{
  "best_global_step": 3272,
  "best_metric": 0.9839899931917298,
  "best_model_checkpoint": "voice_gender_classification/checkpoint-3272",
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 3272,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003056234718826406,
      "grad_norm": 1.0770392417907715,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.6859,
      "step": 1
    },
    {
      "epoch": 0.003056234718826406,
      "grad_norm": 1.2716485261917114,
      "learning_rate": 6e-06,
      "loss": 0.7011,
      "step": 10
    },
    {
      "epoch": 0.006112469437652812,
      "grad_norm": 1.0392799377441406,
      "learning_rate": 1.2e-05,
      "loss": 0.6944,
      "step": 20
    },
    {
      "epoch": 0.009168704156479218,
      "grad_norm": 0.628126859664917,
      "learning_rate": 1.8e-05,
      "loss": 0.6592,
      "step": 30
    },
    {
      "epoch": 0.012224938875305624,
      "grad_norm": 2.359282970428467,
      "learning_rate": 2.4e-05,
      "loss": 0.6314,
      "step": 40
    },
    {
      "epoch": 0.015281173594132029,
      "grad_norm": 4.273380756378174,
      "learning_rate": 3e-05,
      "loss": 0.5976,
      "step": 50
    },
    {
      "epoch": 0.018337408312958436,
      "grad_norm": 1.5336124897003174,
      "learning_rate": 2.9906890130353816e-05,
      "loss": 0.4168,
      "step": 60
    },
    {
      "epoch": 0.021393643031784843,
      "grad_norm": 1.772170066833496,
      "learning_rate": 2.9823091247672255e-05,
      "loss": 0.3254,
      "step": 70
    },
    {
      "epoch": 0.02444987775061125,
      "grad_norm": 1.347057580947876,
      "learning_rate": 2.973929236499069e-05,
      "loss": 0.1641,
      "step": 80
    },
    {
      "epoch": 0.02750611246943765,
      "grad_norm": 0.8756974339485168,
      "learning_rate": 2.9646182495344507e-05,
      "loss": 0.2623,
      "step": 90
    },
    {
      "epoch": 0.030562347188264057,
      "grad_norm": 0.594032347202301,
      "learning_rate": 2.9553072625698325e-05,
      "loss": 0.0924,
      "step": 100
    },
    {
      "epoch": 0.03361858190709047,
      "grad_norm": 15.337114334106445,
      "learning_rate": 2.946927374301676e-05,
      "loss": 0.4399,
      "step": 110
    },
    {
      "epoch": 0.03667481662591687,
      "grad_norm": 0.4488385319709778,
      "learning_rate": 2.937616387337058e-05,
      "loss": 0.044,
      "step": 120
    },
    {
      "epoch": 0.03973105134474328,
      "grad_norm": 1.3066495656967163,
      "learning_rate": 2.9283054003724395e-05,
      "loss": 0.2036,
      "step": 130
    },
    {
      "epoch": 0.042787286063569685,
      "grad_norm": 0.36801964044570923,
      "learning_rate": 2.919925512104283e-05,
      "loss": 0.1376,
      "step": 140
    },
    {
      "epoch": 0.04584352078239609,
      "grad_norm": 0.35944098234176636,
      "learning_rate": 2.910614525139665e-05,
      "loss": 0.2113,
      "step": 150
    },
    {
      "epoch": 0.0488997555012225,
      "grad_norm": 0.32566192746162415,
      "learning_rate": 2.9013035381750465e-05,
      "loss": 0.1613,
      "step": 160
    },
    {
      "epoch": 0.0519559902200489,
      "grad_norm": 0.3194483816623688,
      "learning_rate": 2.8919925512104284e-05,
      "loss": 0.1142,
      "step": 170
    },
    {
      "epoch": 0.0550122249388753,
      "grad_norm": 0.35201185941696167,
      "learning_rate": 2.88268156424581e-05,
      "loss": 0.2022,
      "step": 180
    },
    {
      "epoch": 0.05806845965770171,
      "grad_norm": 0.3429718017578125,
      "learning_rate": 2.8733705772811922e-05,
      "loss": 0.1527,
      "step": 190
    },
    {
      "epoch": 0.061124694376528114,
      "grad_norm": 0.3786994516849518,
      "learning_rate": 2.8640595903165737e-05,
      "loss": 0.1623,
      "step": 200
    },
    {
      "epoch": 0.06418092909535453,
      "grad_norm": 0.4179794192314148,
      "learning_rate": 2.8547486033519556e-05,
      "loss": 0.1551,
      "step": 210
    },
    {
      "epoch": 0.06723716381418093,
      "grad_norm": 83.91354370117188,
      "learning_rate": 2.845437616387337e-05,
      "loss": 0.2994,
      "step": 220
    },
    {
      "epoch": 0.07029339853300734,
      "grad_norm": 0.2769756615161896,
      "learning_rate": 2.8361266294227187e-05,
      "loss": 0.1539,
      "step": 230
    },
    {
      "epoch": 0.07334963325183375,
      "grad_norm": 0.2955838441848755,
      "learning_rate": 2.8268156424581005e-05,
      "loss": 0.1634,
      "step": 240
    },
    {
      "epoch": 0.07640586797066015,
      "grad_norm": 0.2536284029483795,
      "learning_rate": 2.817504655493482e-05,
      "loss": 0.1137,
      "step": 250
    },
    {
      "epoch": 0.07946210268948656,
      "grad_norm": 1.7733937501907349,
      "learning_rate": 2.8081936685288643e-05,
      "loss": 0.3204,
      "step": 260
    },
    {
      "epoch": 0.08251833740831296,
      "grad_norm": 0.36826714873313904,
      "learning_rate": 2.798882681564246e-05,
      "loss": 0.1624,
      "step": 270
    },
    {
      "epoch": 0.08557457212713937,
      "grad_norm": 0.3160645067691803,
      "learning_rate": 2.7895716945996277e-05,
      "loss": 0.1015,
      "step": 280
    },
    {
      "epoch": 0.08863080684596578,
      "grad_norm": 0.24478046596050262,
      "learning_rate": 2.7802607076350093e-05,
      "loss": 0.1955,
      "step": 290
    },
    {
      "epoch": 0.09168704156479218,
      "grad_norm": 14.753984451293945,
      "learning_rate": 2.770949720670391e-05,
      "loss": 0.1996,
      "step": 300
    },
    {
      "epoch": 0.09474327628361859,
      "grad_norm": 1.5840058326721191,
      "learning_rate": 2.7616387337057727e-05,
      "loss": 0.1386,
      "step": 310
    },
    {
      "epoch": 0.097799511002445,
      "grad_norm": 0.23863820731639862,
      "learning_rate": 2.752327746741155e-05,
      "loss": 0.2088,
      "step": 320
    },
    {
      "epoch": 0.1008557457212714,
      "grad_norm": 1.5646458864212036,
      "learning_rate": 2.7430167597765364e-05,
      "loss": 0.16,
      "step": 330
    },
    {
      "epoch": 0.1039119804400978,
      "grad_norm": 0.25270429253578186,
      "learning_rate": 2.7337057728119183e-05,
      "loss": 0.0747,
      "step": 340
    },
    {
      "epoch": 0.1069682151589242,
      "grad_norm": 1.4299941062927246,
      "learning_rate": 2.7243947858473e-05,
      "loss": 0.1075,
      "step": 350
    },
    {
      "epoch": 0.1100244498777506,
      "grad_norm": 20.335739135742188,
      "learning_rate": 2.7150837988826817e-05,
      "loss": 0.2605,
      "step": 360
    },
    {
      "epoch": 0.11308068459657701,
      "grad_norm": 0.2346445471048355,
      "learning_rate": 2.7057728119180633e-05,
      "loss": 0.0954,
      "step": 370
    },
    {
      "epoch": 0.11613691931540342,
      "grad_norm": 0.23116770386695862,
      "learning_rate": 2.696461824953445e-05,
      "loss": 0.0641,
      "step": 380
    },
    {
      "epoch": 0.11919315403422982,
      "grad_norm": 1.6760469675064087,
      "learning_rate": 2.687150837988827e-05,
      "loss": 0.0678,
      "step": 390
    },
    {
      "epoch": 0.12224938875305623,
      "grad_norm": 0.1794356405735016,
      "learning_rate": 2.6778398510242085e-05,
