| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.9988901220865705, | |
| "eval_steps": 500, | |
| "global_step": 225, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.004439511653718091, | |
| "grad_norm": 37.533348083496094, | |
| "learning_rate": 2.8571428571428573e-06, | |
| "loss": 1.824, | |
| "step": 1 | |
| }, | |
| { | |
| "epoch": 0.008879023307436182, | |
| "grad_norm": 46.24335479736328, | |
| "learning_rate": 5.7142857142857145e-06, | |
| "loss": 1.8786, | |
| "step": 2 | |
| }, | |
| { | |
| "epoch": 0.013318534961154272, | |
| "grad_norm": 42.045166015625, | |
| "learning_rate": 8.571428571428571e-06, | |
| "loss": 1.7863, | |
| "step": 3 | |
| }, | |
| { | |
| "epoch": 0.017758046614872364, | |
| "grad_norm": 8.661040306091309, | |
| "learning_rate": 1.1428571428571429e-05, | |
| "loss": 1.2891, | |
| "step": 4 | |
| }, | |
| { | |
| "epoch": 0.022197558268590455, | |
| "grad_norm": 3.85249400138855, | |
| "learning_rate": 1.4285714285714287e-05, | |
| "loss": 1.2127, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.026637069922308545, | |
| "grad_norm": 7.926912307739258, | |
| "learning_rate": 1.7142857142857142e-05, | |
| "loss": 1.2012, | |
| "step": 6 | |
| }, | |
| { | |
| "epoch": 0.03107658157602664, | |
| "grad_norm": 4.07999324798584, | |
| "learning_rate": 2e-05, | |
| "loss": 1.2275, | |
| "step": 7 | |
| }, | |
| { | |
| "epoch": 0.03551609322974473, | |
| "grad_norm": 3.668304681777954, | |
| "learning_rate": 1.9998961636899736e-05, | |
| "loss": 1.2301, | |
| "step": 8 | |
| }, | |
| { | |
| "epoch": 0.03995560488346282, | |
| "grad_norm": 3.8191933631896973, | |
| "learning_rate": 1.9995846763238514e-05, | |
| "loss": 1.18, | |
| "step": 9 | |
| }, | |
| { | |
| "epoch": 0.04439511653718091, | |
| "grad_norm": 3.0913546085357666, | |
| "learning_rate": 1.9990656025890315e-05, | |
| "loss": 1.1692, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.048834628190899, | |
| "grad_norm": 2.7952189445495605, | |
| "learning_rate": 1.9983390502829168e-05, | |
| "loss": 1.2112, | |
| "step": 11 | |
| }, | |
| { | |
| "epoch": 0.05327413984461709, | |
| "grad_norm": 2.480602979660034, | |
| "learning_rate": 1.997405170290528e-05, | |
| "loss": 1.2294, | |
| "step": 12 | |
| }, | |
| { | |
| "epoch": 0.05771365149833518, | |
| "grad_norm": 2.486560344696045, | |
| "learning_rate": 1.9962641565531694e-05, | |
| "loss": 1.222, | |
| "step": 13 | |
| }, | |
| { | |
| "epoch": 0.06215316315205328, | |
| "grad_norm": 2.59287428855896, | |
| "learning_rate": 1.994916246028154e-05, | |
| "loss": 1.1607, | |
| "step": 14 | |
| }, | |
| { | |
| "epoch": 0.06659267480577137, | |
| "grad_norm": 2.5444483757019043, | |
| "learning_rate": 1.9933617186395917e-05, | |
| "loss": 1.2571, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.07103218645948946, | |
| "grad_norm": 2.5133490562438965, | |
| "learning_rate": 1.9916008972202586e-05, | |
| "loss": 1.1338, | |
| "step": 16 | |
| }, | |
| { | |
| "epoch": 0.07547169811320754, | |
| "grad_norm": 2.2021095752716064, | |
| "learning_rate": 1.9896341474445526e-05, | |
| "loss": 1.2253, | |
| "step": 17 | |
| }, | |
| { | |
| "epoch": 0.07991120976692564, | |
| "grad_norm": 2.341919183731079, | |
| "learning_rate": 1.987461877752552e-05, | |
| "loss": 1.2575, | |
| "step": 18 | |
| }, | |
| { | |
| "epoch": 0.08435072142064373, | |
| "grad_norm": 2.27514386177063, | |
| "learning_rate": 1.985084539265195e-05, | |
| "loss": 1.1895, | |
| "step": 19 | |
| }, | |
| { | |
| "epoch": 0.08879023307436182, | |
| "grad_norm": 2.0287599563598633, | |
| "learning_rate": 1.982502625690595e-05, | |
| "loss": 1.122, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0932297447280799, | |
| "grad_norm": 2.33461594581604, | |
| "learning_rate": 1.9797166732215078e-05, | |
| "loss": 1.2319, | |
| "step": 21 | |
| }, | |
| { | |
| "epoch": 0.097669256381798, | |
| "grad_norm": 2.374116897583008, | |
| "learning_rate": 1.9767272604239823e-05, | |
| "loss": 1.2025, | |
| "step": 22 | |
| }, | |
| { | |
| "epoch": 0.10210876803551609, | |
| "grad_norm": 2.2956087589263916, | |
| "learning_rate": 1.973535008117207e-05, | |
| "loss": 1.2078, | |
| "step": 23 | |
| }, | |
| { | |
| "epoch": 0.10654827968923418, | |
| "grad_norm": 2.1211025714874268, | |
| "learning_rate": 1.9701405792445815e-05, | |
| "loss": 1.1912, | |
| "step": 24 | |
| }, | |
| { | |
| "epoch": 0.11098779134295228, | |
| "grad_norm": 2.2014005184173584, | |
| "learning_rate": 1.9665446787360444e-05, | |
| "loss": 1.1932, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.11542730299667037, | |
| "grad_norm": 2.1209704875946045, | |
| "learning_rate": 1.962748053361675e-05, | |
| "loss": 1.2348, | |
| "step": 26 | |
| }, | |
| { | |
| "epoch": 0.11986681465038845, | |
| "grad_norm": 2.0373430252075195, | |
| "learning_rate": 1.9587514915766124e-05, | |
| "loss": 1.2503, | |
| "step": 27 | |
| }, | |
| { | |
| "epoch": 0.12430632630410655, | |
| "grad_norm": 2.3817737102508545, | |
| "learning_rate": 1.9545558233573136e-05, | |
| "loss": 1.2131, | |
