409141540527, |
| "learning_rate": 1.2558115014363592e-05, |
| "epoch": 0.23747248279935262, |
| "step": 7740 |
| }, |
| { |
| "loss": 0.1358, |
| "grad_norm": 0.2537558972835541, |
| "learning_rate": 1.2346426699819458e-05, |
| "epoch": 0.23808610678591427, |
| "step": 7760 |
| }, |
| { |
| "loss": 0.137, |
| "grad_norm": 0.16992440819740295, |
| "learning_rate": 1.2136286176194745e-05, |
| "epoch": 0.2386997307724759, |
| "step": 7780 |
| }, |
| { |
| "loss": 0.1333, |
| "grad_norm": 0.14877456426620483, |
| "learning_rate": 1.1927702081543279e-05, |
| "epoch": 0.23931335475903753, |
| "step": 7800 |
| }, |
| { |
| "loss": 0.1321, |
| "grad_norm": 0.15068818628787994, |
| "learning_rate": 1.1720682989940262e-05, |
| "epoch": 0.23992697874559918, |
| "step": 7820 |
| }, |
| { |
| "loss": 0.1329, |
| "grad_norm": 0.10926194489002228, |
| "learning_rate": 1.1515237411129698e-05, |
| "epoch": 0.2405406027321608, |
| "step": 7840 |
| }, |
| { |
| "loss": 0.1349, |
| "grad_norm": 0.1470404714345932, |
| "learning_rate": 1.1311373790174657e-05, |
| "epoch": 0.24115422671872244, |
| "step": 7860 |
| }, |
| { |
| "loss": 0.1368, |
| "grad_norm": 0.1232927218079567, |
| "learning_rate": 1.1109100507110132e-05, |
| "epoch": 0.24176785070528406, |
| "step": 7880 |
| }, |
| { |
| "loss": 0.1312, |
| "grad_norm": 0.14564509689807892, |
| "learning_rate": 1.090842587659851e-05, |
| "epoch": 0.2423814746918457, |
| "step": 7900 |
| }, |
| { |
| "loss": 0.1319, |
| "grad_norm": 0.1518176794052124, |
| "learning_rate": 1.0709358147587884e-05, |
| "epoch": 0.24299509867840735, |
| "step": 7920 |
| }, |
| { |
| "loss": 0.1377, |
| "grad_norm": 0.11598769575357437, |
| "learning_rate": 1.0511905502972886e-05, |
| "epoch": 0.24360872266496897, |
| "step": 7940 |
| }, |
| { |
| "loss": 0.1271, |
| "grad_norm": 0.13424475491046906, |
| "learning_rate": 1.031607605925839e-05, |
| "epoch": 0.24422234665153061, |
| "step": 7960 |
| }, |
| { |
| "loss": 0.1317, |
| "grad_norm": 0.17036224901676178, |
| "learning_rate": 1.0121877866225781e-05, |
| "epoch": 0.24483597063809223, |
| "step": 7980 |
| }, |
| { |
| "loss": 0.1378, |
| "grad_norm": 0.13854670524597168, |
| "learning_rate": 9.929318906602175e-06, |
| "epoch": 0.24544959462465388, |
| "step": 8000 |
| }, |
| { |
| "eval_loss": 0.13417504727840424, |
| "eval_runtime": 2774.5382, |
| "eval_samples_per_second": 161.519, |
| "eval_steps_per_second": 2.524, |
| "epoch": 0.24544959462465388, |
| "step": 8000 |
| }, |
| { |
| "loss": 0.1381, |
| "grad_norm": 0.12813754379749298, |
| "learning_rate": 9.738407095732195e-06, |
| "epoch": 0.24606321861121552, |
| "step": 8020 |
| }, |
| { |
| "loss": 0.1375, |
| "grad_norm": 0.1952831745147705, |
| "learning_rate": 9.549150281252633e-06, |
| "epoch": 0.24667684259777714, |
| "step": 8040 |
| }, |
| { |
| "loss": 0.1363, |
| "grad_norm": 0.17179648578166962, |
| "learning_rate": 9.36155624276987e-06, |
| "epoch": 0.2472904665843388, |