      "loss": 0.0609,
      "step": 400
    },
    {
      "epoch": 0.12530562347188265,
      "grad_norm": 0.18168634176254272,
      "learning_rate": 2.6685288640595904e-05,
      "loss": 0.0611,
      "step": 410
    },
    {
      "epoch": 0.12836185819070906,
      "grad_norm": 0.2647036612033844,
      "learning_rate": 2.659217877094972e-05,
      "loss": 0.0598,
      "step": 420
    },
    {
      "epoch": 0.13141809290953546,
      "grad_norm": 0.2929624915122986,
      "learning_rate": 2.649906890130354e-05,
      "loss": 0.1172,
      "step": 430
    },
    {
      "epoch": 0.13447432762836187,
      "grad_norm": 0.7274528741836548,
      "learning_rate": 2.6405959031657354e-05,
      "loss": 0.1237,
      "step": 440
    },
    {
      "epoch": 0.13753056234718827,
      "grad_norm": 1.7421795129776,
      "learning_rate": 2.6312849162011176e-05,
      "loss": 0.2341,
      "step": 450
    },
    {
      "epoch": 0.14058679706601468,
      "grad_norm": 0.6214892268180847,
      "learning_rate": 2.621973929236499e-05,
      "loss": 0.1648,
      "step": 460
    },
    {
      "epoch": 0.14364303178484109,
      "grad_norm": 0.21029511094093323,
      "learning_rate": 2.612662942271881e-05,
      "loss": 0.0573,
      "step": 470
    },
    {
      "epoch": 0.1466992665036675,
      "grad_norm": 0.181521475315094,
      "learning_rate": 2.6033519553072625e-05,
      "loss": 0.2243,
      "step": 480
    },
    {
      "epoch": 0.1497555012224939,
      "grad_norm": 0.2883399724960327,
      "learning_rate": 2.5940409683426444e-05,
      "loss": 0.1662,
      "step": 490
    },
    {
      "epoch": 0.1528117359413203,
      "grad_norm": 0.1559794545173645,
      "learning_rate": 2.584729981378026e-05,
      "loss": 0.1764,
      "step": 500
    },
    {
      "epoch": 0.1558679706601467,
      "grad_norm": 0.4264441728591919,
      "learning_rate": 2.575418994413408e-05,
      "loss": 0.0789,
      "step": 510
    },
    {
      "epoch": 0.15892420537897312,
      "grad_norm": 0.320130318403244,
      "learning_rate": 2.5661080074487897e-05,
      "loss": 0.1565,
      "step": 520
    },
    {
      "epoch": 0.16198044009779952,
      "grad_norm": 6.056375503540039,
      "learning_rate": 2.5567970204841716e-05,
      "loss": 0.2424,
      "step": 530
    },
    {
      "epoch": 0.16503667481662593,
      "grad_norm": 4.100960731506348,
      "learning_rate": 2.547486033519553e-05,
      "loss": 0.0672,
      "step": 540
    },
    {
      "epoch": 0.16809290953545233,
      "grad_norm": 4.6566901206970215,
      "learning_rate": 2.538175046554935e-05,
      "loss": 0.1897,
      "step": 550
    },
    {
      "epoch": 0.17114914425427874,
      "grad_norm": 3.481842279434204,
      "learning_rate": 2.5288640595903165e-05,
      "loss": 0.1389,
      "step": 560
    },
    {
      "epoch": 0.17420537897310515,
      "grad_norm": 0.45788800716400146,
      "learning_rate": 2.519553072625698e-05,
      "loss": 0.0967,
      "step": 570
    },
    {
      "epoch": 0.17726161369193155,
      "grad_norm": 0.1905224472284317,
      "learning_rate": 2.5102420856610803e-05,
      "loss": 0.0771,
      "step": 580
    },
    {
      "epoch": 0.18031784841075796,
      "grad_norm": 0.1771785467863083,
      "learning_rate": 2.5009310986964618e-05,
      "loss": 0.1548,
      "step": 590
    },
    {
      "epoch": 0.18337408312958436,
      "grad_norm": 0.18358071148395538,
      "learning_rate": 2.4916201117318437e-05,
      "loss": 0.2571,
      "step": 600
    },
    {
      "epoch": 0.18643031784841077,
      "grad_norm": 0.3116374611854553,
      "learning_rate": 2.4823091247672252e-05,
      "loss": 0.1225,
      "step": 610
    },
    {
      "epoch": 0.18948655256723718,
      "grad_norm": 0.2802579998970032,
      "learning_rate": 2.472998137802607e-05,
      "loss": 0.0988,
      "step": 620
    },
    {
      "epoch": 0.19254278728606358,
      "grad_norm": 43.23971176147461,
      "learning_rate": 2.4636871508379887e-05,
      "loss": 0.2569,
      "step": 630
    },
    {
      "epoch": 0.19559902200489,
      "grad_norm": 0.25425320863723755,
      "learning_rate": 2.454376163873371e-05,
      "loss": 0.0759,
      "step": 640
    },
    {
      "epoch": 0.1986552567237164,
      "grad_norm": 0.16143210232257843,
      "learning_rate": 2.4450651769087524e-05,
      "loss": 0.0666,
      "step": 650
    },
    {
      "epoch": 0.2017114914425428,
      "grad_norm": 0.15112750232219696,
      "learning_rate": 2.4357541899441343e-05,
      "loss": 0.1222,
      "step": 660
    },
    {
      "epoch": 0.2047677261613692,
      "grad_norm": 0.1990450918674469,
      "learning_rate": 2.4264432029795158e-05,
      "loss": 0.1612,
      "step": 670
    },
    {
      "epoch": 0.2078239608801956,
      "grad_norm": 0.23198798298835754,
      "learning_rate": 2.4171322160148977e-05,
      "loss": 0.0139,
      "step": 680
    },
    {
      "epoch": 0.210880195599022,
      "grad_norm": 0.2512992024421692,
      "learning_rate": 2.4078212290502792e-05,
      "loss": 0.2073,
      "step": 690
    },
    {
      "epoch": 0.2139364303178484,
      "grad_norm": 0.29186299443244934,
      "learning_rate": 2.398510242085661e-05,
      "loss": 0.0141,
      "step": 700
    },
    {
      "epoch": 0.2169926650366748,
      "grad_norm": 0.20607629418373108,
      "learning_rate": 2.389199255121043e-05,
      "loss": 0.1886,
      "step": 710
    },
    {
      "epoch": 0.2200488997555012,
      "grad_norm": 0.40326663851737976,
      "learning_rate": 2.379888268156425e-05,
      "loss": 0.082,
      "step": 720
    },
    {
      "epoch": 0.22310513447432762,
      "grad_norm": 0.18325626850128174,
      "learning_rate": 2.3705772811918064e-05,
      "loss": 0.0628,
      "step": 730
    },
    {
      "epoch": 0.22616136919315402,
      "grad_norm": 0.1717202067375183,
      "learning_rate": 2.361266294227188e-05,
      "loss": 0.1665,
      "step": 740
    },
    {
      "epoch": 0.22921760391198043,
      "grad_norm": 1.470982313156128,
      "learning_rate": 2.3519553072625698e-05,
      "loss": 0.1568,
      "step": 750
    },
    {
      "epoch": 0.23227383863080683,
      "grad_norm": 0.1533995270729065,
      "learning_rate": 2.3426443202979514e-05,
      "loss": 0.0565,
      "step": 760
    },
    {
      "epoch": 0.23533007334963324,
      "grad_norm": 0.18549972772598267,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 0.1877,
      "step": 770
    },
    {
      "epoch": 0.23838630806845965,
      "grad_norm": 1.3424246311187744,
      "learning_rate": 2.324022346368715e-05,
      "loss": 0.1371,
      "step": 780
    },
    {
      "epoch": 0.24144254278728605,
      "grad_norm": 1.0893691778182983,
      "learning_rate": 2.314711359404097e-05,
      "loss": 0.1286,
      "step": 790
    },
    {
      "epoch": 0.24449877750611246,
      "grad_norm": 0.309053897857666,
      "learning_rate": 2.3054003724394785e-05,
      "loss": 0.0637,
      "step": 800
    },
    {
      "epoch": 0.24755501222493886,
      "grad_norm": 0.2743883728981018,
      "learning_rate": 2.2960893854748604e-05,
      "loss": 0.0151,
      "step": 810
    },
    {
      "epoch": 0.2506112469437653,
      "grad_norm": 1.4966446161270142,