| "step": 28 | |
| }, | |
| { | |
| "epoch": 0.12874583795782463, | |
| "grad_norm": 2.1462056636810303, | |
| "learning_rate": 1.950161920029191e-05, | |
| "loss": 1.1736, | |
| "step": 29 | |
| }, | |
| { | |
| "epoch": 0.13318534961154274, | |
| "grad_norm": 2.1412036418914795, | |
| "learning_rate": 1.9455706940856602e-05, | |
| "loss": 1.1716, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.13762486126526083, | |
| "grad_norm": 2.2089121341705322, | |
| "learning_rate": 1.940783098998643e-05, | |
| "loss": 1.169, | |
| "step": 31 | |
| }, | |
| { | |
| "epoch": 0.14206437291897892, | |
| "grad_norm": 2.101334571838379, | |
| "learning_rate": 1.9358001290205542e-05, | |
| "loss": 1.2226, | |
| "step": 32 | |
| }, | |
| { | |
| "epoch": 0.146503884572697, | |
| "grad_norm": 2.436990976333618, | |
| "learning_rate": 1.9306228189778255e-05, | |
| "loss": 1.192, | |
| "step": 33 | |
| }, | |
| { | |
| "epoch": 0.1509433962264151, | |
| "grad_norm": 2.069035768508911, | |
| "learning_rate": 1.925252244055998e-05, | |
| "loss": 1.2419, | |
| "step": 34 | |
| }, | |
| { | |
| "epoch": 0.15538290788013318, | |
| "grad_norm": 2.099047899246216, | |
| "learning_rate": 1.9196895195764363e-05, | |
| "loss": 1.2004, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.1598224195338513, | |
| "grad_norm": 2.2311224937438965, | |
| "learning_rate": 1.9139358007647085e-05, | |
| "loss": 1.2065, | |
| "step": 36 | |
| }, | |
| { | |
| "epoch": 0.16426193118756938, | |
| "grad_norm": 2.1131861209869385, | |
| "learning_rate": 1.907992282510675e-05, | |
| "loss": 1.2158, | |
| "step": 37 | |
| }, | |
| { | |
| "epoch": 0.16870144284128746, | |
| "grad_norm": 1.924771785736084, | |
| "learning_rate": 1.901860199120344e-05, | |
| "loss": 1.2714, | |
| "step": 38 | |
| }, | |
| { | |
| "epoch": 0.17314095449500555, | |
| "grad_norm": 1.9812465906143188, | |
| "learning_rate": 1.8955408240595396e-05, | |
| "loss": 1.2216, | |
| "step": 39 | |
| }, | |
| { | |
| "epoch": 0.17758046614872364, | |
| "grad_norm": 1.9320334196090698, | |
| "learning_rate": 1.8890354696894374e-05, | |
| "loss": 1.2164, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.18201997780244172, | |
| "grad_norm": 1.9847488403320312, | |
| "learning_rate": 1.8823454869940243e-05, | |
| "loss": 1.1705, | |
| "step": 41 | |
| }, | |
| { | |
| "epoch": 0.1864594894561598, | |
| "grad_norm": 1.973925232887268, | |
| "learning_rate": 1.8754722652995346e-05, | |
| "loss": 1.1984, | |
| "step": 42 | |
| }, | |
| { | |
| "epoch": 0.19089900110987792, | |
| "grad_norm": 2.0115818977355957, | |
| "learning_rate": 1.8684172319859258e-05, | |
| "loss": 1.1659, | |
| "step": 43 | |
| }, | |
| { | |
| "epoch": 0.195338512763596, | |
| "grad_norm": 2.3217947483062744, | |
| "learning_rate": 1.861181852190451e-05, | |
| "loss": 1.2463, | |
| "step": 44 | |
| }, | |
| { | |
| "epoch": 0.1997780244173141, | |
| "grad_norm": 2.1839258670806885, | |
| "learning_rate": 1.8537676285033886e-05, | |
| "loss": 1.1657, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.20421753607103219, | |
| "grad_norm": 1.9523669481277466, | |
| "learning_rate": 1.8461761006559982e-05, | |
| "loss": 1.1766, | |
| "step": 46 | |
| }, | |
| { | |
| "epoch": 0.20865704772475027, | |
| "grad_norm": 2.1495866775512695, | |
| "learning_rate": 1.838408845200758e-05, | |
| "loss": 1.2138, | |
| "step": 47 | |
| }, | |
| { | |
| "epoch": 0.21309655937846836, | |
| "grad_norm": 2.060826539993286, | |
| "learning_rate": 1.8304674751839583e-05, | |
| "loss": 1.1817, | |
| "step": 48 | |
| }, | |
| { | |
| "epoch": 0.21753607103218647, | |
| "grad_norm": 2.0041329860687256, | |
| "learning_rate": 1.8223536398107177e-05, | |
| "loss": 1.2327, | |
| "step": 49 | |
| }, | |
| { | |
| "epoch": 0.22197558268590456, | |
| "grad_norm": 2.427309989929199, | |
| "learning_rate": 1.8140690241024872e-05, | |
| "loss": 1.2012, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.22641509433962265, | |
| "grad_norm": 2.0693812370300293, | |
| "learning_rate": 1.8056153485471167e-05, | |
| "loss": 1.1806, | |
| "step": 51 | |
| }, | |
| { | |
| "epoch": 0.23085460599334073, | |
| "grad_norm": 2.1914312839508057, | |
| "learning_rate": 1.7969943687415575e-05, | |
| "loss": 1.1908, | |
| "step": 52 | |
| }, | |
| { | |
| "epoch": 0.23529411764705882, | |
| "grad_norm": 2.1551156044006348, | |
| "learning_rate": 1.788207875027274e-05, | |
| "loss": 1.2127, | |
| "step": 53 | |
| }, | |
| { | |
| "epoch": 0.2397336293007769, | |
| "grad_norm": 2.090184211730957, | |
| "learning_rate": 1.7792576921184374e-05, | |
| "loss": 1.1944, | |
| "step": 54 | |
| }, | |
| { | |
| "epoch": 0.244173140954495, | |
| "grad_norm": 1.9871045351028442, | |
| "learning_rate": 1.7701456787229805e-05, | |
| "loss": 1.1736, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.2486126526082131, | |
| "grad_norm": 2.1655359268188477, | |
| "learning_rate": 1.7608737271566004e-05, | |
| "loss": 1.2267, | |
| "step": 56 | |
| }, | |
| { | |
| "epoch": 0.25305216426193117, | |
| "grad_norm": 2.2310895919799805, | |