| "step": 8060 |
| }, |
| { |
| "loss": 0.1307, |
| "grad_norm": 0.2890334129333496, |
| "learning_rate": 9.175632691540065e-06, |
| "epoch": 0.2479040905709004, |
| "step": 8080 |
| }, |
| { |
| "loss": 0.1282, |
| "grad_norm": 0.11331592500209808, |
| "learning_rate": 8.991387270152201e-06, |
| "epoch": 0.24851771455746205, |
| "step": 8100 |
| }, |
| { |
| "loss": 0.1356, |
| "grad_norm": 0.619337797164917, |
| "learning_rate": 8.808827552213916e-06, |
| "epoch": 0.2491313385440237, |
| "step": 8120 |
| }, |
| { |
| "loss": 0.1322, |
| "grad_norm": 0.10491683334112167, |
| "learning_rate": 8.627961042040184e-06, |
| "epoch": 0.2497449625305853, |
| "step": 8140 |
| }, |
| { |
| "loss": 0.1323, |
| "grad_norm": 0.1256422996520996, |
| "learning_rate": 8.448795174344804e-06, |
| "epoch": 0.25035858651714693, |
| "step": 8160 |
| }, |
| { |
| "loss": 0.1325, |
| "grad_norm": 0.11729530245065689, |
| "learning_rate": 8.271337313934869e-06, |
| "epoch": 0.2509722105037086, |
| "step": 8180 |
| }, |
| { |
| "loss": 0.1344, |
| "grad_norm": 0.1296742558479309, |
| "learning_rate": 8.09559475540797e-06, |
| "epoch": 0.2515858344902702, |
| "step": 8200 |
| }, |
| { |
| "loss": 0.1293, |
| "grad_norm": 0.11242369562387466, |
| "learning_rate": 7.921574722852343e-06, |
| "epoch": 0.25219945847683184, |
| "step": 8220 |
| }, |
| { |
| "loss": 0.1318, |
| "grad_norm": 0.13417485356330872, |
| "learning_rate": 7.749284369549953e-06, |
| "epoch": 0.2528130824633935, |
| "step": 8240 |
| }, |
| { |
| "loss": 0.1301, |
| "grad_norm": 0.17316614091396332, |
| "learning_rate": 7.578730777682386e-06, |
| "epoch": 0.25342670644995513, |
| "step": 8260 |
| }, |
| { |
| "loss": 0.1373, |
| "grad_norm": 0.1728428602218628, |
| "learning_rate": 7.409920958039795e-06, |
| "epoch": 0.25404033043651675, |
| "step": 8280 |
| }, |
| { |
| "loss": 0.1336, |
| "grad_norm": 0.1023547574877739, |
| "learning_rate": 7.242861849732696e-06, |
| "epoch": 0.2546539544230784, |
| "step": 8300 |
| }, |
| { |
| "loss": 0.1354, |
| "grad_norm": 0.1599491834640503, |
| "learning_rate": 7.077560319906695e-06, |
| "epoch": 0.25526757840964004, |
| "step": 8320 |
| }, |
| { |
| "loss": 0.134, |
| "grad_norm": 0.12360566109418869, |
| "learning_rate": 6.9140231634602485e-06, |
| "epoch": 0.25588120239620166, |
| "step": 8340 |
| }, |
| { |
| "loss": 0.1289, |
| "grad_norm": 0.12993071973323822, |
| "learning_rate": 6.752257102765325e-06, |
| "epoch": 0.2564948263827633, |
| "step": 8360 |
| }, |
| { |
| "loss": 0.1348, |
| "grad_norm": 0.12023476511240005, |
| "learning_rate": 6.592268787391076e-06, |
| "epoch": 0.25710845036932495, |
| "step": 8380 |
| }, |
| { |
| "loss": 0.1348, |
| "grad_norm": 0.13408417999744415, |
| "learning_rate": 6.43406479383053e-06, |
| "epoch": 0.25772207435588657, |
| "step": 8400 |
| }, |
| { |
| "loss": 0.1362, |
| "grad_norm": 0.12605100870132446, |
| "learning_rate": 6.277651625230219e-06, |
| "epoch": 0.2583356983424482, |