      "learning_rate": 2.286778398510242e-05,
      "loss": 0.1031,
      "step": 820
    },
    {
      "epoch": 0.2536674816625917,
      "grad_norm": 0.22224830090999603,
      "learning_rate": 2.2774674115456238e-05,
      "loss": 0.0618,
      "step": 830
    },
    {
      "epoch": 0.2567237163814181,
      "grad_norm": 0.17919765412807465,
      "learning_rate": 2.2681564245810057e-05,
      "loss": 0.0593,
      "step": 840
    },
    {
      "epoch": 0.2597799511002445,
      "grad_norm": 0.09819658100605011,
      "learning_rate": 2.2588454376163876e-05,
      "loss": 0.0781,
      "step": 850
    },
    {
      "epoch": 0.2628361858190709,
      "grad_norm": 0.22454911470413208,
      "learning_rate": 2.249534450651769e-05,
      "loss": 0.1045,
      "step": 860
    },
    {
      "epoch": 0.2658924205378973,
      "grad_norm": 49.3597526550293,
      "learning_rate": 2.240223463687151e-05,
      "loss": 0.087,
      "step": 870
    },
    {
      "epoch": 0.26894865525672373,
      "grad_norm": 1.6358880996704102,
      "learning_rate": 2.2309124767225325e-05,
      "loss": 0.2507,
      "step": 880
    },
    {
      "epoch": 0.2720048899755501,
      "grad_norm": 0.15142078697681427,
      "learning_rate": 2.2216014897579144e-05,
      "loss": 0.1193,
      "step": 890
    },
    {
      "epoch": 0.27506112469437655,
      "grad_norm": 0.1405174732208252,
      "learning_rate": 2.2122905027932963e-05,
      "loss": 0.0527,
      "step": 900
    },
    {
      "epoch": 0.2781173594132029,
      "grad_norm": 0.1312502920627594,
      "learning_rate": 2.2029795158286778e-05,
      "loss": 0.1309,
      "step": 910
    },
    {
      "epoch": 0.28117359413202936,
      "grad_norm": 0.11085129529237747,
      "learning_rate": 2.1936685288640597e-05,
      "loss": 0.0598,
      "step": 920
    },
    {
      "epoch": 0.28422982885085574,
      "grad_norm": 0.13175728917121887,
      "learning_rate": 2.1843575418994412e-05,
      "loss": 0.0601,
      "step": 930
    },
    {
      "epoch": 0.28728606356968217,
      "grad_norm": 0.13237424194812775,
      "learning_rate": 2.175046554934823e-05,
      "loss": 0.1338,
      "step": 940
    },
    {
      "epoch": 0.29034229828850855,
      "grad_norm": 1.7460986375808716,
      "learning_rate": 2.1657355679702047e-05,
      "loss": 0.1575,
      "step": 950
    },
    {
      "epoch": 0.293398533007335,
      "grad_norm": 0.1591300666332245,
      "learning_rate": 2.156424581005587e-05,
      "loss": 0.0535,
      "step": 960
    },
    {
      "epoch": 0.29645476772616136,
      "grad_norm": 0.1477421522140503,
      "learning_rate": 2.1471135940409684e-05,
      "loss": 0.0701,
      "step": 970
    },
    {
      "epoch": 0.2995110024449878,
      "grad_norm": 0.15832221508026123,
      "learning_rate": 2.1378026070763503e-05,
      "loss": 0.2401,
      "step": 980
    },
    {
      "epoch": 0.3025672371638142,
      "grad_norm": 0.2130369395017624,
      "learning_rate": 2.1284916201117318e-05,
      "loss": 0.2158,
      "step": 990
    },
    {
      "epoch": 0.3056234718826406,
      "grad_norm": 0.1542048305273056,
      "learning_rate": 2.1191806331471137e-05,
      "loss": 0.0971,
      "step": 1000
    },
    {
      "epoch": 0.308679706601467,
      "grad_norm": 3.364166021347046,
      "learning_rate": 2.1098696461824952e-05,
      "loss": 0.1079,
      "step": 1010
    },
    {
      "epoch": 0.3117359413202934,
      "grad_norm": 0.14105622470378876,
      "learning_rate": 2.100558659217877e-05,
      "loss": 0.0597,
      "step": 1020
    },
    {
      "epoch": 0.3147921760391198,
      "grad_norm": 0.1316232830286026,
      "learning_rate": 2.091247672253259e-05,
      "loss": 0.1159,
      "step": 1030
    },
    {
      "epoch": 0.31784841075794623,
      "grad_norm": 0.48321789503097534,
      "learning_rate": 2.081936685288641e-05,
      "loss": 0.078,
      "step": 1040
    },
    {
      "epoch": 0.3209046454767726,
      "grad_norm": 0.1567317247390747,
      "learning_rate": 2.0726256983240224e-05,
      "loss": 0.204,
      "step": 1050
    },
    {
      "epoch": 0.32396088019559904,
      "grad_norm": 6.398591995239258,
      "learning_rate": 2.0633147113594043e-05,
      "loss": 0.3119,
      "step": 1060
    },
    {
      "epoch": 0.3270171149144254,
      "grad_norm": 0.20879434049129486,
      "learning_rate": 2.0540037243947858e-05,
      "loss": 0.1167,
      "step": 1070
    },
    {
      "epoch": 0.33007334963325186,
      "grad_norm": 0.23976916074752808,
      "learning_rate": 2.0446927374301674e-05,
      "loss": 0.1077,
      "step": 1080
    },
    {
      "epoch": 0.33312958435207823,
      "grad_norm": 0.17524395883083344,
      "learning_rate": 2.0353817504655496e-05,
      "loss": 0.0814,
      "step": 1090
    },
    {
      "epoch": 0.33618581907090467,
      "grad_norm": 0.36440449953079224,
      "learning_rate": 2.026070763500931e-05,
      "loss": 0.0589,
      "step": 1100
    },
    {
      "epoch": 0.33924205378973105,
      "grad_norm": 0.15378913283348083,
      "learning_rate": 2.016759776536313e-05,
      "loss": 0.0112,
      "step": 1110
    },
    {
      "epoch": 0.3422982885085575,
      "grad_norm": 0.14528077840805054,
      "learning_rate": 2.0074487895716945e-05,
      "loss": 0.1187,
      "step": 1120
    },
    {
      "epoch": 0.34535452322738386,
      "grad_norm": 0.12367907911539078,
      "learning_rate": 1.9990689013035385e-05,
      "loss": 0.0722,
      "step": 1130
    },
    {
      "epoch": 0.3484107579462103,
      "grad_norm": 0.09661256521940231,
      "learning_rate": 1.98975791433892e-05,
      "loss": 0.0052,
      "step": 1140
    },
    {
      "epoch": 0.35146699266503667,
      "grad_norm": 0.7434007525444031,
      "learning_rate": 1.980446927374302e-05,
      "loss": 0.005,
      "step": 1150
    },
    {
      "epoch": 0.3545232273838631,
      "grad_norm": 0.04170653969049454,
      "learning_rate": 1.9711359404096834e-05,
      "loss": 0.0036,
      "step": 1160
    },
    {
      "epoch": 0.3575794621026895,
      "grad_norm": 0.07466982305049896,
      "learning_rate": 1.961824953445065e-05,
      "loss": 0.0684,
      "step": 1170
    },
    {
      "epoch": 0.3606356968215159,
      "grad_norm": 0.05339043587446213,
      "learning_rate": 1.952513966480447e-05,
      "loss": 0.0064,
      "step": 1180
    },
    {
      "epoch": 0.3636919315403423,
      "grad_norm": 4.3722052574157715,
      "learning_rate": 1.9432029795158287e-05,
      "loss": 0.295,
      "step": 1190
    },
    {
      "epoch": 0.36674816625916873,
      "grad_norm": 0.04944440722465515,
      "learning_rate": 1.9338919925512106e-05,
      "loss": 0.0644,
      "step": 1200
    },
    {
      "epoch": 0.3698044009779951,
      "grad_norm": 0.09254731982946396,
      "learning_rate": 1.924581005586592e-05,
      "loss": 0.2032,
      "step": 1210
    },
    {
      "epoch": 0.37286063569682154,
      "grad_norm": 2.1284565925598145,
      "learning_rate": 1.915270018621974e-05,
      "loss": 0.1224,
      "step": 1220
    },
    {
      "epoch": 0.3759168704156479,
      "grad_norm": 0.1618586629629135,
      "learning_rate": 1.9059590316573555e-05,
      "loss": 0.0694,