| "learning_rate": 1.751443762949772e-05, | |
| "loss": 1.232, | |
| "step": 57 | |
| }, | |
| { | |
| "epoch": 0.25749167591564925, | |
| "grad_norm": 2.07645583152771, | |
| "learning_rate": 1.741857744447869e-05, | |
| "loss": 1.1872, | |
| "step": 58 | |
| }, | |
| { | |
| "epoch": 0.2619311875693674, | |
| "grad_norm": 2.183539628982544, | |
| "learning_rate": 1.732117662404469e-05, | |
| "loss": 1.253, | |
| "step": 59 | |
| }, | |
| { | |
| "epoch": 0.2663706992230855, | |
| "grad_norm": 2.078212261199951, | |
| "learning_rate": 1.7222255395679298e-05, | |
| "loss": 1.2586, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.27081021087680357, | |
| "grad_norm": 1.9478346109390259, | |
| "learning_rate": 1.712183430261319e-05, | |
| "loss": 1.3174, | |
| "step": 61 | |
| }, | |
| { | |
| "epoch": 0.27524972253052166, | |
| "grad_norm": 1.9654473066329956, | |
| "learning_rate": 1.7019934199557868e-05, | |
| "loss": 1.2072, | |
| "step": 62 | |
| }, | |
| { | |
| "epoch": 0.27968923418423974, | |
| "grad_norm": 2.0590426921844482, | |
| "learning_rate": 1.691657624837472e-05, | |
| "loss": 1.2234, | |
| "step": 63 | |
| }, | |
| { | |
| "epoch": 0.28412874583795783, | |
| "grad_norm": 1.9870823621749878, | |
| "learning_rate": 1.6811781913680273e-05, | |
| "loss": 1.2366, | |
| "step": 64 | |
| }, | |
| { | |
| "epoch": 0.2885682574916759, | |
| "grad_norm": 2.0088582038879395, | |
| "learning_rate": 1.6705572958388576e-05, | |
| "loss": 1.1597, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.293007769145394, | |
| "grad_norm": 2.107692241668701, | |
| "learning_rate": 1.659797143919165e-05, | |
| "loss": 1.1738, | |
| "step": 66 | |
| }, | |
| { | |
| "epoch": 0.2974472807991121, | |
| "grad_norm": 1.8893859386444092, | |
| "learning_rate": 1.6488999701978905e-05, | |
| "loss": 1.1214, | |
| "step": 67 | |
| }, | |
| { | |
| "epoch": 0.3018867924528302, | |
| "grad_norm": 1.8780573606491089, | |
| "learning_rate": 1.6378680377196526e-05, | |
| "loss": 1.1799, | |
| "step": 68 | |
| }, | |
| { | |
| "epoch": 0.30632630410654826, | |
| "grad_norm": 2.0093393325805664, | |
| "learning_rate": 1.6267036375147728e-05, | |
| "loss": 1.187, | |
| "step": 69 | |
| }, | |
| { | |
| "epoch": 0.31076581576026635, | |
| "grad_norm": 1.988927960395813, | |
| "learning_rate": 1.615409088123493e-05, | |
| "loss": 1.2668, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.31520532741398444, | |
| "grad_norm": 1.8668824434280396, | |
| "learning_rate": 1.6039867351144778e-05, | |
| "loss": 1.2241, | |
| "step": 71 | |
| }, | |
| { | |
| "epoch": 0.3196448390677026, | |
| "grad_norm": 2.02083420753479, | |
| "learning_rate": 1.5924389505977038e-05, | |
| "loss": 1.228, | |
| "step": 72 | |
| }, | |
| { | |
| "epoch": 0.32408435072142067, | |
| "grad_norm": 2.080263137817383, | |
| "learning_rate": 1.5807681327318372e-05, | |
| "loss": 1.1584, | |
| "step": 73 | |
| }, | |
| { | |
| "epoch": 0.32852386237513875, | |
| "grad_norm": 1.9211077690124512, | |
| "learning_rate": 1.5689767052262028e-05, | |
| "loss": 1.2324, | |
| "step": 74 | |
| }, | |
| { | |
| "epoch": 0.33296337402885684, | |
| "grad_norm": 2.1998391151428223, | |
| "learning_rate": 1.557067116837444e-05, | |
| "loss": 1.1829, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.3374028856825749, | |
| "grad_norm": 1.973344087600708, | |
| "learning_rate": 1.545041840860986e-05, | |
| "loss": 1.233, | |
| "step": 76 | |
| }, | |
| { | |
| "epoch": 0.341842397336293, | |
| "grad_norm": 2.0612385272979736, | |
| "learning_rate": 1.5329033746173975e-05, | |
| "loss": 1.2338, | |
| "step": 77 | |
| }, | |
| { | |
| "epoch": 0.3462819089900111, | |
| "grad_norm": 2.0529823303222656, | |
| "learning_rate": 1.520654238933767e-05, | |
| "loss": 1.1973, | |
| "step": 78 | |
| }, | |
| { | |
| "epoch": 0.3507214206437292, | |
| "grad_norm": 1.8720424175262451, | |
| "learning_rate": 1.5082969776201948e-05, | |
| "loss": 1.2133, | |
| "step": 79 | |
| }, | |
| { | |
| "epoch": 0.3551609322974473, | |
| "grad_norm": 1.9517823457717896, | |
| "learning_rate": 1.4958341569415149e-05, | |
| "loss": 1.2102, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.35960044395116536, | |
| "grad_norm": 1.9612551927566528, | |
| "learning_rate": 1.483268365084351e-05, | |
| "loss": 1.1888, | |
| "step": 81 | |
| }, | |
| { | |
| "epoch": 0.36403995560488345, | |
| "grad_norm": 2.1608946323394775, | |
| "learning_rate": 1.4706022116196208e-05, | |
| "loss": 1.2082, | |
| "step": 82 | |
| }, | |
| { | |
| "epoch": 0.36847946725860153, | |
| "grad_norm": 1.951665997505188, | |
| "learning_rate": 1.4578383269606004e-05, | |
| "loss": 1.2041, | |
| "step": 83 | |
| }, | |
| { | |
| "epoch": 0.3729189789123196, | |
| "grad_norm": 1.9137777090072632, | |
| "learning_rate": 1.4449793618166594e-05, | |
| "loss": 1.2091, | |
| "step": 84 | |
| }, | |
| { | |
| "epoch": 0.37735849056603776, | |
| "grad_norm": 1.9383291006088257, | |
| "learning_rate": 1.4320279866427798e-05, | |