| "step": 8420 |
| }, |
| { |
| "loss": 0.1299, |
| "grad_norm": 0.09935203939676285, |
| "learning_rate": 6.12303571112286e-06, |
| "epoch": 0.25894932232900986, |
| "step": 8440 |
| }, |
| { |
| "loss": 0.1329, |
| "grad_norm": 0.21711465716362, |
| "learning_rate": 5.9702234071631e-06, |
| "epoch": 0.2595629463155715, |
| "step": 8460 |
| }, |
| { |
| "loss": 0.1266, |
| "grad_norm": 0.1281883716583252, |
| "learning_rate": 5.819220994866237e-06, |
| "epoch": 0.2601765703021331, |
| "step": 8480 |
| }, |
| { |
| "loss": 0.1375, |
| "grad_norm": 0.1704270988702774, |
| "learning_rate": 5.670034681349995e-06, |
| "epoch": 0.26079019428869477, |
| "step": 8500 |
| }, |
| { |
| "eval_loss": 0.13329795002937317, |
| "eval_runtime": 2777.002, |
| "eval_samples_per_second": 161.375, |
| "eval_steps_per_second": 2.522, |
| "epoch": 0.26079019428869477, |
| "step": 8500 |
| }, |
| { |
| "loss": 0.1369, |
| "grad_norm": 0.11683149635791779, |
| "learning_rate": 5.5226705990794155e-06, |
| "epoch": 0.2614038182752564, |
| "step": 8520 |
| }, |
| { |
| "loss": 0.1244, |
| "grad_norm": 0.11445898562669754, |
| "learning_rate": 5.377134805614714e-06, |
| "epoch": 0.262017442261818, |
| "step": 8540 |
| }, |
| { |
| "loss": 0.1296, |
| "grad_norm": 0.1652945578098297, |
| "learning_rate": 5.233433283362349e-06, |
| "epoch": 0.2626310662483797, |
| "step": 8560 |
| }, |
| { |
| "loss": 0.1309, |
| "grad_norm": 0.15426833927631378, |
| "learning_rate": 5.091571939329048e-06, |
| "epoch": 0.2632446902349413, |
| "step": 8580 |
| }, |
| { |
| "loss": 0.1302, |
| "grad_norm": 0.1100183054804802, |
| "learning_rate": 4.951556604879048e-06, |
| "epoch": 0.2638583142215029, |
| "step": 8600 |
| }, |
| { |
| "loss": 0.1352, |
| "grad_norm": 0.10951848328113556, |
| "learning_rate": 4.813393035494329e-06, |
| "epoch": 0.2644719382080645, |
| "step": 8620 |
| }, |
| { |
| "loss": 0.1288, |
| "grad_norm": 0.1558559685945511, |
| "learning_rate": 4.677086910538092e-06, |
| "epoch": 0.2650855621946262, |
| "step": 8640 |
| }, |
| { |
| "loss": 0.1344, |
| "grad_norm": 0.14536911249160767, |
| "learning_rate": 4.542643833021254e-06, |
| "epoch": 0.2656991861811878, |
| "step": 8660 |
| }, |
| { |
| "loss": 0.1314, |
| "grad_norm": 0.12489516288042068, |
| "learning_rate": 4.410069329372152e-06, |
| "epoch": 0.26631281016774944, |
| "step": 8680 |
| }, |
| { |
| "loss": 0.1296, |
| "grad_norm": 0.10449978709220886, |
| "learning_rate": 4.279368849209381e-06, |
| "epoch": 0.2669264341543111, |
| "step": 8700 |
| }, |
| { |
| "loss": 0.1282, |
| "grad_norm": 0.13451628386974335, |
| "learning_rate": 4.150547765117746e-06, |
| "epoch": 0.2675400581408727, |
| "step": 8720 |
| }, |
| { |
| "loss": 0.1345, |
| "grad_norm": 0.20787014067173004, |
| "learning_rate": 4.023611372427471e-06, |
| "epoch": 0.26815368212743435, |
| "step": 8740 |
| }, |
| { |
| "loss": 0.1362, |
| "grad_norm": 0.14699150621891022, |
| "learning_rate": 3.898564888996476e-06, |