      "step": 1230
    },
    {
      "epoch": 0.37897310513447435,
      "grad_norm": 0.20215436816215515,
      "learning_rate": 1.8966480446927374e-05,
      "loss": 0.0642,
      "step": 1240
    },
    {
      "epoch": 0.38202933985330073,
      "grad_norm": 0.11647956073284149,
      "learning_rate": 1.8873370577281193e-05,
      "loss": 0.0268,
      "step": 1250
    },
    {
      "epoch": 0.38508557457212717,
      "grad_norm": 2.9735450744628906,
      "learning_rate": 1.878026070763501e-05,
      "loss": 0.1907,
      "step": 1260
    },
    {
      "epoch": 0.38814180929095354,
      "grad_norm": 0.16742634773254395,
      "learning_rate": 1.8687150837988827e-05,
      "loss": 0.068,
      "step": 1270
    },
    {
      "epoch": 0.39119804400978,
      "grad_norm": 0.1201523020863533,
      "learning_rate": 1.8594040968342646e-05,
      "loss": 0.0627,
      "step": 1280
    },
    {
      "epoch": 0.39425427872860636,
      "grad_norm": 0.07464749366044998,
      "learning_rate": 1.850093109869646e-05,
      "loss": 0.0132,
      "step": 1290
    },
    {
      "epoch": 0.3973105134474328,
      "grad_norm": 0.12164530158042908,
      "learning_rate": 1.840782122905028e-05,
      "loss": 0.2341,
      "step": 1300
    },
    {
      "epoch": 0.40036674816625917,
      "grad_norm": 0.09393124282360077,
      "learning_rate": 1.83147113594041e-05,
      "loss": 0.061,
      "step": 1310
    },
    {
      "epoch": 0.4034229828850856,
      "grad_norm": 8.743972778320312,
      "learning_rate": 1.8221601489757917e-05,
      "loss": 0.1509,
      "step": 1320
    },
    {
      "epoch": 0.406479217603912,
      "grad_norm": 0.22810806334018707,
      "learning_rate": 1.8128491620111733e-05,
      "loss": 0.0187,
      "step": 1330
    },
    {
      "epoch": 0.4095354523227384,
      "grad_norm": 0.15852580964565277,
      "learning_rate": 1.8035381750465548e-05,
      "loss": 0.051,
      "step": 1340
    },
    {
      "epoch": 0.4125916870415648,
      "grad_norm": 2.218247652053833,
      "learning_rate": 1.7942271880819367e-05,
      "loss": 0.151,
      "step": 1350
    },
    {
      "epoch": 0.4156479217603912,
      "grad_norm": 0.0791810154914856,
      "learning_rate": 1.7849162011173182e-05,
      "loss": 0.352,
      "step": 1360
    },
    {
      "epoch": 0.4187041564792176,
      "grad_norm": 0.2023414820432663,
      "learning_rate": 1.7756052141527e-05,
      "loss": 0.2002,
      "step": 1370
    },
    {
      "epoch": 0.421760391198044,
      "grad_norm": 4.077162265777588,
      "learning_rate": 1.766294227188082e-05,
      "loss": 0.0924,
      "step": 1380
    },
    {
      "epoch": 0.4248166259168704,
      "grad_norm": 0.16036756336688995,
      "learning_rate": 1.756983240223464e-05,
      "loss": 0.1015,
      "step": 1390
    },
    {
      "epoch": 0.4278728606356968,
      "grad_norm": 12.642723083496094,
      "learning_rate": 1.7476722532588454e-05,
      "loss": 0.0885,
      "step": 1400
    },
    {
      "epoch": 0.43092909535452323,
      "grad_norm": 0.20661431550979614,
      "learning_rate": 1.7383612662942273e-05,
      "loss": 0.1037,
      "step": 1410
    },
    {
      "epoch": 0.4339853300733496,
      "grad_norm": 0.14822793006896973,
      "learning_rate": 1.7290502793296088e-05,
      "loss": 0.0837,
      "step": 1420
    },
    {
      "epoch": 0.43704156479217604,
      "grad_norm": 0.12976513803005219,
      "learning_rate": 1.7197392923649907e-05,
      "loss": 0.2077,
      "step": 1430
    },
    {
      "epoch": 0.4400977995110024,
      "grad_norm": 0.045029882341623306,
      "learning_rate": 1.7104283054003726e-05,
      "loss": 0.0824,
      "step": 1440
    },
    {
      "epoch": 0.44315403422982885,
      "grad_norm": 0.19809339940547943,
      "learning_rate": 1.7011173184357544e-05,
      "loss": 0.1054,
      "step": 1450
    },
    {
      "epoch": 0.44621026894865523,
      "grad_norm": 0.2574075162410736,
      "learning_rate": 1.691806331471136e-05,
      "loss": 0.0878,
      "step": 1460
    },
    {
      "epoch": 0.44926650366748166,
      "grad_norm": 1.8627293109893799,
      "learning_rate": 1.682495344506518e-05,
      "loss": 0.1102,
      "step": 1470
    },
    {
      "epoch": 0.45232273838630804,
      "grad_norm": 0.1813037395477295,
      "learning_rate": 1.6731843575418994e-05,
      "loss": 0.1346,
      "step": 1480
    },
    {
      "epoch": 0.4553789731051345,
      "grad_norm": 0.21230725944042206,
      "learning_rate": 1.6638733705772813e-05,
      "loss": 0.0377,
      "step": 1490
    },
    {
      "epoch": 0.45843520782396086,
      "grad_norm": 0.17976613342761993,
      "learning_rate": 1.654562383612663e-05,
      "loss": 0.1334,
      "step": 1500
    },
    {
      "epoch": 0.4614914425427873,
      "grad_norm": 0.13542696833610535,
      "learning_rate": 1.6452513966480447e-05,
      "loss": 0.0935,
      "step": 1510
    },
    {
      "epoch": 0.46454767726161367,
      "grad_norm": 0.1337755024433136,
      "learning_rate": 1.6359404096834266e-05,
      "loss": 0.0073,
      "step": 1520
    },
    {
      "epoch": 0.4676039119804401,
      "grad_norm": 0.052432287484407425,
      "learning_rate": 1.626629422718808e-05,
      "loss": 0.1279,
      "step": 1530
    },
    {
      "epoch": 0.4706601466992665,
      "grad_norm": 0.06683123856782913,
      "learning_rate": 1.61731843575419e-05,
      "loss": 0.0042,
      "step": 1540
    },
    {
      "epoch": 0.4737163814180929,
      "grad_norm": 0.09111201763153076,
      "learning_rate": 1.6080074487895715e-05,
      "loss": 0.0046,
      "step": 1550
    },
    {
      "epoch": 0.4767726161369193,
      "grad_norm": 0.053649771958589554,
      "learning_rate": 1.5986964618249534e-05,
      "loss": 0.1305,
      "step": 1560
    },
    {
      "epoch": 0.4798288508557457,
      "grad_norm": 0.10080774873495102,
      "learning_rate": 1.5893854748603353e-05,
      "loss": 0.1217,
      "step": 1570
    },
    {
      "epoch": 0.4828850855745721,
      "grad_norm": 34.571510314941406,
      "learning_rate": 1.580074487895717e-05,
      "loss": 0.036,
      "step": 1580
    },
    {
      "epoch": 0.48594132029339854,
      "grad_norm": 0.07344861328601837,
      "learning_rate": 1.5707635009310987e-05,
      "loss": 0.1111,
      "step": 1590
    },
    {
      "epoch": 0.4889975550122249,
      "grad_norm": 4.016622543334961,
      "learning_rate": 1.5614525139664806e-05,
      "loss": 0.1141,
      "step": 1600
    },
    {
      "epoch": 0.49205378973105135,
      "grad_norm": 0.12781716883182526,
      "learning_rate": 1.552141527001862e-05,
      "loss": 0.1436,
      "step": 1610
    },
    {
      "epoch": 0.49511002444987773,
      "grad_norm": 0.28127339482307434,
      "learning_rate": 1.542830540037244e-05,
      "loss": 0.0141,
      "step": 1620
    },
    {
      "epoch": 0.49816625916870416,
      "grad_norm": 110.2169418334961,
      "learning_rate": 1.533519553072626e-05,
      "loss": 0.1011,
      "step": 1630
    },
    {
      "epoch": 0.5012224938875306,
      "grad_norm": 0.11259926855564117,
      "learning_rate": 1.5242085661080076e-05,
      "loss": 0.1317,
      "step": 1640
    },
    {