| "loss": 1.1901, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.38179800221975585, | |
| "grad_norm": 1.8352899551391602, | |
| "learning_rate": 1.4189868910849779e-05, | |
| "loss": 1.1741, | |
| "step": 86 | |
| }, | |
| { | |
| "epoch": 0.38623751387347394, | |
| "grad_norm": 1.9220095872879028, | |
| "learning_rate": 1.4058587834217356e-05, | |
| "loss": 1.1934, | |
| "step": 87 | |
| }, | |
| { | |
| "epoch": 0.390677025527192, | |
| "grad_norm": 1.7801581621170044, | |
| "learning_rate": 1.392646390001569e-05, | |
| "loss": 1.1439, | |
| "step": 88 | |
| }, | |
| { | |
| "epoch": 0.3951165371809101, | |
| "grad_norm": 1.8106623888015747, | |
| "learning_rate": 1.3793524546768358e-05, | |
| "loss": 1.1316, | |
| "step": 89 | |
| }, | |
| { | |
| "epoch": 0.3995560488346282, | |
| "grad_norm": 1.822784423828125, | |
| "learning_rate": 1.3659797382339162e-05, | |
| "loss": 1.2326, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.4039955604883463, | |
| "grad_norm": 2.061378240585327, | |
| "learning_rate": 1.3525310178198707e-05, | |
| "loss": 1.2022, | |
| "step": 91 | |
| }, | |
| { | |
| "epoch": 0.40843507214206437, | |
| "grad_norm": 1.999387264251709, | |
| "learning_rate": 1.3390090863657048e-05, | |
| "loss": 1.2236, | |
| "step": 92 | |
| }, | |
| { | |
| "epoch": 0.41287458379578246, | |
| "grad_norm": 1.7549680471420288, | |
| "learning_rate": 1.325416752006351e-05, | |
| "loss": 1.1421, | |
| "step": 93 | |
| }, | |
| { | |
| "epoch": 0.41731409544950054, | |
| "grad_norm": 1.8363287448883057, | |
| "learning_rate": 1.311756837497499e-05, | |
| "loss": 1.1325, | |
| "step": 94 | |
| }, | |
| { | |
| "epoch": 0.42175360710321863, | |
| "grad_norm": 1.942063808441162, | |
| "learning_rate": 1.2980321796293838e-05, | |
| "loss": 1.1475, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.4261931187569367, | |
| "grad_norm": 2.029550790786743, | |
| "learning_rate": 1.284245628637665e-05, | |
| "loss": 1.1715, | |
| "step": 96 | |
| }, | |
| { | |
| "epoch": 0.4306326304106548, | |
| "grad_norm": 2.0929226875305176, | |
| "learning_rate": 1.2704000476115079e-05, | |
| "loss": 1.1595, | |
| "step": 97 | |
| }, | |
| { | |
| "epoch": 0.43507214206437295, | |
| "grad_norm": 2.0529980659484863, | |
| "learning_rate": 1.256498311899001e-05, | |
| "loss": 1.193, | |
| "step": 98 | |
| }, | |
| { | |
| "epoch": 0.43951165371809103, | |
| "grad_norm": 1.9034109115600586, | |
| "learning_rate": 1.2425433085100224e-05, | |
| "loss": 1.1566, | |
| "step": 99 | |
| }, | |
| { | |
| "epoch": 0.4439511653718091, | |
| "grad_norm": 2.2665998935699463, | |
| "learning_rate": 1.2285379355166893e-05, | |
| "loss": 1.1709, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.4483906770255272, | |
| "grad_norm": 2.0044686794281006, | |
| "learning_rate": 1.2144851014515055e-05, | |
| "loss": 1.1688, | |
| "step": 101 | |
| }, | |
| { | |
| "epoch": 0.4528301886792453, | |
| "grad_norm": 1.9301621913909912, | |
| "learning_rate": 1.2003877247033411e-05, | |
| "loss": 1.1855, | |
| "step": 102 | |
| }, | |
| { | |
| "epoch": 0.4572697003329634, | |
| "grad_norm": 1.996848464012146, | |
| "learning_rate": 1.1862487329113606e-05, | |
| "loss": 1.1603, | |
| "step": 103 | |
| }, | |
| { | |
| "epoch": 0.46170921198668147, | |
| "grad_norm": 2.0279271602630615, | |
| "learning_rate": 1.172071062357035e-05, | |
| "loss": 1.1789, | |
| "step": 104 | |
| }, | |
| { | |
| "epoch": 0.46614872364039955, | |
| "grad_norm": 1.9275884628295898, | |
| "learning_rate": 1.1578576573543541e-05, | |
| "loss": 1.1339, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.47058823529411764, | |
| "grad_norm": 1.9847296476364136, | |
| "learning_rate": 1.1436114696383749e-05, | |
| "loss": 1.186, | |
| "step": 106 | |
| }, | |
| { | |
| "epoch": 0.4750277469478357, | |
| "grad_norm": 1.9284783601760864, | |
| "learning_rate": 1.1293354577522264e-05, | |
| "loss": 1.1697, | |
| "step": 107 | |
| }, | |
| { | |
| "epoch": 0.4794672586015538, | |
| "grad_norm": 1.8462920188903809, | |
| "learning_rate": 1.1150325864327003e-05, | |
| "loss": 1.2233, | |
| "step": 108 | |
| }, | |
| { | |
| "epoch": 0.4839067702552719, | |
| "grad_norm": 1.9206749200820923, | |
| "learning_rate": 1.1007058259945584e-05, | |
| "loss": 1.2148, | |
| "step": 109 | |
| }, | |
| { | |
| "epoch": 0.48834628190899, | |
| "grad_norm": 1.9734727144241333, | |
| "learning_rate": 1.0863581517136776e-05, | |
| "loss": 1.1367, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.49278579356270813, | |
| "grad_norm": 1.83944833278656, | |
| "learning_rate": 1.0719925432091671e-05, | |
| "loss": 1.1834, | |
| "step": 111 | |
| }, | |
| { | |
| "epoch": 0.4972253052164262, | |
| "grad_norm": 1.9524726867675781, | |
| "learning_rate": 1.0576119838245843e-05, | |
| "loss": 1.1046, | |
| "step": 112 | |
| }, | |
| { | |
| "epoch": 0.5016648168701443, | |
| "grad_norm": 1.8533103466033936, | |
| "learning_rate": 1.043219460008374e-05, | |
| "loss": 1.1694, | |
| "step": 113 | |
| }, | |
| { | |
| "epoch": 0.5061043285238623, | |
| "grad_norm": 2.0331192016601562, | |