| "epoch": 0.268767306113996, |
| "step": 8760 |
| }, |
| { |
| "loss": 0.1311, |
| "grad_norm": 0.16810518503189087, |
| "learning_rate": 3.7754134549959297e-06, |
| "epoch": 0.26938093010055764, |
| "step": 8780 |
| }, |
| { |
| "loss": 0.1303, |
| "grad_norm": 0.13315120339393616, |
| "learning_rate": 3.654162132698918e-06, |
| "epoch": 0.26999455408711925, |
| "step": 8800 |
| }, |
| { |
| "loss": 0.1294, |
| "grad_norm": 0.10395628213882446, |
| "learning_rate": 3.534815906272404e-06, |
| "epoch": 0.2706081780736809, |
| "step": 8820 |
| }, |
| { |
| "loss": 0.1315, |
| "grad_norm": 0.12636798620224, |
| "learning_rate": 3.417379681572297e-06, |
| "epoch": 0.27122180206024255, |
| "step": 8840 |
| }, |
| { |
| "loss": 0.133, |
| "grad_norm": 0.1581977903842926, |
| "learning_rate": 3.3018582859418446e-06, |
| "epoch": 0.27183542604680416, |
| "step": 8860 |
| }, |
| { |
| "loss": 0.1266, |
| "grad_norm": 0.13160687685012817, |
| "learning_rate": 3.18825646801314e-06, |
| "epoch": 0.2724490500333658, |
| "step": 8880 |
| }, |
| { |
| "loss": 0.1366, |
| "grad_norm": 0.09358909726142883, |
| "learning_rate": 3.076578897511978e-06, |
| "epoch": 0.27306267401992745, |
| "step": 8900 |
| }, |
| { |
| "loss": 0.1302, |
| "grad_norm": 0.14909835159778595, |
| "learning_rate": 2.966830165065876e-06, |
| "epoch": 0.27367629800648907, |
| "step": 8920 |
| }, |
| { |
| "loss": 0.1344, |
| "grad_norm": 0.11751620471477509, |
| "learning_rate": 2.8590147820153513e-06, |
| "epoch": 0.2742899219930507, |
| "step": 8940 |
| }, |
| { |
| "loss": 0.1315, |
| "grad_norm": 0.12908974289894104, |
| "learning_rate": 2.753137180228543e-06, |
| "epoch": 0.27490354597961236, |
| "step": 8960 |
| }, |
| { |
| "loss": 0.136, |
| "grad_norm": 0.12726104259490967, |
| "learning_rate": 2.6492017119189417e-06, |
| "epoch": 0.275517169966174, |
| "step": 8980 |
| }, |
| { |
| "loss": 0.1351, |
| "grad_norm": 0.16233795881271362, |
| "learning_rate": 2.547212649466568e-06, |
| "epoch": 0.2761307939527356, |
| "step": 9000 |
| }, |
| { |
| "eval_loss": 0.13288874924182892, |
| "eval_runtime": 2779.1785, |
| "eval_samples_per_second": 161.249, |
| "eval_steps_per_second": 2.52, |
| "epoch": 0.2761307939527356, |
| "step": 9000 |
| }, |
| { |
| "loss": 0.1356, |
| "grad_norm": 0.15901215374469757, |
| "learning_rate": 2.4471741852423237e-06, |
| "epoch": 0.27674441793929727, |
| "step": 9020 |
| }, |
| { |
| "loss": 0.1329, |
| "grad_norm": 0.11479828506708145, |
| "learning_rate": 2.349090431435641e-06, |
| "epoch": 0.2773580419258589, |
| "step": 9040 |
| }, |
| { |
| "loss": 0.1347, |
| "grad_norm": 0.1306300163269043, |
| "learning_rate": 2.2529654198854835e-06, |
| "epoch": 0.2779716659124205, |
| "step": 9060 |
| }, |
| { |
| "loss": 0.137, |
| "grad_norm": 0.14198105037212372, |
| "learning_rate": 2.1588031019145636e-06, |
| "epoch": 0.2785852898989821, |
| "step": 9080 |
| }, |
| { |
| "loss": 0.1345, |
| "grad_norm": 0.1466785967350006, |