      "epoch": 0.5042787286063569,
      "grad_norm": 54.546661376953125,
      "learning_rate": 1.5148975791433893e-05,
      "loss": 0.1059,
      "step": 1650
    },
    {
      "epoch": 0.5073349633251834,
      "grad_norm": 0.039869241416454315,
      "learning_rate": 1.5055865921787711e-05,
      "loss": 0.0576,
      "step": 1660
    },
    {
      "epoch": 0.5103911980440098,
      "grad_norm": 0.09830798953771591,
      "learning_rate": 1.4962756052141527e-05,
      "loss": 0.0096,
      "step": 1670
    },
    {
      "epoch": 0.5134474327628362,
      "grad_norm": 0.07597620040178299,
      "learning_rate": 1.4869646182495346e-05,
      "loss": 0.051,
      "step": 1680
    },
    {
      "epoch": 0.5165036674816625,
      "grad_norm": 0.05168398097157478,
      "learning_rate": 1.4776536312849163e-05,
      "loss": 0.1166,
      "step": 1690
    },
    {
      "epoch": 0.519559902200489,
      "grad_norm": 93.93925476074219,
      "learning_rate": 1.468342644320298e-05,
      "loss": 0.0429,
      "step": 1700
    },
    {
      "epoch": 0.5226161369193154,
      "grad_norm": 0.9389681220054626,
      "learning_rate": 1.4590316573556799e-05,
      "loss": 0.0065,
      "step": 1710
    },
    {
      "epoch": 0.5256723716381418,
      "grad_norm": 0.09277983009815216,
      "learning_rate": 1.4497206703910616e-05,
      "loss": 0.0635,
      "step": 1720
    },
    {
      "epoch": 0.5287286063569682,
      "grad_norm": 0.09135166555643082,
      "learning_rate": 1.4404096834264433e-05,
      "loss": 0.1598,
      "step": 1730
    },
    {
      "epoch": 0.5317848410757946,
      "grad_norm": 0.07920227199792862,
      "learning_rate": 1.431098696461825e-05,
      "loss": 0.0586,
      "step": 1740
    },
    {
      "epoch": 0.534841075794621,
      "grad_norm": 0.0759720727801323,
      "learning_rate": 1.4217877094972069e-05,
      "loss": 0.1053,
      "step": 1750
    },
    {
      "epoch": 0.5378973105134475,
      "grad_norm": 0.07295390963554382,
      "learning_rate": 1.4124767225325884e-05,
      "loss": 0.0636,
      "step": 1760
    },
    {
      "epoch": 0.5409535452322738,
      "grad_norm": 0.1437848061323166,
      "learning_rate": 1.4031657355679701e-05,
      "loss": 0.1748,
      "step": 1770
    },
    {
      "epoch": 0.5440097799511002,
      "grad_norm": 0.0709507092833519,
      "learning_rate": 1.393854748603352e-05,
      "loss": 0.0113,
      "step": 1780
    },
    {
      "epoch": 0.5470660146699267,
      "grad_norm": 0.06068747863173485,
      "learning_rate": 1.3845437616387337e-05,
      "loss": 0.1883,
      "step": 1790
    },
    {
      "epoch": 0.5501222493887531,
      "grad_norm": 0.2686610221862793,
      "learning_rate": 1.3752327746741154e-05,
      "loss": 0.1691,
      "step": 1800
    },
    {
      "epoch": 0.5531784841075794,
      "grad_norm": 0.09411593526601791,
      "learning_rate": 1.3659217877094973e-05,
      "loss": 0.0983,
      "step": 1810
    },
    {
      "epoch": 0.5562347188264058,
      "grad_norm": 0.22031310200691223,
      "learning_rate": 1.356610800744879e-05,
      "loss": 0.0591,
      "step": 1820
    },
    {
      "epoch": 0.5592909535452323,
      "grad_norm": 0.09072048962116241,
      "learning_rate": 1.3472998137802607e-05,
      "loss": 0.0111,
      "step": 1830
    },
    {
      "epoch": 0.5623471882640587,
      "grad_norm": 0.10641049593687057,
      "learning_rate": 1.3379888268156426e-05,
      "loss": 0.1188,
      "step": 1840
    },
    {
      "epoch": 0.565403422982885,
      "grad_norm": 0.07925461232662201,
      "learning_rate": 1.3286778398510243e-05,
      "loss": 0.0603,
      "step": 1850
    },
    {
      "epoch": 0.5684596577017115,
      "grad_norm": 0.16015411913394928,
      "learning_rate": 1.319366852886406e-05,
      "loss": 0.062,
      "step": 1860
    },
    {
      "epoch": 0.5715158924205379,
      "grad_norm": 1.8697471618652344,
      "learning_rate": 1.3100558659217879e-05,
      "loss": 0.1608,
      "step": 1870
    },
    {
      "epoch": 0.5745721271393643,
      "grad_norm": 0.1413722038269043,
      "learning_rate": 1.3007448789571696e-05,
      "loss": 0.061,
      "step": 1880
    },
    {
      "epoch": 0.5776283618581907,
      "grad_norm": 0.1544978767633438,
      "learning_rate": 1.2914338919925513e-05,
      "loss": 0.0677,
      "step": 1890
    },
    {
      "epoch": 0.5806845965770171,
      "grad_norm": 0.1221311092376709,
      "learning_rate": 1.282122905027933e-05,
      "loss": 0.1088,
      "step": 1900
    },
    {
      "epoch": 0.5837408312958435,
      "grad_norm": 1.616721272468567,
      "learning_rate": 1.2728119180633149e-05,
      "loss": 0.1209,
      "step": 1910
    },
    {
      "epoch": 0.58679706601467,
      "grad_norm": 6.469699382781982,
      "learning_rate": 1.2635009310986966e-05,
      "loss": 0.1047,
      "step": 1920
    },
    {
      "epoch": 0.5898533007334963,
      "grad_norm": 0.13257375359535217,
      "learning_rate": 1.2541899441340781e-05,
      "loss": 0.1008,
      "step": 1930
    },
    {
      "epoch": 0.5929095354523227,
      "grad_norm": 5.230037689208984,
      "learning_rate": 1.24487895716946e-05,
      "loss": 0.1843,
      "step": 1940
    },
    {
      "epoch": 0.5959657701711492,
      "grad_norm": 5.640364170074463,
      "learning_rate": 1.2355679702048417e-05,
      "loss": 0.1216,
      "step": 1950
    },
    {
      "epoch": 0.5990220048899756,
      "grad_norm": 0.05993470177054405,
      "learning_rate": 1.2262569832402234e-05,
      "loss": 0.0991,
      "step": 1960
    },
    {
      "epoch": 0.6020782396088019,
      "grad_norm": 0.12238603085279465,
      "learning_rate": 1.2169459962756053e-05,
      "loss": 0.0109,
      "step": 1970
    },
    {
      "epoch": 0.6051344743276283,
      "grad_norm": 0.16531647741794586,
      "learning_rate": 1.207635009310987e-05,
      "loss": 0.0523,
      "step": 1980
    },
    {
      "epoch": 0.6081907090464548,
      "grad_norm": 0.07218731194734573,
      "learning_rate": 1.1983240223463687e-05,
      "loss": 0.0054,
      "step": 1990
    },
    {
      "epoch": 0.6112469437652812,
      "grad_norm": 0.04113167151808739,
      "learning_rate": 1.1890130353817506e-05,
      "loss": 0.0674,
      "step": 2000
    },
    {
      "epoch": 0.6143031784841075,
      "grad_norm": 13.622767448425293,
      "learning_rate": 1.1797020484171323e-05,
      "loss": 0.1255,
      "step": 2010
    },
    {
      "epoch": 0.617359413202934,
      "grad_norm": 0.06221568211913109,
      "learning_rate": 1.170391061452514e-05,
      "loss": 0.0043,
      "step": 2020
    },
    {
      "epoch": 0.6204156479217604,
      "grad_norm": 0.04879147931933403,
      "learning_rate": 1.1610800744878958e-05,
      "loss": 0.0603,
      "step": 2030
    },
    {
      "epoch": 0.6234718826405868,
      "grad_norm": 0.09293865412473679,
      "learning_rate": 1.1517690875232776e-05,
      "loss": 0.1467,
      "step": 2040
    },
    {
      "epoch": 0.6265281173594132,
      "grad_norm": 0.10396531224250793,
      "learning_rate": 1.1424581005586593e-05,
      "loss": 0.0688,
      "step": 2050
    },
    {
      "epoch": 0.6295843520782396,
      "grad_norm": 0.2048744261264801,