| "learning_rate": 1.0288179606936666e-05, | |
| "loss": 1.2525, | |
| "step": 114 | |
| }, | |
| { | |
| "epoch": 0.5105438401775805, | |
| "grad_norm": 1.7501027584075928, | |
| "learning_rate": 1.0144104766775574e-05, | |
| "loss": 1.1787, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.5149833518312985, | |
| "grad_norm": 1.8049185276031494, | |
| "learning_rate": 1e-05, | |
| "loss": 1.1388, | |
| "step": 116 | |
| }, | |
| { | |
| "epoch": 0.5194228634850167, | |
| "grad_norm": 1.9093164205551147, | |
| "learning_rate": 9.855895233224431e-06, | |
| "loss": 1.1042, | |
| "step": 117 | |
| }, | |
| { | |
| "epoch": 0.5238623751387348, | |
| "grad_norm": 1.7633061408996582, | |
| "learning_rate": 9.711820393063337e-06, | |
| "loss": 1.1286, | |
| "step": 118 | |
| }, | |
| { | |
| "epoch": 0.5283018867924528, | |
| "grad_norm": 1.8089503049850464, | |
| "learning_rate": 9.56780539991626e-06, | |
| "loss": 1.1385, | |
| "step": 119 | |
| }, | |
| { | |
| "epoch": 0.532741398446171, | |
| "grad_norm": 1.827928066253662, | |
| "learning_rate": 9.423880161754158e-06, | |
| "loss": 1.1864, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.537180910099889, | |
| "grad_norm": 1.9098917245864868, | |
| "learning_rate": 9.28007456790833e-06, | |
| "loss": 1.1412, | |
| "step": 121 | |
| }, | |
| { | |
| "epoch": 0.5416204217536071, | |
| "grad_norm": 1.7700555324554443, | |
| "learning_rate": 9.13641848286323e-06, | |
| "loss": 1.1733, | |
| "step": 122 | |
| }, | |
| { | |
| "epoch": 0.5460599334073252, | |
| "grad_norm": 1.829474687576294, | |
| "learning_rate": 8.992941740054418e-06, | |
| "loss": 1.1297, | |
| "step": 123 | |
| }, | |
| { | |
| "epoch": 0.5504994450610433, | |
| "grad_norm": 1.9246844053268433, | |
| "learning_rate": 8.849674135672999e-06, | |
| "loss": 1.2096, | |
| "step": 124 | |
| }, | |
| { | |
| "epoch": 0.5549389567147613, | |
| "grad_norm": 1.9404821395874023, | |
| "learning_rate": 8.706645422477739e-06, | |
| "loss": 1.1531, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.5593784683684795, | |
| "grad_norm": 1.8768739700317383, | |
| "learning_rate": 8.563885303616253e-06, | |
| "loss": 1.1896, | |
| "step": 126 | |
| }, | |
| { | |
| "epoch": 0.5638179800221975, | |
| "grad_norm": 1.9277173280715942, | |
| "learning_rate": 8.42142342645646e-06, | |
| "loss": 1.112, | |
| "step": 127 | |
| }, | |
| { | |
| "epoch": 0.5682574916759157, | |
| "grad_norm": 1.9244784116744995, | |
| "learning_rate": 8.279289376429653e-06, | |
| "loss": 1.1457, | |
| "step": 128 | |
| }, | |
| { | |
| "epoch": 0.5726970033296337, | |
| "grad_norm": 1.856616497039795, | |
| "learning_rate": 8.137512670886397e-06, | |
| "loss": 1.139, | |
| "step": 129 | |
| }, | |
| { | |
| "epoch": 0.5771365149833518, | |
| "grad_norm": 1.7110472917556763, | |
| "learning_rate": 7.996122752966596e-06, | |
| "loss": 1.1181, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.58157602663707, | |
| "grad_norm": 1.7752193212509155, | |
| "learning_rate": 7.855148985484946e-06, | |
| "loss": 1.1517, | |
| "step": 131 | |
| }, | |
| { | |
| "epoch": 0.586015538290788, | |
| "grad_norm": 1.789296269416809, | |
| "learning_rate": 7.71462064483311e-06, | |
| "loss": 1.0982, | |
| "step": 132 | |
| }, | |
| { | |
| "epoch": 0.5904550499445061, | |
| "grad_norm": 2.0241036415100098, | |
| "learning_rate": 7.574566914899779e-06, | |
| "loss": 1.1787, | |
| "step": 133 | |
| }, | |
| { | |
| "epoch": 0.5948945615982242, | |
| "grad_norm": 2.217411518096924, | |
| "learning_rate": 7.4350168810099955e-06, | |
| "loss": 1.1871, | |
| "step": 134 | |
| }, | |
| { | |
| "epoch": 0.5993340732519423, | |
| "grad_norm": 1.8291338682174683, | |
| "learning_rate": 7.295999523884921e-06, | |
| "loss": 1.1396, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.6037735849056604, | |
| "grad_norm": 1.818039059638977, | |
| "learning_rate": 7.157543713623353e-06, | |
| "loss": 1.1719, | |
| "step": 136 | |
| }, | |
| { | |
| "epoch": 0.6082130965593785, | |
| "grad_norm": 1.8456182479858398, | |
| "learning_rate": 7.019678203706164e-06, | |
| "loss": 1.1714, | |
| "step": 137 | |
| }, | |
| { | |
| "epoch": 0.6126526082130965, | |
| "grad_norm": 1.9036660194396973, | |
| "learning_rate": 6.882431625025016e-06, | |
| "loss": 1.1439, | |
| "step": 138 | |
| }, | |
| { | |
| "epoch": 0.6170921198668147, | |
| "grad_norm": 1.7687510251998901, | |
| "learning_rate": 6.745832479936492e-06, | |
| "loss": 1.1511, | |
| "step": 139 | |
| }, | |
| { | |
| "epoch": 0.6215316315205327, | |
| "grad_norm": 1.9004747867584229, | |
| "learning_rate": 6.609909136342956e-06, | |
| "loss": 1.172, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.6259711431742508, | |
| "grad_norm": 1.736116886138916, | |
| "learning_rate": 6.474689821801295e-06, | |
| "loss": 1.1422, | |
| "step": 141 | |
| }, | |
| { | |
| "epoch": 0.6304106548279689, | |
| "grad_norm": 2.1278252601623535, | |
| "learning_rate": 6.340202617660842e-06, | |
| "loss": 1.1553, | |
| "step": 142 | |
| }, | |
| { | |
| "epoch": 0.634850166481687, | |