| "learning_rate": 2.066607348166971e-06, |
| "epoch": 0.2791989138855438, |
| "step": 9100 |
| }, |
| { |
| "loss": 0.1324, |
| "grad_norm": 0.1188543438911438, |
| "learning_rate": 1.9763819484490355e-06, |
| "epoch": 0.2798125378721054, |
| "step": 9120 |
| }, |
| { |
| "loss": 0.1374, |
| "grad_norm": 0.14264015853405, |
| "learning_rate": 1.888130611573563e-06, |
| "epoch": 0.28042616185866703, |
| "step": 9140 |
| }, |
| { |
| "loss": 0.133, |
| "grad_norm": 0.10809975862503052, |
| "learning_rate": 1.8018569652073381e-06, |
| "epoch": 0.2810397858452287, |
| "step": 9160 |
| }, |
| { |
| "loss": 0.1323, |
| "grad_norm": 0.14256234467029572, |
| "learning_rate": 1.7175645557220566e-06, |
| "epoch": 0.2816534098317903, |
| "step": 9180 |
| }, |
| { |
| "loss": 0.1327, |
| "grad_norm": 0.12473766505718231, |
| "learning_rate": 1.6352568480485276e-06, |
| "epoch": 0.28226703381835194, |
| "step": 9200 |
| }, |
| { |
| "loss": 0.1311, |
| "grad_norm": 0.0975056141614914, |
| "learning_rate": 1.5549372255342366e-06, |
| "epoch": 0.2828806578049136, |
| "step": 9220 |
| }, |
| { |
| "loss": 0.1349, |
| "grad_norm": 0.13847889006137848, |
| "learning_rate": 1.4766089898042678e-06, |
| "epoch": 0.28349428179147523, |
| "step": 9240 |
| }, |
| { |
| "loss": 0.1332, |
| "grad_norm": 0.10670981556177139, |
| "learning_rate": 1.400275360625608e-06, |
| "epoch": 0.28410790577803685, |
| "step": 9260 |
| }, |
| { |
| "loss": 0.1308, |
| "grad_norm": 0.13052289187908173, |
| "learning_rate": 1.325939475774768e-06, |
| "epoch": 0.2847215297645985, |
| "step": 9280 |
| }, |
| { |
| "loss": 0.1359, |
| "grad_norm": 0.1260124295949936, |
| "learning_rate": 1.2536043909088191e-06, |
| "epoch": 0.28533515375116014, |
| "step": 9300 |
| }, |
| { |
| "loss": 0.129, |
| "grad_norm": 0.11741944402456284, |
| "learning_rate": 1.183273079439795e-06, |
| "epoch": 0.28594877773772176, |
| "step": 9320 |
| }, |
| { |
| "loss": 0.1364, |
| "grad_norm": 0.13720561563968658, |
| "learning_rate": 1.1149484324124327e-06, |
| "epoch": 0.2865624017242834, |
| "step": 9340 |
| }, |
| { |
| "loss": 0.14, |
| "grad_norm": 0.15556901693344116, |
| "learning_rate": 1.0486332583853563e-06, |
| "epoch": 0.28717602571084505, |
| "step": 9360 |
| }, |
| { |
| "loss": 0.1358, |
| "grad_norm": 0.10475285351276398, |
| "learning_rate": 9.843302833156376e-07, |
| "epoch": 0.28778964969740667, |
| "step": 9380 |
| }, |
| { |
| "loss": 0.1302, |
| "grad_norm": 0.1152934804558754, |
| "learning_rate": 9.220421504467281e-07, |
| "epoch": 0.2884032736839683, |
| "step": 9400 |
| }, |
| { |
| "loss": 0.1376, |
| "grad_norm": 0.13443966209888458, |
| "learning_rate": 8.617714201998084e-07, |
| "epoch": 0.28901689767052996, |
| "step": 9420 |
| }, |
| { |
| "loss": 0.1363, |
| "grad_norm": 0.11530063301324844, |
| "learning_rate": 8.035205700685167e-07, |
| "epoch": 0.2896305216570916, |
| "step": 9440 |
| }, |
| { |
| "loss": 0.134, |
| "grad_norm": 0.1621532291173935, |