      "learning_rate": 1.133147113594041e-05,
      "loss": 0.0627,
      "step": 2060
    },
    {
      "epoch": 0.632640586797066,
      "grad_norm": 0.05055548995733261,
      "learning_rate": 1.1238361266294228e-05,
      "loss": 0.0662,
      "step": 2070
    },
    {
      "epoch": 0.6356968215158925,
      "grad_norm": 2.0886988639831543,
      "learning_rate": 1.1145251396648046e-05,
      "loss": 0.1094,
      "step": 2080
    },
    {
      "epoch": 0.6387530562347188,
      "grad_norm": 0.11593261361122131,
      "learning_rate": 1.1052141527001861e-05,
      "loss": 0.3611,
      "step": 2090
    },
    {
      "epoch": 0.6418092909535452,
      "grad_norm": 0.07993444800376892,
      "learning_rate": 1.095903165735568e-05,
      "loss": 0.0055,
      "step": 2100
    },
    {
      "epoch": 0.6448655256723717,
      "grad_norm": 0.13663113117218018,
      "learning_rate": 1.0865921787709497e-05,
      "loss": 0.0692,
      "step": 2110
    },
    {
      "epoch": 0.6479217603911981,
      "grad_norm": 0.09905444830656052,
      "learning_rate": 1.0772811918063314e-05,
      "loss": 0.008,
      "step": 2120
    },
    {
      "epoch": 0.6509779951100244,
      "grad_norm": 0.06681796908378601,
      "learning_rate": 1.0679702048417133e-05,
      "loss": 0.007,
      "step": 2130
    },
    {
      "epoch": 0.6540342298288508,
      "grad_norm": 0.08089947700500488,
      "learning_rate": 1.058659217877095e-05,
      "loss": 0.0875,
      "step": 2140
    },
    {
      "epoch": 0.6570904645476773,
      "grad_norm": 0.11339417099952698,
      "learning_rate": 1.0493482309124767e-05,
      "loss": 0.1986,
      "step": 2150
    },
    {
      "epoch": 0.6601466992665037,
      "grad_norm": 0.05615869536995888,
      "learning_rate": 1.0400372439478586e-05,
      "loss": 0.1701,
      "step": 2160
    },
    {
      "epoch": 0.66320293398533,
      "grad_norm": 0.09708325564861298,
      "learning_rate": 1.0307262569832403e-05,
      "loss": 0.059,
      "step": 2170
    },
    {
      "epoch": 0.6662591687041565,
      "grad_norm": 0.08505033701658249,
      "learning_rate": 1.021415270018622e-05,
      "loss": 0.2405,
      "step": 2180
    },
    {
      "epoch": 0.6693154034229829,
      "grad_norm": 0.12613414227962494,
      "learning_rate": 1.0121042830540038e-05,
      "loss": 0.0379,
      "step": 2190
    },
    {
      "epoch": 0.6723716381418093,
      "grad_norm": 0.11440026015043259,
      "learning_rate": 1.0027932960893856e-05,
      "loss": 0.0096,
      "step": 2200
    },
    {
      "epoch": 0.6754278728606357,
      "grad_norm": 1.7247629165649414,
      "learning_rate": 9.934823091247673e-06,
      "loss": 0.2441,
      "step": 2210
    },
    {
      "epoch": 0.6784841075794621,
      "grad_norm": 0.2626115679740906,
      "learning_rate": 9.84171322160149e-06,
      "loss": 0.1266,
      "step": 2220
    },
    {
      "epoch": 0.6815403422982885,
      "grad_norm": 0.08933160454034805,
      "learning_rate": 9.748603351955308e-06,
      "loss": 0.0932,
      "step": 2230
    },
    {
      "epoch": 0.684596577017115,
      "grad_norm": 0.18056859076023102,
      "learning_rate": 9.655493482309126e-06,
      "loss": 0.0528,
      "step": 2240
    },
    {
      "epoch": 0.6876528117359413,
      "grad_norm": 0.719739556312561,
      "learning_rate": 9.562383612662943e-06,
      "loss": 0.0848,
      "step": 2250
    },
    {
      "epoch": 0.6907090464547677,
      "grad_norm": 0.11645276844501495,
      "learning_rate": 9.46927374301676e-06,
      "loss": 0.0073,
      "step": 2260
    },
    {
      "epoch": 0.6937652811735942,
      "grad_norm": 0.05040796473622322,
      "learning_rate": 9.376163873370577e-06,
      "loss": 0.1393,
      "step": 2270
    },
    {
      "epoch": 0.6968215158924206,
      "grad_norm": 17.505489349365234,
      "learning_rate": 9.283054003724394e-06,
      "loss": 0.0733,
      "step": 2280
    },
    {
      "epoch": 0.6998777506112469,
      "grad_norm": 1.6304818391799927,
      "learning_rate": 9.189944134078213e-06,
      "loss": 0.0513,
      "step": 2290
    },
    {
      "epoch": 0.7029339853300733,
      "grad_norm": 0.1561206579208374,
      "learning_rate": 9.09683426443203e-06,
      "loss": 0.0575,
      "step": 2300
    },
    {
      "epoch": 0.7059902200488998,
      "grad_norm": 0.07211354374885559,
      "learning_rate": 9.003724394785847e-06,
      "loss": 0.1541,
      "step": 2310
    },
    {
      "epoch": 0.7090464547677262,
      "grad_norm": 0.12376312911510468,
      "learning_rate": 8.910614525139666e-06,
      "loss": 0.0061,
      "step": 2320
    },
    {
      "epoch": 0.7121026894865525,
      "grad_norm": 0.3487497568130493,
      "learning_rate": 8.817504655493483e-06,
      "loss": 0.0553,
      "step": 2330
    },
    {
      "epoch": 0.715158924205379,
      "grad_norm": 0.10781729221343994,
      "learning_rate": 8.7243947858473e-06,
      "loss": 0.0069,
      "step": 2340
    },
    {
      "epoch": 0.7182151589242054,
      "grad_norm": 5.861069679260254,
      "learning_rate": 8.631284916201118e-06,
      "loss": 0.1086,
      "step": 2350
    },
    {
      "epoch": 0.7212713936430318,
      "grad_norm": 0.2850921154022217,
      "learning_rate": 8.538175046554935e-06,
      "loss": 0.0912,
      "step": 2360
    },
    {
      "epoch": 0.7243276283618582,
      "grad_norm": 3.2116122245788574,
      "learning_rate": 8.445065176908753e-06,
      "loss": 0.1441,
      "step": 2370
    },
    {
      "epoch": 0.7273838630806846,
      "grad_norm": 0.4662390947341919,
      "learning_rate": 8.35195530726257e-06,
      "loss": 0.1017,
      "step": 2380
    },
    {
      "epoch": 0.730440097799511,
      "grad_norm": 0.05119713768362999,
      "learning_rate": 8.258845437616388e-06,
      "loss": 0.0102,
      "step": 2390
    },
    {
      "epoch": 0.7334963325183375,
      "grad_norm": 0.0849548801779747,
      "learning_rate": 8.165735567970205e-06,
      "loss": 0.0061,
      "step": 2400
    },
    {
      "epoch": 0.7365525672371638,
      "grad_norm": 0.035956285893917084,
      "learning_rate": 8.072625698324023e-06,
      "loss": 0.0035,
      "step": 2410
    },
    {
      "epoch": 0.7396088019559902,
      "grad_norm": 0.06464776396751404,
      "learning_rate": 7.979515828677841e-06,
      "loss": 0.0466,
      "step": 2420
    },
    {
      "epoch": 0.7426650366748166,
      "grad_norm": 0.5524429082870483,
      "learning_rate": 7.886405959031657e-06,
      "loss": 0.0704,
      "step": 2430
    },
    {
      "epoch": 0.7457212713936431,
      "grad_norm": 0.04895765706896782,
      "learning_rate": 7.793296089385474e-06,
      "loss": 0.0036,
      "step": 2440
    },
    {
      "epoch": 0.7487775061124694,
      "grad_norm": 0.03954509273171425,
      "learning_rate": 7.700186219739293e-06,
      "loss": 0.0032,
      "step": 2450
    },
    {
      "epoch": 0.7518337408312958,
      "grad_norm": 13.51663589477539,
      "learning_rate": 7.60707635009311e-06,
      "loss": 0.063,
      "step": 2460
    },
    {
      "epoch": 0.7548899755501223,
      "grad_norm": 0.11551866680383682,
      "learning_rate": 7.5139664804469275e-06,
      "loss": 0.0028,
      "step": 2470
    },
    {