| "grad_norm": 1.7536156177520752, | |
| "learning_rate": 6.206475453231644e-06, | |
| "loss": 1.2252, | |
| "step": 143 | |
| }, | |
| { | |
| "epoch": 0.6392896781354052, | |
| "grad_norm": 1.6779848337173462, | |
| "learning_rate": 6.073536099984314e-06, | |
| "loss": 1.193, | |
| "step": 144 | |
| }, | |
| { | |
| "epoch": 0.6437291897891232, | |
| "grad_norm": 1.8742091655731201, | |
| "learning_rate": 5.941412165782645e-06, | |
| "loss": 1.1921, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.6481687014428413, | |
| "grad_norm": 1.8728663921356201, | |
| "learning_rate": 5.810131089150228e-06, | |
| "loss": 1.1906, | |
| "step": 146 | |
| }, | |
| { | |
| "epoch": 0.6526082130965594, | |
| "grad_norm": 1.6480120420455933, | |
| "learning_rate": 5.6797201335722064e-06, | |
| "loss": 1.1297, | |
| "step": 147 | |
| }, | |
| { | |
| "epoch": 0.6570477247502775, | |
| "grad_norm": 1.6961437463760376, | |
| "learning_rate": 5.550206381833409e-06, | |
| "loss": 1.213, | |
| "step": 148 | |
| }, | |
| { | |
| "epoch": 0.6614872364039955, | |
| "grad_norm": 1.7690813541412354, | |
| "learning_rate": 5.421616730394e-06, | |
| "loss": 1.1643, | |
| "step": 149 | |
| }, | |
| { | |
| "epoch": 0.6659267480577137, | |
| "grad_norm": 1.621023178100586, | |
| "learning_rate": 5.293977883803797e-06, | |
| "loss": 1.1866, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.6703662597114317, | |
| "grad_norm": 1.6755465269088745, | |
| "learning_rate": 5.167316349156495e-06, | |
| "loss": 1.1487, | |
| "step": 151 | |
| }, | |
| { | |
| "epoch": 0.6748057713651499, | |
| "grad_norm": 1.739989995956421, | |
| "learning_rate": 5.041658430584852e-06, | |
| "loss": 1.148, | |
| "step": 152 | |
| }, | |
| { | |
| "epoch": 0.6792452830188679, | |
| "grad_norm": 1.6027244329452515, | |
| "learning_rate": 4.917030223798057e-06, | |
| "loss": 1.1462, | |
| "step": 153 | |
| }, | |
| { | |
| "epoch": 0.683684794672586, | |
| "grad_norm": 1.7501620054244995, | |
| "learning_rate": 4.793457610662334e-06, | |
| "loss": 1.1887, | |
| "step": 154 | |
| }, | |
| { | |
| "epoch": 0.6881243063263041, | |
| "grad_norm": 1.8939120769500732, | |
| "learning_rate": 4.670966253826027e-06, | |
| "loss": 1.0909, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.6925638179800222, | |
| "grad_norm": 1.7964297533035278, | |
| "learning_rate": 4.549581591390142e-06, | |
| "loss": 1.1357, | |
| "step": 156 | |
| }, | |
| { | |
| "epoch": 0.6970033296337403, | |
| "grad_norm": 1.8703395128250122, | |
| "learning_rate": 4.429328831625565e-06, | |
| "loss": 1.1529, | |
| "step": 157 | |
| }, | |
| { | |
| "epoch": 0.7014428412874584, | |
| "grad_norm": 1.676789402961731, | |
| "learning_rate": 4.310232947737979e-06, | |
| "loss": 1.1435, | |
| "step": 158 | |
| }, | |
| { | |
| "epoch": 0.7058823529411765, | |
| "grad_norm": 1.8037596940994263, | |
| "learning_rate": 4.192318672681631e-06, | |
| "loss": 1.1878, | |
| "step": 159 | |
| }, | |
| { | |
| "epoch": 0.7103218645948945, | |
| "grad_norm": 1.6602219343185425, | |
| "learning_rate": 4.0756104940229645e-06, | |
| "loss": 1.1887, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.7147613762486127, | |
| "grad_norm": 1.8436843156814575, | |
| "learning_rate": 3.960132648855226e-06, | |
| "loss": 1.1406, | |
| "step": 161 | |
| }, | |
| { | |
| "epoch": 0.7192008879023307, | |
| "grad_norm": 1.7904314994812012, | |
| "learning_rate": 3.845909118765073e-06, | |
| "loss": 1.149, | |
| "step": 162 | |
| }, | |
| { | |
| "epoch": 0.7236403995560489, | |
| "grad_norm": 1.7600526809692383, | |
| "learning_rate": 3.732963624852275e-06, | |
| "loss": 1.136, | |
| "step": 163 | |
| }, | |
| { | |
| "epoch": 0.7280799112097669, | |
| "grad_norm": 1.912719488143921, | |
| "learning_rate": 3.6213196228034796e-06, | |
| "loss": 1.1052, | |
| "step": 164 | |
| }, | |
| { | |
| "epoch": 0.732519422863485, | |
| "grad_norm": 1.8316577672958374, | |
| "learning_rate": 3.511000298021098e-06, | |
| "loss": 1.115, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.7369589345172031, | |
| "grad_norm": 1.9175002574920654, | |
| "learning_rate": 3.402028560808357e-06, | |
| "loss": 1.1997, | |
| "step": 166 | |
| }, | |
| { | |
| "epoch": 0.7413984461709212, | |
| "grad_norm": 1.7501415014266968, | |
| "learning_rate": 3.2944270416114256e-06, | |
| "loss": 1.1601, | |
| "step": 167 | |
| }, | |
| { | |
| "epoch": 0.7458379578246392, | |
| "grad_norm": 1.8293687105178833, | |
| "learning_rate": 3.1882180863197308e-06, | |
| "loss": 1.131, | |
| "step": 168 | |
| }, | |
| { | |
| "epoch": 0.7502774694783574, | |
| "grad_norm": 1.7824769020080566, | |
| "learning_rate": 3.0834237516252817e-06, | |
| "loss": 1.1655, | |
| "step": 169 | |
| }, | |
| { | |
| "epoch": 0.7547169811320755, | |
| "grad_norm": 1.672102689743042, | |
| "learning_rate": 2.980065800442137e-06, | |
| "loss": 1.1496, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.7591564927857936, | |
| "grad_norm": 1.8063626289367676, | |
| "learning_rate": 2.878165697386812e-06, | |
| "loss": 1.1314, | |