| "learning_rate": 7.472919945171631e-07, |
| "epoch": 0.2902441456436532, |
| "step": 9460 |
| }, |
| { |
| "loss": 0.1346, |
| "grad_norm": 0.15441814064979553, |
| "learning_rate": 6.93088004882253e-07, |
| "epoch": 0.29085776963021487, |
| "step": 9480 |
| }, |
| { |
| "loss": 0.1273, |
| "grad_norm": 0.13251030445098877, |
| "learning_rate": 6.409108292774913e-07, |
| "epoch": 0.2914713936167765, |
| "step": 9500 |
| }, |
| { |
| "eval_loss": 0.1326712667942047, |
| "eval_runtime": 2779.8317, |
| "eval_samples_per_second": 161.211, |
| "eval_steps_per_second": 2.519, |
| "epoch": 0.2914713936167765, |
| "step": 9500 |
| }, |
| { |
| "loss": 0.1357, |
| "grad_norm": 0.13315628468990326, |
| "learning_rate": 5.907626125022159e-07, |
| "epoch": 0.2920850176033381, |
| "step": 9520 |
| }, |
| { |
| "loss": 0.1344, |
| "grad_norm": 0.11808905005455017, |
| "learning_rate": 5.426454159531913e-07, |
| "epoch": 0.2926986415898997, |
| "step": 9540 |
| }, |
| { |
| "loss": 0.1332, |
| "grad_norm": 0.11968034505844116, |
| "learning_rate": 4.965612175399092e-07, |
| "epoch": 0.2933122655764614, |
| "step": 9560 |
| }, |
| { |
| "loss": 0.1285, |
| "grad_norm": 0.16589680314064026, |
| "learning_rate": 4.52511911603265e-07, |
| "epoch": 0.293925889563023, |
| "step": 9580 |
| }, |
| { |
| "loss": 0.1317, |
| "grad_norm": 0.1595570296049118, |
| "learning_rate": 4.104993088376974e-07, |
| "epoch": 0.29453951354958463, |
| "step": 9600 |
| }, |
| { |
| "loss": 0.135, |
| "grad_norm": 0.10978726297616959, |
| "learning_rate": 3.7052513621674833e-07, |
| "epoch": 0.2951531375361463, |
| "step": 9620 |
| }, |
| { |
| "loss": 0.1319, |
| "grad_norm": 0.10192031413316727, |
| "learning_rate": 3.3259103692209747e-07, |
| "epoch": 0.2957667615227079, |
| "step": 9640 |
| }, |
| { |
| "loss": 0.1324, |
| "grad_norm": 0.10372564941644669, |
| "learning_rate": 2.966985702759828e-07, |
| "epoch": 0.29638038550926954, |
| "step": 9660 |
| }, |
| { |
| "loss": 0.1313, |
| "grad_norm": 0.17446647584438324, |
| "learning_rate": 2.6284921167712973e-07, |
| "epoch": 0.2969940094958312, |
| "step": 9680 |
| }, |
| { |
| "loss": 0.1347, |
| "grad_norm": 0.1544017642736435, |
| "learning_rate": 2.310443525400885e-07, |
| "epoch": 0.29760763348239283, |
| "step": 9700 |
| }, |
| { |
| "loss": 0.1292, |
| "grad_norm": 0.29529738426208496, |
| "learning_rate": 2.012853002380466e-07, |
| "epoch": 0.29822125746895445, |
| "step": 9720 |
| }, |
| { |
| "loss": 0.1311, |
| "grad_norm": 0.14359545707702637, |
| "learning_rate": 1.735732780490884e-07, |
| "epoch": 0.2988348814555161, |
| "step": 9740 |
| }, |
| { |
| "loss": 0.1333, |
| "grad_norm": 0.10971110314130783, |
| "learning_rate": 1.4790942510590766e-07, |
| "epoch": 0.29944850544207774, |
| "step": 9760 |
| }, |
| { |
| "loss": 0.122, |
| "grad_norm": 0.15253949165344238, |
| "learning_rate": 1.2429479634897267e-07, |
| "epoch": 0.30006212942863936, |
| "step": 9780 |
| }, |