      "epoch": 0.7579462102689487,
      "grad_norm": 7.975831031799316,
      "learning_rate": 7.420856610800745e-06,
      "loss": 0.2147,
      "step": 2480
    },
    {
      "epoch": 0.761002444987775,
      "grad_norm": 0.10641849786043167,
      "learning_rate": 7.3277467411545625e-06,
      "loss": 0.1884,
      "step": 2490
    },
    {
      "epoch": 0.7640586797066015,
      "grad_norm": 0.3172556757926941,
      "learning_rate": 7.23463687150838e-06,
      "loss": 0.0692,
      "step": 2500
    },
    {
      "epoch": 0.7671149144254279,
      "grad_norm": 0.12498392164707184,
      "learning_rate": 7.1415270018621975e-06,
      "loss": 0.1202,
      "step": 2510
    },
    {
      "epoch": 0.7701711491442543,
      "grad_norm": 0.1759582906961441,
      "learning_rate": 7.0484171322160155e-06,
      "loss": 0.0624,
      "step": 2520
    },
    {
      "epoch": 0.7732273838630807,
      "grad_norm": 0.09976211935281754,
      "learning_rate": 6.9553072625698325e-06,
      "loss": 0.1345,
      "step": 2530
    },
    {
      "epoch": 0.7762836185819071,
      "grad_norm": 0.10823414474725723,
      "learning_rate": 6.8621973929236505e-06,
      "loss": 0.0963,
      "step": 2540
    },
    {
      "epoch": 0.7793398533007335,
      "grad_norm": 0.09119368344545364,
      "learning_rate": 6.7690875232774675e-06,
      "loss": 0.0065,
      "step": 2550
    },
    {
      "epoch": 0.78239608801956,
      "grad_norm": 0.22418056428432465,
      "learning_rate": 6.675977653631285e-06,
      "loss": 0.1055,
      "step": 2560
    },
    {
      "epoch": 0.7854523227383863,
      "grad_norm": 0.16302266716957092,
      "learning_rate": 6.5828677839851025e-06,
      "loss": 0.1129,
      "step": 2570
    },
    {
      "epoch": 0.7885085574572127,
      "grad_norm": 0.12889882922172546,
      "learning_rate": 6.48975791433892e-06,
      "loss": 0.1578,
      "step": 2580
    },
    {
      "epoch": 0.7915647921760391,
      "grad_norm": 0.4161546528339386,
      "learning_rate": 6.3966480446927375e-06,
      "loss": 0.0546,
      "step": 2590
    },
    {
      "epoch": 0.7946210268948656,
      "grad_norm": 0.4652371108531952,
      "learning_rate": 6.3035381750465554e-06,
      "loss": 0.0736,
      "step": 2600
    },
    {
      "epoch": 0.7976772616136919,
      "grad_norm": 0.1676311194896698,
      "learning_rate": 6.2104283054003725e-06,
      "loss": 0.1377,
      "step": 2610
    },
    {
      "epoch": 0.8007334963325183,
      "grad_norm": 0.49725136160850525,
      "learning_rate": 6.1173184357541904e-06,
      "loss": 0.0478,
      "step": 2620
    },
    {
      "epoch": 0.8037897310513448,
      "grad_norm": 0.3372693359851837,
      "learning_rate": 6.0242085661080075e-06,
      "loss": 0.107,
      "step": 2630
    },
    {
      "epoch": 0.8068459657701712,
      "grad_norm": 11.673458099365234,
      "learning_rate": 5.9310986964618254e-06,
      "loss": 0.1794,
      "step": 2640
    },
    {
      "epoch": 0.8099022004889975,
      "grad_norm": 0.13454145193099976,
      "learning_rate": 5.8379888268156425e-06,
      "loss": 0.005,
      "step": 2650
    },
    {
      "epoch": 0.812958435207824,
      "grad_norm": 0.10115115344524384,
      "learning_rate": 5.74487895716946e-06,
      "loss": 0.0428,
      "step": 2660
    },
    {
      "epoch": 0.8160146699266504,
      "grad_norm": 0.1693265289068222,
      "learning_rate": 5.6517690875232775e-06,
      "loss": 0.0553,
      "step": 2670
    },
    {
      "epoch": 0.8190709046454768,
      "grad_norm": 0.040254298597574234,
      "learning_rate": 5.558659217877095e-06,
      "loss": 0.0041,
      "step": 2680
    },
    {
      "epoch": 0.8221271393643031,
      "grad_norm": 0.06211506202816963,
      "learning_rate": 5.4655493482309125e-06,
      "loss": 0.0301,
      "step": 2690
    },
    {
      "epoch": 0.8251833740831296,
      "grad_norm": 0.03791659697890282,
      "learning_rate": 5.37243947858473e-06,
      "loss": 0.1197,
      "step": 2700
    },
    {
      "epoch": 0.828239608801956,
      "grad_norm": 0.06369005143642426,
      "learning_rate": 5.2793296089385475e-06,
      "loss": 0.065,
      "step": 2710
    },
    {
      "epoch": 0.8312958435207825,
      "grad_norm": 0.15808682143688202,
      "learning_rate": 5.186219739292365e-06,
      "loss": 0.0554,
      "step": 2720
    },
    {
      "epoch": 0.8343520782396088,
      "grad_norm": 0.05038405582308769,
      "learning_rate": 5.0931098696461825e-06,
      "loss": 0.082,
      "step": 2730
    },
    {
      "epoch": 0.8374083129584352,
      "grad_norm": 0.26016944646835327,
      "learning_rate": 4.9999999999999996e-06,
      "loss": 0.0731,
      "step": 2740
    },
    {
      "epoch": 0.8404645476772616,
      "grad_norm": 0.03949306160211563,
      "learning_rate": 4.9068901303538175e-06,
      "loss": 0.2097,
      "step": 2750
    },
    {
      "epoch": 0.843520782396088,
      "grad_norm": 0.10841944068670273,
      "learning_rate": 4.813780260707635e-06,
      "loss": 0.1337,
      "step": 2760
    },
    {
      "epoch": 0.8465770171149144,
      "grad_norm": 0.08589453250169754,
      "learning_rate": 4.7206703910614525e-06,
      "loss": 0.1875,
      "step": 2770
    },
    {
      "epoch": 0.8496332518337408,
      "grad_norm": 7.176298141479492,
      "learning_rate": 4.62756052141527e-06,
      "loss": 0.1634,
      "step": 2780
    },
    {
      "epoch": 0.8526894865525673,
      "grad_norm": 0.20672304928302765,
      "learning_rate": 4.5344506517690875e-06,
      "loss": 0.0686,
      "step": 2790
    },
    {
      "epoch": 0.8557457212713936,
      "grad_norm": 0.03756504878401756,
      "learning_rate": 4.441340782122905e-06,
      "loss": 0.121,
      "step": 2800
    },
    {
      "epoch": 0.85880195599022,
      "grad_norm": 0.509375274181366,
      "learning_rate": 4.348230912476723e-06,
      "loss": 0.0217,
      "step": 2810
    },
    {
      "epoch": 0.8618581907090465,
      "grad_norm": 13.620018005371094,
      "learning_rate": 4.2551210428305395e-06,
      "loss": 0.119,
      "step": 2820
    },
    {
      "epoch": 0.8649144254278729,
      "grad_norm": 0.0516686849296093,
      "learning_rate": 4.1620111731843575e-06,
      "loss": 0.0654,
      "step": 2830
    },
    {
      "epoch": 0.8679706601466992,
      "grad_norm": 0.04489528760313988,
      "learning_rate": 4.068901303538175e-06,
      "loss": 0.0572,
      "step": 2840
    },
    {
      "epoch": 0.8710268948655256,
      "grad_norm": 0.06253698468208313,
      "learning_rate": 3.9757914338919925e-06,
      "loss": 0.0448,
      "step": 2850
    },
    {
      "epoch": 0.8740831295843521,
      "grad_norm": 0.06379766017198563,
      "learning_rate": 3.88268156424581e-06,
      "loss": 0.0062,
      "step": 2860
    },
    {
      "epoch": 0.8771393643031785,
      "grad_norm": 0.058733634650707245,
      "learning_rate": 3.789571694599628e-06,
      "loss": 0.0691,
      "step": 2870
    },
    {
      "epoch": 0.8801955990220048,
      "grad_norm": 0.06447366625070572,
      "learning_rate": 3.696461824953445e-06,
      "loss": 0.1691,
      "step": 2880
    },
    {
      "epoch": 0.8832518337408313,
      "grad_norm": 0.04590935632586479,
      "learning_rate": 3.6033519553072625e-06,
      "loss": 0.1227,
      "step": 2890
    },
    {
      "epoch": 0.8863080684596577,
      "grad_norm": 0.39994603395462036,
      "learning_rate": 3.5102420856610804e-06,
      "loss": 0.1183,
      "step": 2900
    },
    {
      "epoch": 0.8893643031784841,
      "grad_norm": 0.396999329328537,
      "learning_rate": 3.417132216014898e-06,
      "loss": 0.0406,
      "step": 2910
    },
    {
      "epoch": 0.8924205378973105,
      "grad_norm": 0.10507692396640778,
      "learning_rate": 3.324022346368715e-06,
      "loss": 0.053,
      "step": 2920
    },
    {
      "epoch": 0.8954767726161369,
      "grad_norm": 0.3681839406490326,
      "learning_rate": 3.2309124767225324e-06,
      "loss": 0.0048,
      "step": 2930
    },
    {
      "epoch": 0.8985330073349633,
      "grad_norm": 0.21031515300273895,
      "learning_rate": 3.1378026070763504e-06,
      "loss": 0.0053,
      "step": 2940
    },
    {
      "epoch": 0.9015892420537898,
      "grad_norm": 6.659632682800293,
      "learning_rate": 3.044692737430168e-06,
      "loss": 0.0991,
      "step": 2950
    },
    {
      "epoch": 0.9046454767726161,
      "grad_norm": 0.049913499504327774,
      "learning_rate": 2.951582867783985e-06,
      "loss": 0.0038,
      "step": 2960
    },
    {
      "epoch": 0.9077017114914425,
      "grad_norm": 0.09325915575027466,
      "learning_rate": 2.8584729981378024e-06,
      "loss": 0.0033,
      "step": 2970
    },
    {
      "epoch": 0.910757946210269,
      "grad_norm": 0.08547964692115784,
      "learning_rate": 2.7653631284916204e-06,
      "loss": 0.0684,
      "step": 2980
    },
    {
      "epoch": 0.9138141809290954,
      "grad_norm": 9.980097770690918,
      "learning_rate": 2.672253258845438e-06,
      "loss": 0.0735,
      "step": 2990
    },
    {
      "epoch": 0.9168704156479217,
      "grad_norm": 0.03152007982134819,
      "learning_rate": 2.5791433891992554e-06,
      "loss": 0.049,
      "step": 3000
    },
    {
      "epoch": 0.9199266503667481,
      "grad_norm": 0.039837196469306946,
      "learning_rate": 2.4860335195530724e-06,
      "loss": 0.0686,
      "step": 3010
    },
    {
      "epoch": 0.9229828850855746,
      "grad_norm": 12.22442626953125,
      "learning_rate": 2.3929236499068903e-06,
      "loss": 0.0554,
      "step": 3020
    },
    {
      "epoch": 0.926039119804401,
      "grad_norm": 0.13149628043174744,
      "learning_rate": 2.299813780260708e-06,
      "loss": 0.1026,
      "step": 3030
    },
    {
      "epoch": 0.9290953545232273,
      "grad_norm": 0.044785063713788986,
      "learning_rate": 2.2067039106145253e-06,
      "loss": 0.0042,
      "step": 3040
    },
    {
      "epoch": 0.9321515892420538,
      "grad_norm": 0.042241331189870834,
      "learning_rate": 2.1135940409683424e-06,
      "loss": 0.0681,
      "step": 3050
    },
    {
      "epoch": 0.9352078239608802,
      "grad_norm": 0.0555758960545063,
      "learning_rate": 2.0204841713221603e-06,
      "loss": 0.0282,
      "step": 3060
    },
    {
      "epoch": 0.9382640586797066,
      "grad_norm": 0.07724986225366592,
      "learning_rate": 1.927374301675978e-06,
      "loss": 0.1247,
      "step": 3070
    },
    {
      "epoch": 0.941320293398533,
      "grad_norm": 0.08575576543807983,
      "learning_rate": 1.8342644320297951e-06,
      "loss": 0.0689,
      "step": 3080
    },
    {
      "epoch": 0.9443765281173594,
      "grad_norm": 0.2837558686733246,
      "learning_rate": 1.7411545623836128e-06,
      "loss": 0.102,
      "step": 3090
    },
    {
      "epoch": 0.9474327628361858,
      "grad_norm": 0.1205146312713623,
      "learning_rate": 1.6480446927374301e-06,
      "loss": 0.1657,
      "step": 3100
    },
    {
      "epoch": 0.9504889975550123,
      "grad_norm": 0.16669517755508423,
      "learning_rate": 1.5549348230912478e-06,
      "loss": 0.0078,
      "step": 3110
    },
    {
      "epoch": 0.9535452322738386,
      "grad_norm": 15.247810363769531,
      "learning_rate": 1.4618249534450651e-06,
      "loss": 0.0175,
      "step": 3120
    },
    {
      "epoch": 0.956601466992665,
      "grad_norm": 0.20827865600585938,
      "learning_rate": 1.3687150837988828e-06,
      "loss": 0.0353,
      "step": 3130
    },
    {
      "epoch": 0.9596577017114915,
      "grad_norm": 0.07442985475063324,
      "learning_rate": 1.2756052141527001e-06,
      "loss": 0.0415,
      "step": 3140
    },
    {
      "epoch": 0.9627139364303179,
      "grad_norm": 0.040485650300979614,
      "learning_rate": 1.1824953445065178e-06,
      "loss": 0.1062,
      "step": 3150
    },
    {
      "epoch": 0.9657701711491442,
      "grad_norm": 0.12251528352499008,
      "learning_rate": 1.0893854748603353e-06,
      "loss": 0.043,
      "step": 3160
    },
    {
      "epoch": 0.9688264058679706,
      "grad_norm": 0.050180476158857346,
      "learning_rate": 9.962756052141528e-07,
      "loss": 0.03,
      "step": 3170
    },
    {
      "epoch": 0.9718826405867971,
      "grad_norm": 0.0506335012614727,
      "learning_rate": 9.031657355679702e-07,
      "loss": 0.2524,
      "step": 3180
    },
    {
      "epoch": 0.9749388753056235,
      "grad_norm": 28.770177841186523,
      "learning_rate": 8.100558659217877e-07,
      "loss": 0.0968,
      "step": 3190
    },
    {
      "epoch": 0.9779951100244498,
      "grad_norm": 0.03078501671552658,
      "learning_rate": 7.169459962756052e-07,
      "loss": 0.0758,
      "step": 3200
    },
    {
      "epoch": 0.9810513447432763,
      "grad_norm": 16.90084457397461,
      "learning_rate": 6.238361266294227e-07,
      "loss": 0.1547,
      "step": 3210
    },
    {
      "epoch": 0.9841075794621027,
      "grad_norm": 0.1555572897195816,
      "learning_rate": 5.307262569832402e-07,
      "loss": 0.0491,
      "step": 3220
    },
    {
      "epoch": 0.9871638141809291,
      "grad_norm": 0.04730997607111931,
      "learning_rate": 4.3761638733705775e-07,
      "loss": 0.0039,
      "step": 3230
    },
    {
      "epoch": 0.9902200488997555,
      "grad_norm": 3.162146806716919,
      "learning_rate": 3.4450651769087525e-07,
      "loss": 0.1387,
      "step": 3240
    },
    {
      "epoch": 0.9932762836185819,
      "grad_norm": 2.292222499847412,
      "learning_rate": 2.5139664804469275e-07,
      "loss": 0.1184,
      "step": 3250
    },
    {
      "epoch": 0.9963325183374083,
      "grad_norm": 0.054889462888240814,
      "learning_rate": 1.5828677839851024e-07,
      "loss": 0.0065,
      "step": 3260
    },
    {
      "epoch": 0.9993887530562348,
      "grad_norm": 0.310970664024353,
      "learning_rate": 6.517690875232774e-08,
      "loss": 0.0087,
      "step": 3270
    },
    {
      "epoch": 1.0,
      "eval_accuracy": 0.9845683728036669,
      "eval_f1": 0.9839899931917298,
      "eval_loss": 0.06383595615625381,
      "eval_model_preparation_time": 0.0032,
      "eval_runtime": 411.4096,
      "eval_samples_per_second": 15.909,
      "eval_steps_per_second": 1.991,
      "step": 3272
    }
  ],
  "logging_steps": 10,
  "max_steps": 3272,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 2.2958418080824934e+18,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}