| "step": 171 | |
| }, | |
| { | |
| "epoch": 0.7635960044395117, | |
| "grad_norm": 1.9365928173065186, | |
| "learning_rate": 2.777744604320706e-06, | |
| "loss": 1.2001, | |
| "step": 172 | |
| }, | |
| { | |
| "epoch": 0.7680355160932297, | |
| "grad_norm": 2.0963966846466064, | |
| "learning_rate": 2.678823375955314e-06, | |
| "loss": 1.1451, | |
| "step": 173 | |
| }, | |
| { | |
| "epoch": 0.7724750277469479, | |
| "grad_norm": 1.6925102472305298, | |
| "learning_rate": 2.581422555521316e-06, | |
| "loss": 1.1666, | |
| "step": 174 | |
| }, | |
| { | |
| "epoch": 0.7769145394006659, | |
| "grad_norm": 1.7280200719833374, | |
| "learning_rate": 2.485562370502279e-06, | |
| "loss": 1.1394, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.781354051054384, | |
| "grad_norm": 2.9482946395874023, | |
| "learning_rate": 2.391262728433995e-06, | |
| "loss": 1.1453, | |
| "step": 176 | |
| }, | |
| { | |
| "epoch": 0.7857935627081021, | |
| "grad_norm": 1.7591135501861572, | |
| "learning_rate": 2.2985432127701945e-06, | |
| "loss": 1.1422, | |
| "step": 177 | |
| }, | |
| { | |
| "epoch": 0.7902330743618202, | |
| "grad_norm": 1.9076871871948242, | |
| "learning_rate": 2.2074230788156325e-06, | |
| "loss": 1.133, | |
| "step": 178 | |
| }, | |
| { | |
| "epoch": 0.7946725860155383, | |
| "grad_norm": 1.7988766431808472, | |
| "learning_rate": 2.1179212497272582e-06, | |
| "loss": 1.1535, | |
| "step": 179 | |
| }, | |
| { | |
| "epoch": 0.7991120976692564, | |
| "grad_norm": 1.9537118673324585, | |
| "learning_rate": 2.030056312584424e-06, | |
| "loss": 1.1871, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.8035516093229744, | |
| "grad_norm": 1.6785671710968018, | |
| "learning_rate": 1.9438465145288377e-06, | |
| "loss": 1.0986, | |
| "step": 181 | |
| }, | |
| { | |
| "epoch": 0.8079911209766926, | |
| "grad_norm": 1.8023606538772583, | |
| "learning_rate": 1.8593097589751318e-06, | |
| "loss": 1.118, | |
| "step": 182 | |
| }, | |
| { | |
| "epoch": 0.8124306326304107, | |
| "grad_norm": 1.6982240676879883, | |
| "learning_rate": 1.7764636018928249e-06, | |
| "loss": 1.1521, | |
| "step": 183 | |
| }, | |
| { | |
| "epoch": 0.8168701442841287, | |
| "grad_norm": 1.7196639776229858, | |
| "learning_rate": 1.6953252481604198e-06, | |
| "loss": 1.1243, | |
| "step": 184 | |
| }, | |
| { | |
| "epoch": 0.8213096559378469, | |
| "grad_norm": 1.6617742776870728, | |
| "learning_rate": 1.6159115479924259e-06, | |
| "loss": 1.1496, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.8257491675915649, | |
| "grad_norm": 1.8764078617095947, | |
| "learning_rate": 1.5382389934400199e-06, | |
| "loss": 1.1417, | |
| "step": 186 | |
| }, | |
| { | |
| "epoch": 0.8301886792452831, | |
| "grad_norm": 1.743654489517212, | |
| "learning_rate": 1.462323714966114e-06, | |
| "loss": 1.1322, | |
| "step": 187 | |
| }, | |
| { | |
| "epoch": 0.8346281908990011, | |
| "grad_norm": 1.691676139831543, | |
| "learning_rate": 1.3881814780954916e-06, | |
| "loss": 1.104, | |
| "step": 188 | |
| }, | |
| { | |
| "epoch": 0.8390677025527192, | |
| "grad_norm": 1.6910086870193481, | |
| "learning_rate": 1.3158276801407432e-06, | |
| "loss": 1.181, | |
| "step": 189 | |
| }, | |
| { | |
| "epoch": 0.8435072142064373, | |
| "grad_norm": 1.7811787128448486, | |
| "learning_rate": 1.2452773470046543e-06, | |
| "loss": 1.2015, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.8479467258601554, | |
| "grad_norm": 1.8758302927017212, | |
| "learning_rate": 1.1765451300597574e-06, | |
| "loss": 1.1697, | |
| "step": 191 | |
| }, | |
| { | |
| "epoch": 0.8523862375138734, | |
| "grad_norm": 1.7708605527877808, | |
| "learning_rate": 1.1096453031056265e-06, | |
| "loss": 1.1007, | |
| "step": 192 | |
| }, | |
| { | |
| "epoch": 0.8568257491675916, | |
| "grad_norm": 1.7193588018417358, | |
| "learning_rate": 1.0445917594046073e-06, | |
| "loss": 1.1135, | |
| "step": 193 | |
| }, | |
| { | |
| "epoch": 0.8612652608213096, | |
| "grad_norm": 1.6113449335098267, | |
| "learning_rate": 9.813980087965625e-07, | |
| "loss": 1.1501, | |
| "step": 194 | |
| }, | |
| { | |
| "epoch": 0.8657047724750278, | |
| "grad_norm": 1.726749062538147, | |
| "learning_rate": 9.200771748932513e-07, | |
| "loss": 1.0861, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.8701442841287459, | |
| "grad_norm": 1.7274186611175537, | |
| "learning_rate": 8.606419923529175e-07, | |
| "loss": 1.1256, | |
| "step": 196 | |
| }, | |
| { | |
| "epoch": 0.8745837957824639, | |
| "grad_norm": 1.7909616231918335, | |
| "learning_rate": 8.031048042356393e-07, | |
| "loss": 1.1844, | |
| "step": 197 | |
| }, | |
| { | |
| "epoch": 0.8790233074361821, | |
| "grad_norm": 1.769081473350525, | |
| "learning_rate": 7.474775594400252e-07, | |
| "loss": 1.1142, | |
| "step": 198 | |
| }, | |
| { | |
| "epoch": 0.8834628190899001, | |
| "grad_norm": 1.5884616374969482, | |
| "learning_rate": 6.937718102217461e-07, | |
| "loss": 1.1735, | |
| "step": 199 | |
| }, | |
| { | |
| "epoch": 0.8879023307436182, | |