| { |
| "loss": 0.1309, |
| "grad_norm": 0.12937602400779724, |
| "learning_rate": 1.0273036248318324e-07, |
| "epoch": 0.300675753415201, |
| "step": 9800 |
| }, |
| { |
| "loss": 0.1267, |
| "grad_norm": 0.12739217281341553, |
| "learning_rate": 8.321700993795811e-08, |
| "epoch": 0.30128937740176265, |
| "step": 9820 |
| }, |
| { |
| "loss": 0.1305, |
| "grad_norm": 0.11524944007396698, |
| "learning_rate": 6.575554083078084e-08, |
| "epoch": 0.30190300138832427, |
| "step": 9840 |
| }, |
| { |
| "loss": 0.1307, |
| "grad_norm": 0.14692504703998566, |
| "learning_rate": 5.0346672934270534e-08, |
| "epoch": 0.3025166253748859, |
| "step": 9860 |
| }, |
| { |
| "loss": 0.1297, |
| "grad_norm": 0.14273245632648468, |
| "learning_rate": 3.699103964661665e-08, |
| "epoch": 0.30313024936144756, |
| "step": 9880 |
| }, |
| { |
| "loss": 0.1271, |
| "grad_norm": 0.13972416520118713, |
| "learning_rate": 2.568918996560532e-08, |
| "epoch": 0.3037438733480092, |
| "step": 9900 |
| }, |
| { |
| "loss": 0.1342, |
| "grad_norm": 0.12904135882854462, |
| "learning_rate": 1.644158846600963e-08, |
| "epoch": 0.3043574973345708, |
| "step": 9920 |
| }, |
| { |
| "loss": 0.1334, |
| "grad_norm": 0.09770902246236801, |
| "learning_rate": 9.248615280499361e-09, |
| "epoch": 0.30497112132113247, |
| "step": 9940 |
| }, |
| { |
| "loss": 0.1388, |
| "grad_norm": 0.10842837393283844, |
| "learning_rate": 4.110566084036816e-09, |
| "epoch": 0.3055847453076941, |
| "step": 9960 |
| }, |
| { |
| "loss": 0.1264, |
| "grad_norm": 0.10802903026342392, |
| "learning_rate": 1.0276520816976387e-09, |
| "epoch": 0.3061983692942557, |
| "step": 9980 |
| }, |
| { |
| "loss": 0.1283, |
| "grad_norm": 0.14172281324863434, |
| "learning_rate": 0.0, |
| "epoch": 0.3068119932808174, |
| "step": 10000 |
| }, |
| { |
| "eval_loss": 0.1326160877943039, |
| "eval_runtime": 2780.1651, |
| "eval_samples_per_second": 161.192, |
| "eval_steps_per_second": 2.519, |
| "epoch": 0.3068119932808174, |
| "step": 10000 |
| }, |
| { |
| "train_runtime": 73306.6559, |
| "train_samples_per_second": 17.461, |
| "train_steps_per_second": 0.136, |
| "total_flos": 4.748331528373666e+18, |
| "train_loss": 0.17932859930992126, |
| "epoch": 0.3068119932808174, |
| "step": 10000 |
| } |
| ], |
| "best_metric": 0.1326160877943039, |
| "best_model_checkpoint": "/content/drive/MyDrive/ptm-llama/saves/ptm-llama/checkpoint-10000", |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "is_hyper_param_search": false, |
| "trial_name": null, |
| "trial_params": null, |
| "stateful_callbacks": { |
| "EarlyStoppingCallback": { |
| "args": { |
| "early_stopping_patience": 5, |
| "early_stopping_threshold": 0.0 |
| }, |
| "attributes": { |
| "early_stopping_patience_counter": 0 |
| } |
| }, |
| "TrainerControl": { |
| "args": { |
| "should_training_stop": true, |
| "should_epoch_stop": false, |
| "should_save": true, |
| "should_evaluate": false, |
| "should_log": false |
| }, |
| "attributes": {} |
| } |
| }, |
| "model": "PeftModelForCausalLM(\n (base_model): LoraModel(\n (model): LlamaForCausalLM(\n (model): LlamaModel(\n (embed_tokens): Embedding(32000, 4096, padding_idx=0)\n (layers): ModuleList(\n (0-31): 32 x LlamaDecoderLayer(\n (self_attn): LlamaAttention(\n (q_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=4096, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=4096, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (k_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=4096, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=4096, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (v_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=4096, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=4096, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (o_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=4096, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=4096, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n )\n (mlp): LlamaMLP(\n (gate_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=11008, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=11008, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (up_proj): lora.Linear(\n (base_layer): Linear(in_features=4096, out_features=11008, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=4096, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=11008, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (down_proj): lora.Linear(\n (base_layer): Linear(in_features=11008, out_features=4096, bias=False)\n (lora_dropout): ModuleDict(\n (default): Dropout(p=0.05, inplace=False)\n )\n (lora_A): ModuleDict(\n (default): Linear(in_features=11008, out_features=64, bias=False)\n )\n (lora_B): ModuleDict(\n (default): Linear(in_features=64, out_features=4096, bias=False)\n )\n (lora_embedding_A): ParameterDict()\n (lora_embedding_B): ParameterDict()\n (lora_magnitude_vector): ModuleDict()\n )\n (act_fn): SiLU()\n )\n (input_layernorm): LlamaRMSNorm((4096,), eps=1e-05)\n (post_attention_layernorm): LlamaRMSNorm((4096,), eps=1e-05)\n )\n )\n (norm): LlamaRMSNorm((4096,), eps=1e-05)\n (rotary_emb): LlamaRotaryEmbedding()\n )\n (lm_head): Linear(in_features=4096, out_features=32000, bias=False)\n )\n )\n)", |
| "optimizer": "AcceleratedOptimizer (\nParameter Group 0\n amsgrad: False\n betas: (0.9, 0.999)\n capturable: False\n decoupled_weight_decay: True\n differentiable: False\n eps: 1e-08\n foreach: None\n fused: None\n initial_lr: 0.0001\n lr: 0.0\n maximize: False\n weight_decay: 0.0\n\nParameter Group 1\n amsgrad: False\n betas: (0.9, 0.999)\n capturable: False\n decoupled_weight_decay: True\n differentiable: False\n eps: 1e-08\n foreach: None\n fused: None\n initial_lr: 0.0001\n lr: 0.0\n maximize: False\n weight_decay: 0.0\n)", |
| "lr_scheduler": "<torch.optim.lr_scheduler.LambdaLR object at 0x7f9595badc40>", |
| "train_dataloader": "<accelerate.data_loader.DataLoaderShard object at 0x7f95a07191c0>" |
| } |