| "grad_norm": 1.8353629112243652, | |
| "learning_rate": 6.41998709794458e-07, | |
| "loss": 1.1656, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.8923418423973363, | |
| "grad_norm": 1.836683988571167, | |
| "learning_rate": 5.921690100135713e-07, | |
| "loss": 1.1665, | |
| "step": 201 | |
| }, | |
| { | |
| "epoch": 0.8967813540510544, | |
| "grad_norm": 1.7683050632476807, | |
| "learning_rate": 5.442930591433992e-07, | |
| "loss": 1.0818, | |
| "step": 202 | |
| }, | |
| { | |
| "epoch": 0.9012208657047724, | |
| "grad_norm": 1.7694001197814941, | |
| "learning_rate": 4.983807997080925e-07, | |
| "loss": 1.1625, | |
| "step": 203 | |
| }, | |
| { | |
| "epoch": 0.9056603773584906, | |
| "grad_norm": 1.639148473739624, | |
| "learning_rate": 4.544417664268652e-07, | |
| "loss": 1.1001, | |
| "step": 204 | |
| }, | |
| { | |
| "epoch": 0.9100998890122086, | |
| "grad_norm": 1.682405710220337, | |
| "learning_rate": 4.124850842338779e-07, | |
| "loss": 1.1392, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.9145394006659268, | |
| "grad_norm": 1.705399751663208, | |
| "learning_rate": 3.725194663832521e-07, | |
| "loss": 1.1308, | |
| "step": 206 | |
| }, | |
| { | |
| "epoch": 0.9189789123196448, | |
| "grad_norm": 1.77009117603302, | |
| "learning_rate": 3.345532126395579e-07, | |
| "loss": 1.1334, | |
| "step": 207 | |
| }, | |
| { | |
| "epoch": 0.9234184239733629, | |
| "grad_norm": 1.6911001205444336, | |
| "learning_rate": 2.985942075541848e-07, | |
| "loss": 1.1135, | |
| "step": 208 | |
| }, | |
| { | |
| "epoch": 0.9278579356270811, | |
| "grad_norm": 1.6834074258804321, | |
| "learning_rate": 2.646499188279328e-07, | |
| "loss": 1.1604, | |
| "step": 209 | |
| }, | |
| { | |
| "epoch": 0.9322974472807991, | |
| "grad_norm": 1.6770280599594116, | |
| "learning_rate": 2.3272739576017945e-07, | |
| "loss": 1.1296, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.9367369589345172, | |
| "grad_norm": 1.8066668510437012, | |
| "learning_rate": 2.028332677849254e-07, | |
| "loss": 1.1261, | |
| "step": 211 | |
| }, | |
| { | |
| "epoch": 0.9411764705882353, | |
| "grad_norm": 1.8163933753967285, | |
| "learning_rate": 1.7497374309405346e-07, | |
| "loss": 1.174, | |
| "step": 212 | |
| }, | |
| { | |
| "epoch": 0.9456159822419534, | |
| "grad_norm": 1.6583095788955688, | |
| "learning_rate": 1.49154607348051e-07, | |
| "loss": 1.1626, | |
| "step": 213 | |
| }, | |
| { | |
| "epoch": 0.9500554938956715, | |
| "grad_norm": 1.7477093935012817, | |
| "learning_rate": 1.2538122247448325e-07, | |
| "loss": 1.1512, | |
| "step": 214 | |
| }, | |
| { | |
| "epoch": 0.9544950055493896, | |
| "grad_norm": 1.6833035945892334, | |
| "learning_rate": 1.0365852555447642e-07, | |
| "loss": 1.116, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.9589345172031076, | |
| "grad_norm": 1.6046544313430786, | |
| "learning_rate": 8.39910277974132e-08, | |
| "loss": 1.1263, | |
| "step": 216 | |
| }, | |
| { | |
| "epoch": 0.9633740288568258, | |
| "grad_norm": 1.6821755170822144, | |
| "learning_rate": 6.638281360408339e-08, | |
| "loss": 1.1357, | |
| "step": 217 | |
| }, | |
| { | |
| "epoch": 0.9678135405105438, | |
| "grad_norm": 1.7433803081512451, | |
| "learning_rate": 5.083753971846239e-08, | |
| "loss": 1.0744, | |
| "step": 218 | |
| }, | |
| { | |
| "epoch": 0.9722530521642619, | |
| "grad_norm": 1.7408146858215332, | |
| "learning_rate": 3.735843446830867e-08, | |
| "loss": 1.1356, | |
| "step": 219 | |
| }, | |
| { | |
| "epoch": 0.97669256381798, | |
| "grad_norm": 1.7871052026748657, | |
| "learning_rate": 2.5948297094724463e-08, | |
| "loss": 1.1227, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.9811320754716981, | |
| "grad_norm": 1.5904314517974854, | |
| "learning_rate": 1.6609497170834154e-08, | |
| "loss": 1.0993, | |
| "step": 221 | |
| }, | |
| { | |
| "epoch": 0.9855715871254163, | |
| "grad_norm": 1.8538672924041748, | |
| "learning_rate": 9.343974109685684e-09, | |
| "loss": 1.1632, | |
| "step": 222 | |
| }, | |
| { | |
| "epoch": 0.9900110987791343, | |
| "grad_norm": 1.6794443130493164, | |
| "learning_rate": 4.153236761488266e-09, | |
| "loss": 1.0994, | |
| "step": 223 | |
| }, | |
| { | |
| "epoch": 0.9944506104328524, | |
| "grad_norm": 1.6343417167663574, | |
| "learning_rate": 1.0383631002686133e-09, | |
| "loss": 1.1072, | |
| "step": 224 | |
| }, | |
| { | |
| "epoch": 0.9988901220865705, | |
| "grad_norm": 1.9944053888320923, | |
| "learning_rate": 0.0, | |
| "loss": 1.1393, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.9988901220865705, | |
| "step": 225, | |
| "total_flos": 8.515440547840655e+17, | |
| "train_loss": 1.1814873923195732, | |
| "train_runtime": 2241.1241, | |
| "train_samples_per_second": 25.717, | |
| "train_steps_per_second": 0.1 | |
| } | |
| ], | |
| "logging_steps": 1.0, | |
| "max_steps": 225, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 400, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 8.515440547840655e+17, | |
| "train_batch_size": 4, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |