{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5476701200310347,
  "eval_steps": 500,
  "global_step": 24000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00022819588334626443,
      "grad_norm": 38.25,
      "learning_rate": 1.5209125475285173e-07,
      "loss": 1.723,
      "step": 10
    },
    {
      "epoch": 0.00045639176669252886,
      "grad_norm": 26.125,
      "learning_rate": 3.0418250950570346e-07,
      "loss": 1.7514,
      "step": 20
    },
    {
      "epoch": 0.0006845876500387933,
      "grad_norm": 45.75,
      "learning_rate": 4.562737642585552e-07,
      "loss": 1.8018,
      "step": 30
    },
    {
      "epoch": 0.0009127835333850577,
      "grad_norm": 73.0,
      "learning_rate": 6.083650190114069e-07,
      "loss": 1.6866,
      "step": 40
    },
    {
      "epoch": 0.0011409794167313223,
      "grad_norm": 33.0,
      "learning_rate": 7.604562737642586e-07,
      "loss": 1.7271,
      "step": 50
    },
    {
      "epoch": 0.0013691753000775865,
      "grad_norm": 55.5,
      "learning_rate": 9.125475285171104e-07,
      "loss": 1.7047,
      "step": 60
    },
    {
      "epoch": 0.001597371183423851,
      "grad_norm": 20.5,
      "learning_rate": 1.064638783269962e-06,
      "loss": 1.6827,
      "step": 70
    },
    {
      "epoch": 0.0018255670667701154,
      "grad_norm": 33.0,
      "learning_rate": 1.2167300380228138e-06,
      "loss": 1.7259,
      "step": 80
    },
    {
      "epoch": 0.00205376295011638,
      "grad_norm": 32.75,
      "learning_rate": 1.3688212927756656e-06,
      "loss": 1.7137,
      "step": 90
    },
    {
      "epoch": 0.0022819588334626446,
      "grad_norm": 4.65625,
      "learning_rate": 1.5209125475285172e-06,
      "loss": 1.6195,
      "step": 100
    },
    {
      "epoch": 0.002510154716808909,
      "grad_norm": 4.4375,
      "learning_rate": 1.6730038022813688e-06,
      "loss": 1.668,
      "step": 110
    },
    {
      "epoch": 0.002738350600155173,
      "grad_norm": 3.953125,
      "learning_rate": 1.8250950570342208e-06,
      "loss": 1.6049,
      "step": 120
    },
    {
      "epoch": 0.0029665464835014377,
      "grad_norm": 3.84375,
      "learning_rate": 1.9771863117870722e-06,
      "loss": 1.6246,
      "step": 130
    },
    {
      "epoch": 0.003194742366847702,
      "grad_norm": 4.0625,
      "learning_rate": 2.129277566539924e-06,
      "loss": 1.5938,
      "step": 140
    },
    {
      "epoch": 0.0034229382501939666,
      "grad_norm": 3.75,
      "learning_rate": 2.281368821292776e-06,
      "loss": 1.572,
      "step": 150
    },
    {
      "epoch": 0.003651134133540231,
      "grad_norm": 3.890625,
      "learning_rate": 2.4334600760456276e-06,
      "loss": 1.5447,
      "step": 160
    },
    {
      "epoch": 0.0038793300168864955,
      "grad_norm": 3.671875,
      "learning_rate": 2.585551330798479e-06,
      "loss": 1.5597,
      "step": 170
    },
    {
      "epoch": 0.00410752590023276,
      "grad_norm": 3.484375,
      "learning_rate": 2.7376425855513313e-06,
      "loss": 1.5742,
      "step": 180
    },
    {
      "epoch": 0.004335721783579024,
      "grad_norm": 3.4375,
      "learning_rate": 2.8897338403041826e-06,
      "loss": 1.5406,
      "step": 190
    },
    {
      "epoch": 0.004563917666925289,
      "grad_norm": 5.0,
      "learning_rate": 3.0418250950570345e-06,
      "loss": 1.5553,
      "step": 200
    },
    {
      "epoch": 0.004792113550271553,
      "grad_norm": 3.625,
      "learning_rate": 3.1939163498098863e-06,
      "loss": 1.577,
      "step": 210
    },
    {
      "epoch": 0.005020309433617818,
      "grad_norm": 3.984375,
      "learning_rate": 3.3460076045627376e-06,
      "loss": 1.5313,
      "step": 220
    },
    {
      "epoch": 0.005248505316964082,
      "grad_norm": 4.4375,
      "learning_rate": 3.4980988593155894e-06,
      "loss": 1.5889,
      "step": 230
    },
    {
      "epoch": 0.005476701200310346,
      "grad_norm": 3.78125,
      "learning_rate": 3.6501901140684417e-06,
      "loss": 1.5283,
      "step": 240
    },
    {
      "epoch": 0.005704897083656611,
      "grad_norm": 3.9375,
      "learning_rate": 3.8022813688212926e-06,
      "loss": 1.5268,
      "step": 250
    },
    {
      "epoch": 0.005933092967002875,
      "grad_norm": 3.6875,
      "learning_rate": 3.9543726235741444e-06,
      "loss": 1.5106,
      "step": 260
    },
    {
      "epoch": 0.00616128885034914,
      "grad_norm": 3.96875,
      "learning_rate": 4.106463878326996e-06,
      "loss": 1.5173,
      "step": 270
    },
    {
      "epoch": 0.006389484733695404,
      "grad_norm": 4.3125,
      "learning_rate": 4.258555133079848e-06,
      "loss": 1.5211,
      "step": 280
    },
    {
      "epoch": 0.0066176806170416685,
      "grad_norm": 4.0,
      "learning_rate": 4.4106463878327e-06,
      "loss": 1.5162,
      "step": 290
    },
    {
      "epoch": 0.006845876500387933,
      "grad_norm": 3.796875,
      "learning_rate": 4.562737642585552e-06,
      "loss": 1.516,
      "step": 300
    },
    {
      "epoch": 0.007074072383734197,
      "grad_norm": 4.15625,
      "learning_rate": 4.7148288973384035e-06,
      "loss": 1.5391,
      "step": 310
    },
    {
      "epoch": 0.007302268267080462,
      "grad_norm": 3.34375,
      "learning_rate": 4.866920152091255e-06,
      "loss": 1.5261,
      "step": 320
    },
    {
      "epoch": 0.007530464150426726,
      "grad_norm": 4.0625,
      "learning_rate": 5.019011406844106e-06,
      "loss": 1.5017,
      "step": 330
    },
    {
      "epoch": 0.007758660033772991,
      "grad_norm": 3.59375,
      "learning_rate": 5.171102661596958e-06,
      "loss": 1.5055,
      "step": 340
    },
    {
      "epoch": 0.007986855917119255,
      "grad_norm": 3.6875,
      "learning_rate": 5.323193916349811e-06,
      "loss": 1.5308,
      "step": 350
    },
    {
      "epoch": 0.00821505180046552,
      "grad_norm": 3.890625,
      "learning_rate": 5.4752851711026625e-06,
      "loss": 1.5245,
      "step": 360
    },
    {
      "epoch": 0.008443247683811784,
      "grad_norm": 3.875,
      "learning_rate": 5.627376425855514e-06,
      "loss": 1.458,
      "step": 370
    },
    {
      "epoch": 0.008671443567158049,
      "grad_norm": 3.921875,
      "learning_rate": 5.779467680608365e-06,
      "loss": 1.4163,
      "step": 380
    },
    {
      "epoch": 0.008899639450504314,
      "grad_norm": 3.734375,
      "learning_rate": 5.931558935361217e-06,
      "loss": 1.4617,
      "step": 390
    },
    {
      "epoch": 0.009127835333850578,
      "grad_norm": 3.578125,
      "learning_rate": 6.083650190114069e-06,
      "loss": 1.478,
      "step": 400
    },
    {
      "epoch": 0.009356031217196841,
      "grad_norm": 4.25,
      "learning_rate": 6.235741444866921e-06,
      "loss": 1.4673,
      "step": 410
    },
    {
      "epoch": 0.009584227100543106,
      "grad_norm": 3.609375,
      "learning_rate": 6.3878326996197725e-06,
      "loss": 1.4664,
      "step": 420
    },
    {
      "epoch": 0.00981242298388937,
      "grad_norm": 3.90625,
      "learning_rate": 6.539923954372624e-06,
      "loss": 1.4617,
      "step": 430
    },
    {
      "epoch": 0.010040618867235635,
      "grad_norm": 3.953125,
      "learning_rate": 6.692015209125475e-06,
      "loss": 1.5364,
      "step": 440
    },
    {
      "epoch": 0.0102688147505819,
      "grad_norm": 3.890625,
      "learning_rate": 6.844106463878327e-06,
      "loss": 1.4874,
      "step": 450
    },
    {
      "epoch": 0.010497010633928164,
      "grad_norm": 3.59375,
      "learning_rate": 6.996197718631179e-06,
      "loss": 1.4275,
      "step": 460
    },
    {
      "epoch": 0.01072520651727443,
      "grad_norm": 3.703125,
      "learning_rate": 7.148288973384031e-06,
      "loss": 1.3942,
      "step": 470
    },
    {
      "epoch": 0.010953402400620692,
      "grad_norm": 3.859375,
      "learning_rate": 7.300380228136883e-06,
      "loss": 1.462,
      "step": 480
    },
    {
      "epoch": 0.011181598283966957,
      "grad_norm": 3.796875,
      "learning_rate": 7.452471482889735e-06,
      "loss": 1.4269,
      "step": 490
    },
    {
      "epoch": 0.011409794167313221,
      "grad_norm": 3.75,
      "learning_rate": 7.604562737642585e-06,
      "loss": 1.4625,
      "step": 500
    },
    {
      "epoch": 0.011637990050659486,
      "grad_norm": 4.28125,
      "learning_rate": 7.756653992395437e-06,
      "loss": 1.4103,
      "step": 510
    },
    {
      "epoch": 0.01186618593400575,
      "grad_norm": 4.125,
      "learning_rate": 7.908745247148289e-06,
      "loss": 1.4796,
      "step": 520
    },
    {
      "epoch": 0.012094381817352015,
      "grad_norm": 3.796875,
      "learning_rate": 8.06083650190114e-06,
      "loss": 1.4429,
      "step": 530
    },
    {
      "epoch": 0.01232257770069828,
      "grad_norm": 3.8125,
      "learning_rate": 8.212927756653993e-06,
      "loss": 1.3771,
      "step": 540
    },
    {
      "epoch": 0.012550773584044543,
      "grad_norm": 3.84375,
      "learning_rate": 8.365019011406846e-06,
      "loss": 1.4054,
      "step": 550
    },
    {
      "epoch": 0.012778969467390808,
      "grad_norm": 3.59375,
      "learning_rate": 8.517110266159696e-06,
      "loss": 1.4619,
      "step": 560
    },
    {
      "epoch": 0.013007165350737072,
      "grad_norm": 4.0,
      "learning_rate": 8.669201520912548e-06,
      "loss": 1.4362,
      "step": 570
    },
    {
      "epoch": 0.013235361234083337,
      "grad_norm": 3.765625,
      "learning_rate": 8.8212927756654e-06,
      "loss": 1.3942,
      "step": 580
    },
    {
      "epoch": 0.013463557117429602,
      "grad_norm": 3.59375,
      "learning_rate": 8.973384030418252e-06,
      "loss": 1.3715,
      "step": 590
    },
    {
      "epoch": 0.013691753000775866,
      "grad_norm": 3.59375,
      "learning_rate": 9.125475285171103e-06,
      "loss": 1.3841,
      "step": 600
    },
    {
      "epoch": 0.013919948884122131,
      "grad_norm": 3.5625,
      "learning_rate": 9.277566539923955e-06,
      "loss": 1.4372,
      "step": 610
    },
    {
      "epoch": 0.014148144767468394,
      "grad_norm": 3.4375,
      "learning_rate": 9.429657794676807e-06,
      "loss": 1.4196,
      "step": 620
    },
    {
      "epoch": 0.014376340650814659,
      "grad_norm": 3.578125,
      "learning_rate": 9.581749049429659e-06,
      "loss": 1.4417,
      "step": 630
    },
    {
      "epoch": 0.014604536534160923,
      "grad_norm": 3.6875,
      "learning_rate": 9.73384030418251e-06,
      "loss": 1.4401,
      "step": 640
    },
    {
      "epoch": 0.014832732417507188,
      "grad_norm": 3.640625,
      "learning_rate": 9.885931558935362e-06,
      "loss": 1.4314,
      "step": 650
    },
    {
      "epoch": 0.015060928300853453,
      "grad_norm": 3.53125,
      "learning_rate": 1.0038022813688212e-05,
      "loss": 1.3939,
      "step": 660
    },
    {
      "epoch": 0.015289124184199717,
      "grad_norm": 3.53125,
      "learning_rate": 1.0190114068441066e-05,
      "loss": 1.3851,
      "step": 670
    },
    {
      "epoch": 0.015517320067545982,
      "grad_norm": 3.390625,
      "learning_rate": 1.0342205323193916e-05,
      "loss": 1.4583,
      "step": 680
    },
    {
      "epoch": 0.015745515950892245,
      "grad_norm": 3.828125,
      "learning_rate": 1.049429657794677e-05,
      "loss": 1.3869,
      "step": 690
    },
    {
      "epoch": 0.01597371183423851,
      "grad_norm": 4.09375,
      "learning_rate": 1.0646387832699621e-05,
      "loss": 1.3684,
      "step": 700
    },
    {
      "epoch": 0.016201907717584774,
      "grad_norm": 4.21875,
      "learning_rate": 1.0798479087452472e-05,
      "loss": 1.3874,
      "step": 710
    },
    {
      "epoch": 0.01643010360093104,
      "grad_norm": 3.828125,
      "learning_rate": 1.0950570342205325e-05,
      "loss": 1.3773,
      "step": 720
    },
    {
      "epoch": 0.016658299484277304,
      "grad_norm": 3.5,
      "learning_rate": 1.1102661596958175e-05,
      "loss": 1.3338,
      "step": 730
    },
    {
      "epoch": 0.01688649536762357,
      "grad_norm": 3.875,
      "learning_rate": 1.1254752851711029e-05,
      "loss": 1.4083,
      "step": 740
    },
    {
      "epoch": 0.017114691250969833,
      "grad_norm": 3.703125,
      "learning_rate": 1.1406844106463879e-05,
      "loss": 1.4171,
      "step": 750
    },
    {
      "epoch": 0.017342887134316098,
      "grad_norm": 3.765625,
      "learning_rate": 1.155893536121673e-05,
      "loss": 1.4492,
      "step": 760
    },
    {
      "epoch": 0.017571083017662362,
      "grad_norm": 3.6875,
      "learning_rate": 1.1711026615969582e-05,
      "loss": 1.3698,
      "step": 770
    },
    {
      "epoch": 0.017799278901008627,
      "grad_norm": 4.875,
      "learning_rate": 1.1863117870722434e-05,
      "loss": 1.4096,
      "step": 780
    },
    {
      "epoch": 0.01802747478435489,
      "grad_norm": 3.90625,
      "learning_rate": 1.2015209125475286e-05,
      "loss": 1.4024,
      "step": 790
    },
    {
      "epoch": 0.018255670667701156,
      "grad_norm": 3.359375,
      "learning_rate": 1.2167300380228138e-05,
      "loss": 1.4165,
      "step": 800
    },
    {
      "epoch": 0.018483866551047418,
      "grad_norm": 3.40625,
      "learning_rate": 1.2319391634980991e-05,
      "loss": 1.3782,
      "step": 810
    },
    {
      "epoch": 0.018712062434393682,
      "grad_norm": 3.703125,
      "learning_rate": 1.2471482889733841e-05,
      "loss": 1.4021,
      "step": 820
    },
    {
      "epoch": 0.018940258317739947,
      "grad_norm": 3.546875,
      "learning_rate": 1.2623574144486692e-05,
      "loss": 1.3975,
      "step": 830
    },
    {
      "epoch": 0.01916845420108621,
      "grad_norm": 3.265625,
      "learning_rate": 1.2775665399239545e-05,
      "loss": 1.3528,
      "step": 840
    },
    {
      "epoch": 0.019396650084432476,
      "grad_norm": 3.546875,
      "learning_rate": 1.2927756653992395e-05,
      "loss": 1.3684,
      "step": 850
    },
    {
      "epoch": 0.01962484596777874,
      "grad_norm": 3.578125,
      "learning_rate": 1.3079847908745249e-05,
      "loss": 1.3472,
      "step": 860
    },
    {
      "epoch": 0.019853041851125006,
      "grad_norm": 4.0,
      "learning_rate": 1.32319391634981e-05,
      "loss": 1.4346,
      "step": 870
    },
    {
      "epoch": 0.02008123773447127,
      "grad_norm": 3.8125,
      "learning_rate": 1.338403041825095e-05,
      "loss": 1.4504,
      "step": 880
    },
    {
      "epoch": 0.020309433617817535,
      "grad_norm": 3.859375,
      "learning_rate": 1.3536121673003804e-05,
      "loss": 1.3516,
      "step": 890
    },
    {
      "epoch": 0.0205376295011638,
      "grad_norm": 3.640625,
      "learning_rate": 1.3688212927756654e-05,
      "loss": 1.37,
      "step": 900
    },
    {
      "epoch": 0.020765825384510064,
      "grad_norm": 3.46875,
      "learning_rate": 1.3840304182509508e-05,
      "loss": 1.4166,
      "step": 910
    },
    {
      "epoch": 0.02099402126785633,
      "grad_norm": 3.765625,
      "learning_rate": 1.3992395437262358e-05,
      "loss": 1.4551,
      "step": 920
    },
    {
      "epoch": 0.021222217151202594,
      "grad_norm": 3.75,
      "learning_rate": 1.4144486692015211e-05,
      "loss": 1.3976,
      "step": 930
    },
    {
      "epoch": 0.02145041303454886,
      "grad_norm": 3.90625,
      "learning_rate": 1.4296577946768061e-05,
      "loss": 1.3726,
      "step": 940
    },
    {
      "epoch": 0.02167860891789512,
      "grad_norm": 3.4375,
      "learning_rate": 1.4448669201520913e-05,
      "loss": 1.412,
      "step": 950
    },
    {
      "epoch": 0.021906804801241384,
      "grad_norm": 3.375,
      "learning_rate": 1.4600760456273767e-05,
      "loss": 1.3502,
      "step": 960
    },
    {
      "epoch": 0.02213500068458765,
      "grad_norm": 3.4375,
      "learning_rate": 1.4752851711026617e-05,
      "loss": 1.4424,
      "step": 970
    },
    {
      "epoch": 0.022363196567933914,
      "grad_norm": 3.375,
      "learning_rate": 1.490494296577947e-05,
      "loss": 1.3747,
      "step": 980
    },
    {
      "epoch": 0.022591392451280178,
      "grad_norm": 3.46875,
      "learning_rate": 1.505703422053232e-05,
      "loss": 1.3558,
      "step": 990
    },
    {
      "epoch": 0.022819588334626443,
      "grad_norm": 3.203125,
      "learning_rate": 1.520912547528517e-05,
      "loss": 1.3847,
      "step": 1000
    },
    {
      "epoch": 0.023047784217972708,
      "grad_norm": 3.40625,
      "learning_rate": 1.5361216730038024e-05,
      "loss": 1.3874,
      "step": 1010
    },
    {
      "epoch": 0.023275980101318972,
      "grad_norm": 3.890625,
      "learning_rate": 1.5513307984790874e-05,
      "loss": 1.3789,
      "step": 1020
    },
    {
      "epoch": 0.023504175984665237,
      "grad_norm": 3.40625,
      "learning_rate": 1.5665399239543728e-05,
      "loss": 1.3891,
      "step": 1030
    },
    {
      "epoch": 0.0237323718680115,
      "grad_norm": 3.75,
      "learning_rate": 1.5817490494296578e-05,
      "loss": 1.3984,
      "step": 1040
    },
    {
      "epoch": 0.023960567751357766,
      "grad_norm": 3.8125,
      "learning_rate": 1.596958174904943e-05,
      "loss": 1.3308,
      "step": 1050
    },
    {
      "epoch": 0.02418876363470403,
      "grad_norm": 3.625,
      "learning_rate": 1.612167300380228e-05,
      "loss": 1.3975,
      "step": 1060
    },
    {
      "epoch": 0.024416959518050296,
      "grad_norm": 3.46875,
      "learning_rate": 1.6273764258555135e-05,
      "loss": 1.427,
      "step": 1070
    },
    {
      "epoch": 0.02464515540139656,
      "grad_norm": 4.46875,
      "learning_rate": 1.6425855513307985e-05,
      "loss": 1.3605,
      "step": 1080
    },
    {
      "epoch": 0.024873351284742825,
      "grad_norm": 3.1875,
      "learning_rate": 1.657794676806084e-05,
      "loss": 1.4108,
      "step": 1090
    },
    {
      "epoch": 0.025101547168089086,
      "grad_norm": 3.671875,
      "learning_rate": 1.6730038022813692e-05,
      "loss": 1.3061,
      "step": 1100
    },
    {
      "epoch": 0.02532974305143535,
      "grad_norm": 3.625,
      "learning_rate": 1.6882129277566542e-05,
      "loss": 1.3819,
      "step": 1110
    },
    {
      "epoch": 0.025557938934781615,
      "grad_norm": 3.484375,
      "learning_rate": 1.7034220532319392e-05,
      "loss": 1.3756,
      "step": 1120
    },
    {
      "epoch": 0.02578613481812788,
      "grad_norm": 3.59375,
      "learning_rate": 1.7186311787072246e-05,
      "loss": 1.3613,
      "step": 1130
    },
    {
      "epoch": 0.026014330701474145,
      "grad_norm": 3.46875,
      "learning_rate": 1.7338403041825096e-05,
      "loss": 1.3646,
      "step": 1140
    },
    {
      "epoch": 0.02624252658482041,
      "grad_norm": 3.609375,
      "learning_rate": 1.749049429657795e-05,
      "loss": 1.3491,
      "step": 1150
    },
    {
      "epoch": 0.026470722468166674,
      "grad_norm": 3.28125,
      "learning_rate": 1.76425855513308e-05,
      "loss": 1.3289,
      "step": 1160
    },
    {
      "epoch": 0.02669891835151294,
      "grad_norm": 3.703125,
      "learning_rate": 1.779467680608365e-05,
      "loss": 1.4166,
      "step": 1170
    },
    {
      "epoch": 0.026927114234859204,
      "grad_norm": 3.28125,
      "learning_rate": 1.7946768060836503e-05,
      "loss": 1.3441,
      "step": 1180
    },
    {
      "epoch": 0.027155310118205468,
      "grad_norm": 3.46875,
      "learning_rate": 1.8098859315589353e-05,
      "loss": 1.3552,
      "step": 1190
    },
    {
      "epoch": 0.027383506001551733,
      "grad_norm": 3.28125,
      "learning_rate": 1.8250950570342207e-05,
      "loss": 1.4087,
      "step": 1200
    },
    {
      "epoch": 0.027611701884897998,
      "grad_norm": 3.609375,
      "learning_rate": 1.8403041825095057e-05,
      "loss": 1.358,
      "step": 1210
    },
    {
      "epoch": 0.027839897768244262,
      "grad_norm": 3.34375,
      "learning_rate": 1.855513307984791e-05,
      "loss": 1.4267,
      "step": 1220
    },
    {
      "epoch": 0.028068093651590527,
      "grad_norm": 3.359375,
      "learning_rate": 1.870722433460076e-05,
      "loss": 1.4027,
      "step": 1230
    },
    {
      "epoch": 0.028296289534936788,
      "grad_norm": 3.484375,
      "learning_rate": 1.8859315589353614e-05,
      "loss": 1.4192,
      "step": 1240
    },
    {
      "epoch": 0.028524485418283053,
      "grad_norm": 3.40625,
      "learning_rate": 1.9011406844106467e-05,
      "loss": 1.348,
      "step": 1250
    },
    {
      "epoch": 0.028752681301629317,
      "grad_norm": 3.34375,
      "learning_rate": 1.9163498098859318e-05,
      "loss": 1.3163,
      "step": 1260
    },
    {
      "epoch": 0.028980877184975582,
      "grad_norm": 3.390625,
      "learning_rate": 1.931558935361217e-05,
      "loss": 1.3834,
      "step": 1270
    },
    {
      "epoch": 0.029209073068321847,
      "grad_norm": 3.390625,
      "learning_rate": 1.946768060836502e-05,
      "loss": 1.4104,
      "step": 1280
    },
    {
      "epoch": 0.02943726895166811,
      "grad_norm": 3.390625,
      "learning_rate": 1.961977186311787e-05,
      "loss": 1.3416,
      "step": 1290
    },
    {
      "epoch": 0.029665464835014376,
      "grad_norm": 3.296875,
      "learning_rate": 1.9771863117870725e-05,
      "loss": 1.3557,
      "step": 1300
    },
    {
      "epoch": 0.02989366071836064,
      "grad_norm": 3.5,
      "learning_rate": 1.9923954372623575e-05,
      "loss": 1.4189,
      "step": 1310
    },
    {
      "epoch": 0.030121856601706905,
      "grad_norm": 3.21875,
      "learning_rate": 1.9997647446302963e-05,
      "loss": 1.384,
      "step": 1320
    },
    {
      "epoch": 0.03035005248505317,
      "grad_norm": 3.265625,
      "learning_rate": 1.9992942338908888e-05,
      "loss": 1.3405,
      "step": 1330
    },
    {
      "epoch": 0.030578248368399435,
      "grad_norm": 3.375,
      "learning_rate": 1.998823723151481e-05,
      "loss": 1.345,
      "step": 1340
    },
    {
      "epoch": 0.0308064442517457,
      "grad_norm": 3.15625,
      "learning_rate": 1.9983532124120735e-05,
      "loss": 1.3545,
      "step": 1350
    },
    {
      "epoch": 0.031034640135091964,
      "grad_norm": 3.484375,
      "learning_rate": 1.9978827016726657e-05,
      "loss": 1.4016,
      "step": 1360
    },
    {
      "epoch": 0.031262836018438225,
      "grad_norm": 3.0625,
      "learning_rate": 1.9974121909332583e-05,
      "loss": 1.3485,
      "step": 1370
    },
    {
      "epoch": 0.03149103190178449,
      "grad_norm": 3.203125,
      "learning_rate": 1.9969416801938505e-05,
      "loss": 1.3508,
      "step": 1380
    },
    {
      "epoch": 0.031719227785130755,
      "grad_norm": 3.296875,
      "learning_rate": 1.996471169454443e-05,
      "loss": 1.3914,
      "step": 1390
    },
    {
      "epoch": 0.03194742366847702,
      "grad_norm": 3.5,
      "learning_rate": 1.9960006587150352e-05,
      "loss": 1.3361,
      "step": 1400
    },
    {
      "epoch": 0.032175619551823284,
      "grad_norm": 3.078125,
      "learning_rate": 1.9955301479756278e-05,
      "loss": 1.3417,
      "step": 1410
    },
    {
      "epoch": 0.03240381543516955,
      "grad_norm": 3.0625,
      "learning_rate": 1.99505963723622e-05,
      "loss": 1.345,
      "step": 1420
    },
    {
      "epoch": 0.03263201131851581,
      "grad_norm": 2.984375,
      "learning_rate": 1.9945891264968125e-05,
      "loss": 1.3148,
      "step": 1430
    },
    {
      "epoch": 0.03286020720186208,
      "grad_norm": 3.09375,
      "learning_rate": 1.9941186157574047e-05,
      "loss": 1.3093,
      "step": 1440
    },
    {
      "epoch": 0.03308840308520834,
      "grad_norm": 3.0625,
      "learning_rate": 1.9936481050179973e-05,
      "loss": 1.343,
      "step": 1450
    },
    {
      "epoch": 0.03331659896855461,
      "grad_norm": 3.15625,
      "learning_rate": 1.9931775942785895e-05,
      "loss": 1.3747,
      "step": 1460
    },
    {
      "epoch": 0.03354479485190087,
      "grad_norm": 3.09375,
      "learning_rate": 1.992707083539182e-05,
      "loss": 1.3472,
      "step": 1470
    },
    {
      "epoch": 0.03377299073524714,
      "grad_norm": 3.296875,
      "learning_rate": 1.9922365727997742e-05,
      "loss": 1.3915,
      "step": 1480
    },
    {
      "epoch": 0.0340011866185934,
      "grad_norm": 3.265625,
      "learning_rate": 1.9917660620603667e-05,
      "loss": 1.3595,
      "step": 1490
    },
    {
      "epoch": 0.034229382501939666,
      "grad_norm": 3.484375,
      "learning_rate": 1.991295551320959e-05,
      "loss": 1.3305,
      "step": 1500
    },
    {
      "epoch": 0.03445757838528593,
      "grad_norm": 3.296875,
      "learning_rate": 1.9908250405815515e-05,
      "loss": 1.3188,
      "step": 1510
    },
    {
      "epoch": 0.034685774268632195,
      "grad_norm": 3.375,
      "learning_rate": 1.9903545298421437e-05,
      "loss": 1.3128,
      "step": 1520
    },
    {
      "epoch": 0.03491397015197846,
      "grad_norm": 3.078125,
      "learning_rate": 1.9898840191027362e-05,
      "loss": 1.3442,
      "step": 1530
    },
    {
      "epoch": 0.035142166035324725,
      "grad_norm": 3.109375,
      "learning_rate": 1.9894135083633284e-05,
      "loss": 1.3796,
      "step": 1540
    },
    {
      "epoch": 0.03537036191867099,
      "grad_norm": 3.171875,
      "learning_rate": 1.988942997623921e-05,
      "loss": 1.3631,
      "step": 1550
    },
    {
      "epoch": 0.035598557802017254,
      "grad_norm": 3.28125,
      "learning_rate": 1.9884724868845132e-05,
      "loss": 1.3231,
      "step": 1560
    },
    {
      "epoch": 0.03582675368536352,
      "grad_norm": 3.203125,
      "learning_rate": 1.9880019761451057e-05,
      "loss": 1.3384,
      "step": 1570
    },
    {
      "epoch": 0.03605494956870978,
      "grad_norm": 3.25,
      "learning_rate": 1.987531465405698e-05,
      "loss": 1.3778,
      "step": 1580
    },
    {
      "epoch": 0.03628314545205605,
      "grad_norm": 3.34375,
      "learning_rate": 1.9870609546662905e-05,
      "loss": 1.2998,
      "step": 1590
    },
    {
      "epoch": 0.03651134133540231,
      "grad_norm": 3.234375,
      "learning_rate": 1.9865904439268827e-05,
      "loss": 1.3991,
      "step": 1600
    },
    {
      "epoch": 0.03673953721874857,
      "grad_norm": 3.546875,
      "learning_rate": 1.9861199331874752e-05,
      "loss": 1.2882,
      "step": 1610
    },
    {
      "epoch": 0.036967733102094835,
      "grad_norm": 3.578125,
      "learning_rate": 1.9856494224480677e-05,
      "loss": 1.3104,
      "step": 1620
    },
    {
      "epoch": 0.0371959289854411,
      "grad_norm": 3.21875,
      "learning_rate": 1.98517891170866e-05,
      "loss": 1.3528,
      "step": 1630
    },
    {
      "epoch": 0.037424124868787365,
      "grad_norm": 3.328125,
      "learning_rate": 1.9847084009692525e-05,
      "loss": 1.3201,
      "step": 1640
    },
    {
      "epoch": 0.03765232075213363,
      "grad_norm": 3.15625,
      "learning_rate": 1.9842378902298447e-05,
      "loss": 1.3611,
      "step": 1650
    },
    {
      "epoch": 0.037880516635479894,
      "grad_norm": 3.296875,
      "learning_rate": 1.9837673794904372e-05,
      "loss": 1.3765,
      "step": 1660
    },
    {
      "epoch": 0.03810871251882616,
      "grad_norm": 3.34375,
      "learning_rate": 1.9832968687510294e-05,
      "loss": 1.3333,
      "step": 1670
    },
    {
      "epoch": 0.03833690840217242,
      "grad_norm": 3.609375,
      "learning_rate": 1.982826358011622e-05,
      "loss": 1.4382,
      "step": 1680
    },
    {
      "epoch": 0.03856510428551869,
      "grad_norm": 3.1875,
      "learning_rate": 1.9823558472722142e-05,
      "loss": 1.2967,
      "step": 1690
    },
    {
      "epoch": 0.03879330016886495,
      "grad_norm": 3.109375,
      "learning_rate": 1.9818853365328067e-05,
      "loss": 1.3432,
      "step": 1700
    },
    {
      "epoch": 0.03902149605221122,
      "grad_norm": 2.921875,
      "learning_rate": 1.981414825793399e-05,
      "loss": 1.3983,
      "step": 1710
    },
    {
      "epoch": 0.03924969193555748,
      "grad_norm": 3.28125,
      "learning_rate": 1.9809443150539915e-05,
      "loss": 1.3312,
      "step": 1720
    },
    {
      "epoch": 0.03947788781890375,
      "grad_norm": 3.140625,
      "learning_rate": 1.9804738043145837e-05,
      "loss": 1.3478,
      "step": 1730
    },
    {
      "epoch": 0.03970608370225001,
      "grad_norm": 3.03125,
      "learning_rate": 1.9800032935751762e-05,
      "loss": 1.3477,
      "step": 1740
    },
    {
      "epoch": 0.039934279585596276,
      "grad_norm": 3.125,
      "learning_rate": 1.9795327828357684e-05,
      "loss": 1.3439,
      "step": 1750
    },
    {
      "epoch": 0.04016247546894254,
      "grad_norm": 3.046875,
      "learning_rate": 1.979062272096361e-05,
      "loss": 1.3563,
      "step": 1760
    },
    {
      "epoch": 0.040390671352288805,
      "grad_norm": 3.375,
      "learning_rate": 1.978591761356953e-05,
      "loss": 1.4,
      "step": 1770
    },
    {
      "epoch": 0.04061886723563507,
      "grad_norm": 3.25,
      "learning_rate": 1.9781212506175457e-05,
      "loss": 1.3364,
      "step": 1780
    },
    {
      "epoch": 0.040847063118981335,
      "grad_norm": 3.203125,
      "learning_rate": 1.977650739878138e-05,
      "loss": 1.3672,
      "step": 1790
    },
    {
      "epoch": 0.0410752590023276,
      "grad_norm": 3.03125,
      "learning_rate": 1.9771802291387304e-05,
      "loss": 1.3825,
      "step": 1800
    },
    {
      "epoch": 0.041303454885673864,
      "grad_norm": 3.203125,
      "learning_rate": 1.9767097183993226e-05,
      "loss": 1.3331,
      "step": 1810
    },
    {
      "epoch": 0.04153165076902013,
      "grad_norm": 2.921875,
      "learning_rate": 1.976239207659915e-05,
      "loss": 1.3049,
      "step": 1820
    },
    {
      "epoch": 0.04175984665236639,
      "grad_norm": 2.875,
      "learning_rate": 1.9757686969205074e-05,
      "loss": 1.2942,
      "step": 1830
    },
    {
      "epoch": 0.04198804253571266,
      "grad_norm": 3.140625,
      "learning_rate": 1.9752981861811e-05,
      "loss": 1.3519,
      "step": 1840
    },
    {
      "epoch": 0.04221623841905892,
      "grad_norm": 3.28125,
      "learning_rate": 1.974827675441692e-05,
      "loss": 1.3123,
      "step": 1850
    },
    {
      "epoch": 0.04244443430240519,
      "grad_norm": 3.15625,
      "learning_rate": 1.9743571647022847e-05,
      "loss": 1.3317,
      "step": 1860
    },
    {
      "epoch": 0.04267263018575145,
      "grad_norm": 3.296875,
      "learning_rate": 1.973886653962877e-05,
      "loss": 1.3286,
      "step": 1870
    },
    {
      "epoch": 0.04290082606909772,
      "grad_norm": 3.1875,
      "learning_rate": 1.9734161432234694e-05,
      "loss": 1.2829,
      "step": 1880
    },
    {
      "epoch": 0.04312902195244398,
      "grad_norm": 3.234375,
      "learning_rate": 1.9729456324840616e-05,
      "loss": 1.3229,
      "step": 1890
    },
    {
      "epoch": 0.04335721783579024,
      "grad_norm": 3.0,
      "learning_rate": 1.972475121744654e-05,
      "loss": 1.3511,
      "step": 1900
    },
    {
      "epoch": 0.043585413719136504,
      "grad_norm": 3.234375,
      "learning_rate": 1.9720046110052463e-05,
      "loss": 1.3444,
      "step": 1910
    },
    {
      "epoch": 0.04381360960248277,
      "grad_norm": 3.109375,
      "learning_rate": 1.971534100265839e-05,
      "loss": 1.3569,
      "step": 1920
    },
    {
      "epoch": 0.04404180548582903,
      "grad_norm": 3.3125,
      "learning_rate": 1.971063589526431e-05,
      "loss": 1.3211,
      "step": 1930
    },
    {
      "epoch": 0.0442700013691753,
      "grad_norm": 3.46875,
      "learning_rate": 1.9705930787870236e-05,
      "loss": 1.3437,
      "step": 1940
    },
    {
      "epoch": 0.04449819725252156,
      "grad_norm": 2.78125,
      "learning_rate": 1.9701225680476158e-05,
      "loss": 1.313,
      "step": 1950
    },
    {
      "epoch": 0.04472639313586783,
      "grad_norm": 2.859375,
      "learning_rate": 1.9696520573082084e-05,
      "loss": 1.3624,
      "step": 1960
    },
    {
      "epoch": 0.04495458901921409,
      "grad_norm": 2.9375,
      "learning_rate": 1.9691815465688006e-05,
      "loss": 1.3802,
      "step": 1970
    },
    {
      "epoch": 0.045182784902560356,
      "grad_norm": 3.015625,
      "learning_rate": 1.968711035829393e-05,
      "loss": 1.3185,
      "step": 1980
    },
    {
      "epoch": 0.04541098078590662,
      "grad_norm": 3.0625,
      "learning_rate": 1.9682405250899853e-05,
      "loss": 1.3172,
      "step": 1990
    },
    {
      "epoch": 0.045639176669252886,
      "grad_norm": 3.21875,
      "learning_rate": 1.967770014350578e-05,
      "loss": 1.3595,
      "step": 2000
    },
    {
      "epoch": 0.04586737255259915,
      "grad_norm": 3.1875,
      "learning_rate": 1.96729950361117e-05,
      "loss": 1.3756,
      "step": 2010
    },
    {
      "epoch": 0.046095568435945415,
      "grad_norm": 3.015625,
      "learning_rate": 1.9668289928717626e-05,
      "loss": 1.3744,
      "step": 2020
    },
    {
      "epoch": 0.04632376431929168,
      "grad_norm": 3.4375,
      "learning_rate": 1.9663584821323548e-05,
      "loss": 1.3532,
      "step": 2030
    },
    {
      "epoch": 0.046551960202637944,
      "grad_norm": 3.078125,
      "learning_rate": 1.9658879713929473e-05,
      "loss": 1.3375,
      "step": 2040
    },
    {
      "epoch": 0.04678015608598421,
      "grad_norm": 3.15625,
      "learning_rate": 1.9654174606535395e-05,
      "loss": 1.3547,
      "step": 2050
    },
    {
      "epoch": 0.047008351969330474,
      "grad_norm": 3.0625,
      "learning_rate": 1.964946949914132e-05,
      "loss": 1.3437,
      "step": 2060
    },
    {
      "epoch": 0.04723654785267674,
      "grad_norm": 2.90625,
      "learning_rate": 1.9644764391747243e-05,
      "loss": 1.3712,
      "step": 2070
    },
    {
      "epoch": 0.047464743736023,
      "grad_norm": 2.734375,
      "learning_rate": 1.9640059284353168e-05,
      "loss": 1.3295,
      "step": 2080
    },
    {
      "epoch": 0.04769293961936927,
      "grad_norm": 3.09375,
      "learning_rate": 1.963535417695909e-05,
      "loss": 1.2965,
      "step": 2090
    },
    {
      "epoch": 0.04792113550271553,
      "grad_norm": 2.78125,
      "learning_rate": 1.9630649069565016e-05,
      "loss": 1.3355,
      "step": 2100
    },
    {
      "epoch": 0.0481493313860618,
      "grad_norm": 3.03125,
      "learning_rate": 1.9625943962170938e-05,
      "loss": 1.2739,
      "step": 2110
    },
    {
      "epoch": 0.04837752726940806,
      "grad_norm": 2.921875,
      "learning_rate": 1.9621238854776863e-05,
      "loss": 1.3527,
      "step": 2120
    },
    {
      "epoch": 0.04860572315275433,
      "grad_norm": 3.1875,
      "learning_rate": 1.9616533747382785e-05,
      "loss": 1.3844,
      "step": 2130
    },
    {
      "epoch": 0.04883391903610059,
      "grad_norm": 3.390625,
      "learning_rate": 1.961182863998871e-05,
      "loss": 1.266,
      "step": 2140
    },
    {
      "epoch": 0.049062114919446856,
      "grad_norm": 3.015625,
      "learning_rate": 1.9607123532594632e-05,
      "loss": 1.289,
      "step": 2150
    },
    {
      "epoch": 0.04929031080279312,
      "grad_norm": 2.921875,
      "learning_rate": 1.9602418425200558e-05,
      "loss": 1.3261,
      "step": 2160
    },
    {
      "epoch": 0.049518506686139385,
      "grad_norm": 3.09375,
      "learning_rate": 1.959771331780648e-05,
      "loss": 1.3243,
      "step": 2170
    },
    {
      "epoch": 0.04974670256948565,
      "grad_norm": 3.109375,
      "learning_rate": 1.9593008210412405e-05,
      "loss": 1.3062,
      "step": 2180
    },
    {
      "epoch": 0.04997489845283191,
      "grad_norm": 3.015625,
      "learning_rate": 1.958830310301833e-05,
      "loss": 1.3586,
      "step": 2190
    },
    {
      "epoch": 0.05020309433617817,
      "grad_norm": 3.15625,
      "learning_rate": 1.9583597995624253e-05,
      "loss": 1.2829,
      "step": 2200
    },
    {
      "epoch": 0.05043129021952444,
      "grad_norm": 3.046875,
      "learning_rate": 1.9578892888230178e-05,
      "loss": 1.3036,
      "step": 2210
    },
    {
      "epoch": 0.0506594861028707,
      "grad_norm": 3.03125,
      "learning_rate": 1.95741877808361e-05,
      "loss": 1.2972,
      "step": 2220
    },
    {
      "epoch": 0.050887681986216966,
      "grad_norm": 3.0625,
      "learning_rate": 1.9569482673442026e-05,
      "loss": 1.3014,
      "step": 2230
    },
    {
      "epoch": 0.05111587786956323,
      "grad_norm": 3.109375,
      "learning_rate": 1.9564777566047944e-05,
      "loss": 1.2874,
      "step": 2240
    },
    {
      "epoch": 0.051344073752909496,
      "grad_norm": 2.96875,
      "learning_rate": 1.956007245865387e-05,
      "loss": 1.2832,
      "step": 2250
    },
    {
      "epoch": 0.05157226963625576,
      "grad_norm": 3.09375,
      "learning_rate": 1.955536735125979e-05,
      "loss": 1.2873,
      "step": 2260
    },
    {
      "epoch": 0.051800465519602025,
      "grad_norm": 3.375,
      "learning_rate": 1.9550662243865717e-05,
      "loss": 1.3528,
      "step": 2270
    },
    {
      "epoch": 0.05202866140294829,
      "grad_norm": 3.015625,
      "learning_rate": 1.954595713647164e-05,
      "loss": 1.3258,
      "step": 2280
    },
    {
      "epoch": 0.052256857286294554,
      "grad_norm": 3.484375,
      "learning_rate": 1.9541252029077564e-05,
      "loss": 1.333,
      "step": 2290
    },
    {
      "epoch": 0.05248505316964082,
      "grad_norm": 3.0,
      "learning_rate": 1.9536546921683486e-05,
      "loss": 1.3283,
      "step": 2300
    },
    {
      "epoch": 0.052713249052987084,
      "grad_norm": 2.90625,
      "learning_rate": 1.9531841814289412e-05,
      "loss": 1.3391,
      "step": 2310
    },
    {
      "epoch": 0.05294144493633335,
      "grad_norm": 3.453125,
      "learning_rate": 1.9527136706895334e-05,
      "loss": 1.3023,
      "step": 2320
    },
    {
      "epoch": 0.05316964081967961,
      "grad_norm": 2.828125,
      "learning_rate": 1.952243159950126e-05,
      "loss": 1.2578,
      "step": 2330
    },
    {
      "epoch": 0.05339783670302588,
      "grad_norm": 3.359375,
      "learning_rate": 1.951772649210718e-05,
      "loss": 1.3337,
      "step": 2340
    },
    {
      "epoch": 0.05362603258637214,
      "grad_norm": 3.015625,
      "learning_rate": 1.9513021384713107e-05,
      "loss": 1.3553,
      "step": 2350
    },
    {
      "epoch": 0.05385422846971841,
      "grad_norm": 3.046875,
      "learning_rate": 1.9508316277319032e-05,
      "loss": 1.327,
      "step": 2360
    },
    {
      "epoch": 0.05408242435306467,
      "grad_norm": 3.140625,
      "learning_rate": 1.9503611169924954e-05,
      "loss": 1.335,
      "step": 2370
    },
    {
      "epoch": 0.054310620236410936,
      "grad_norm": 3.09375,
      "learning_rate": 1.949890606253088e-05,
      "loss": 1.3125,
      "step": 2380
    },
    {
      "epoch": 0.0545388161197572,
      "grad_norm": 2.828125,
      "learning_rate": 1.94942009551368e-05,
      "loss": 1.3499,
      "step": 2390
    },
    {
      "epoch": 0.054767012003103466,
      "grad_norm": 2.90625,
      "learning_rate": 1.9489495847742727e-05,
      "loss": 1.3259,
      "step": 2400
    },
    {
      "epoch": 0.05499520788644973,
      "grad_norm": 2.921875,
      "learning_rate": 1.948479074034865e-05,
      "loss": 1.2998,
      "step": 2410
    },
    {
      "epoch": 0.055223403769795995,
      "grad_norm": 2.96875,
      "learning_rate": 1.9480085632954574e-05,
      "loss": 1.3369,
      "step": 2420
    },
    {
      "epoch": 0.05545159965314226,
      "grad_norm": 2.78125,
      "learning_rate": 1.9475380525560496e-05,
      "loss": 1.3212,
      "step": 2430
    },
    {
      "epoch": 0.055679795536488524,
      "grad_norm": 3.046875,
      "learning_rate": 1.9470675418166422e-05,
      "loss": 1.3363,
      "step": 2440
    },
    {
      "epoch": 0.05590799141983479,
      "grad_norm": 2.90625,
      "learning_rate": 1.9465970310772344e-05,
      "loss": 1.2463,
      "step": 2450
    },
    {
      "epoch": 0.056136187303181054,
      "grad_norm": 2.984375,
      "learning_rate": 1.946126520337827e-05,
      "loss": 1.262,
      "step": 2460
    },
    {
      "epoch": 0.05636438318652732,
      "grad_norm": 2.78125,
      "learning_rate": 1.945656009598419e-05,
      "loss": 1.2973,
      "step": 2470
    },
    {
      "epoch": 0.056592579069873576,
      "grad_norm": 3.25,
      "learning_rate": 1.9451854988590117e-05,
      "loss": 1.2681,
      "step": 2480
    },
    {
      "epoch": 0.05682077495321984,
      "grad_norm": 2.984375,
      "learning_rate": 1.944714988119604e-05,
      "loss": 1.348,
      "step": 2490
    },
    {
      "epoch": 0.057048970836566106,
      "grad_norm": 2.921875,
      "learning_rate": 1.9442444773801964e-05,
      "loss": 1.3223,
      "step": 2500
    },
    {
      "epoch": 0.05727716671991237,
      "grad_norm": 2.875,
      "learning_rate": 1.9437739666407886e-05,
      "loss": 1.3331,
      "step": 2510
    },
    {
      "epoch": 0.057505362603258635,
      "grad_norm": 3.0625,
      "learning_rate": 1.943303455901381e-05,
      "loss": 1.3448,
      "step": 2520
    },
    {
      "epoch": 0.0577335584866049,
      "grad_norm": 2.828125,
      "learning_rate": 1.9428329451619734e-05,
      "loss": 1.2804,
      "step": 2530
    },
    {
      "epoch": 0.057961754369951164,
      "grad_norm": 2.96875,
      "learning_rate": 1.942362434422566e-05,
      "loss": 1.3091,
      "step": 2540
    },
    {
      "epoch": 0.05818995025329743,
      "grad_norm": 3.015625,
      "learning_rate": 1.941891923683158e-05,
      "loss": 1.298,
      "step": 2550
    },
    {
      "epoch": 0.058418146136643694,
      "grad_norm": 3.0,
      "learning_rate": 1.9414214129437506e-05,
      "loss": 1.3311,
      "step": 2560
    },
    {
      "epoch": 0.05864634201998996,
      "grad_norm": 2.9375,
      "learning_rate": 1.940950902204343e-05,
      "loss": 1.3316,
      "step": 2570
    },
    {
      "epoch": 0.05887453790333622,
      "grad_norm": 2.921875,
      "learning_rate": 1.9404803914649354e-05,
      "loss": 1.2494,
      "step": 2580
    },
    {
      "epoch": 0.05910273378668249,
      "grad_norm": 2.96875,
      "learning_rate": 1.9400098807255276e-05,
      "loss": 1.3575,
      "step": 2590
    },
    {
      "epoch": 0.05933092967002875,
      "grad_norm": 2.90625,
      "learning_rate": 1.93953936998612e-05,
      "loss": 1.3136,
      "step": 2600
    },
    {
      "epoch": 0.05955912555337502,
      "grad_norm": 2.96875,
      "learning_rate": 1.9390688592467123e-05,
      "loss": 1.2939,
      "step": 2610
    },
    {
      "epoch": 0.05978732143672128,
      "grad_norm": 3.3125,
      "learning_rate": 1.938598348507305e-05,
      "loss": 1.2837,
      "step": 2620
    },
    {
      "epoch": 0.060015517320067546,
      "grad_norm": 2.796875,
      "learning_rate": 1.938127837767897e-05,
      "loss": 1.3491,
      "step": 2630
    },
    {
      "epoch": 0.06024371320341381,
      "grad_norm": 3.0,
      "learning_rate": 1.9376573270284896e-05,
      "loss": 1.3313,
      "step": 2640
    },
    {
      "epoch": 0.060471909086760076,
      "grad_norm": 3.046875,
      "learning_rate": 1.9371868162890818e-05,
      "loss": 1.2966,
      "step": 2650
    },
    {
      "epoch": 0.06070010497010634,
      "grad_norm": 2.96875,
      "learning_rate": 1.9367163055496743e-05,
      "loss": 1.3495,
      "step": 2660
    },
    {
      "epoch": 0.060928300853452605,
      "grad_norm": 3.0625,
      "learning_rate": 1.9362457948102666e-05,
      "loss": 1.3175,
      "step": 2670
    },
    {
      "epoch": 0.06115649673679887,
      "grad_norm": 2.90625,
      "learning_rate": 1.935775284070859e-05,
      "loss": 1.3328,
      "step": 2680
    },
    {
      "epoch": 0.061384692620145134,
      "grad_norm": 3.140625,
      "learning_rate": 1.9353047733314513e-05,
      "loss": 1.3052,
      "step": 2690
    },
    {
      "epoch": 0.0616128885034914,
      "grad_norm": 3.171875,
      "learning_rate": 1.934834262592044e-05,
      "loss": 1.3565,
      "step": 2700
    },
    {
      "epoch": 0.061841084386837664,
      "grad_norm": 2.921875,
      "learning_rate": 1.934363751852636e-05,
      "loss": 1.2375,
      "step": 2710
    },
    {
      "epoch": 0.06206928027018393,
      "grad_norm": 3.1875,
      "learning_rate": 1.9338932411132286e-05,
      "loss": 1.3403,
      "step": 2720
    },
    {
      "epoch": 0.06229747615353019,
      "grad_norm": 3.0,
      "learning_rate": 1.9334227303738208e-05,
      "loss": 1.2978,
      "step": 2730
    },
    {
      "epoch": 0.06252567203687645,
      "grad_norm": 2.796875,
      "learning_rate": 1.9329522196344133e-05,
      "loss": 1.2961,
      "step": 2740
    },
    {
      "epoch": 0.06275386792022272,
      "grad_norm": 3.0,
      "learning_rate": 1.9324817088950055e-05,
      "loss": 1.2535,
      "step": 2750
    },
    {
      "epoch": 0.06298206380356898,
      "grad_norm": 3.0625,
      "learning_rate": 1.932011198155598e-05,
      "loss": 1.3331,
      "step": 2760
    },
    {
      "epoch": 0.06321025968691525,
      "grad_norm": 2.875,
      "learning_rate": 1.9315406874161903e-05,
      "loss": 1.3123,
      "step": 2770
    },
    {
      "epoch": 0.06343845557026151,
      "grad_norm": 3.109375,
      "learning_rate": 1.9310701766767828e-05,
      "loss": 1.3033,
      "step": 2780
    },
    {
      "epoch": 0.06366665145360778,
      "grad_norm": 3.09375,
      "learning_rate": 1.930599665937375e-05,
      "loss": 1.2968,
      "step": 2790
    },
    {
      "epoch": 0.06389484733695404,
      "grad_norm": 3.015625,
      "learning_rate": 1.9301291551979675e-05,
      "loss": 1.2564,
      "step": 2800
    },
    {
      "epoch": 0.06412304322030031,
      "grad_norm": 2.890625,
      "learning_rate": 1.9296586444585597e-05,
      "loss": 1.3422,
      "step": 2810
    },
    {
      "epoch": 0.06435123910364657,
      "grad_norm": 3.28125,
      "learning_rate": 1.9291881337191523e-05,
      "loss": 1.3109,
      "step": 2820
    },
    {
      "epoch": 0.06457943498699284,
      "grad_norm": 3.03125,
      "learning_rate": 1.9287176229797445e-05,
      "loss": 1.3506,
      "step": 2830
    },
    {
      "epoch": 0.0648076308703391,
      "grad_norm": 2.984375,
      "learning_rate": 1.928247112240337e-05,
      "loss": 1.3651,
      "step": 2840
    },
    {
      "epoch": 0.06503582675368537,
      "grad_norm": 3.03125,
      "learning_rate": 1.9277766015009292e-05,
      "loss": 1.2158,
      "step": 2850
    },
    {
      "epoch": 0.06526402263703163,
      "grad_norm": 3.125,
      "learning_rate": 1.9273060907615218e-05,
      "loss": 1.3632,
      "step": 2860
    },
    {
      "epoch": 0.0654922185203779,
      "grad_norm": 3.09375,
      "learning_rate": 1.926835580022114e-05,
      "loss": 1.2916,
      "step": 2870
    },
    {
      "epoch": 0.06572041440372416,
      "grad_norm": 3.15625,
      "learning_rate": 1.9263650692827065e-05,
      "loss": 1.2579,
      "step": 2880
    },
    {
      "epoch": 0.06594861028707043,
      "grad_norm": 3.0625,
      "learning_rate": 1.9258945585432987e-05,
      "loss": 1.3358,
      "step": 2890
    },
    {
      "epoch": 0.06617680617041669,
      "grad_norm": 2.875,
      "learning_rate": 1.9254240478038913e-05,
      "loss": 1.296,
      "step": 2900
    },
    {
      "epoch": 0.06640500205376294,
      "grad_norm": 3.109375,
      "learning_rate": 1.9249535370644835e-05,
      "loss": 1.3481,
      "step": 2910
    },
    {
      "epoch": 0.06663319793710921,
      "grad_norm": 3.046875,
      "learning_rate": 1.924483026325076e-05,
      "loss": 1.3228,
      "step": 2920
    },
    {
      "epoch": 0.06686139382045547,
      "grad_norm": 2.78125,
      "learning_rate": 1.9240125155856682e-05,
      "loss": 1.315,
      "step": 2930
    },
    {
      "epoch": 0.06708958970380174,
      "grad_norm": 3.140625,
      "learning_rate": 1.9235420048462607e-05,
      "loss": 1.2657,
      "step": 2940
    },
    {
      "epoch": 0.067317785587148,
      "grad_norm": 2.9375,
      "learning_rate": 1.9230714941068533e-05,
      "loss": 1.2938,
      "step": 2950
    },
    {
      "epoch": 0.06754598147049427,
      "grad_norm": 2.859375,
      "learning_rate": 1.9226009833674455e-05,
      "loss": 1.2649,
      "step": 2960
    },
    {
      "epoch": 0.06777417735384053,
      "grad_norm": 3.078125,
      "learning_rate": 1.922130472628038e-05,
      "loss": 1.3182,
      "step": 2970
    },
    {
      "epoch": 0.0680023732371868,
      "grad_norm": 3.015625,
      "learning_rate": 1.9216599618886302e-05,
      "loss": 1.2931,
      "step": 2980
    },
    {
      "epoch": 0.06823056912053306,
      "grad_norm": 2.875,
      "learning_rate": 1.9211894511492228e-05,
      "loss": 1.2929,
      "step": 2990
    },
    {
      "epoch": 0.06845876500387933,
      "grad_norm": 2.90625,
      "learning_rate": 1.920718940409815e-05,
      "loss": 1.3254,
      "step": 3000
    },
    {
      "epoch": 0.06868696088722559,
      "grad_norm": 3.09375,
      "learning_rate": 1.9202484296704075e-05,
      "loss": 1.295,
      "step": 3010
    },
    {
      "epoch": 0.06891515677057186,
      "grad_norm": 2.96875,
      "learning_rate": 1.9197779189309997e-05,
      "loss": 1.2974,
      "step": 3020
    },
    {
      "epoch": 0.06914335265391812,
      "grad_norm": 2.765625,
      "learning_rate": 1.9193074081915923e-05,
      "loss": 1.3092,
      "step": 3030
    },
    {
      "epoch": 0.06937154853726439,
      "grad_norm": 2.828125,
      "learning_rate": 1.9188368974521845e-05,
      "loss": 1.2749,
      "step": 3040
    },
    {
      "epoch": 0.06959974442061065,
      "grad_norm": 2.984375,
      "learning_rate": 1.918366386712777e-05,
      "loss": 1.3363,
      "step": 3050
    },
    {
      "epoch": 0.06982794030395692,
      "grad_norm": 3.375,
      "learning_rate": 1.9178958759733692e-05,
      "loss": 1.2407,
      "step": 3060
    },
    {
      "epoch": 0.07005613618730318,
      "grad_norm": 3.3125,
      "learning_rate": 1.9174253652339617e-05,
      "loss": 1.2827,
      "step": 3070
    },
    {
      "epoch": 0.07028433207064945,
      "grad_norm": 3.265625,
      "learning_rate": 1.916954854494554e-05,
      "loss": 1.3172,
      "step": 3080
    },
    {
      "epoch": 0.07051252795399571,
      "grad_norm": 3.0,
      "learning_rate": 1.9164843437551465e-05,
      "loss": 1.3089,
      "step": 3090
    },
    {
      "epoch": 0.07074072383734198,
      "grad_norm": 3.34375,
      "learning_rate": 1.9160138330157387e-05,
      "loss": 1.2981,
      "step": 3100
    },
    {
      "epoch": 0.07096891972068824,
      "grad_norm": 2.703125,
      "learning_rate": 1.9155433222763312e-05,
      "loss": 1.2745,
      "step": 3110
    },
    {
      "epoch": 0.07119711560403451,
      "grad_norm": 2.8125,
      "learning_rate": 1.9150728115369234e-05,
      "loss": 1.2705,
      "step": 3120
    },
    {
      "epoch": 0.07142531148738077,
      "grad_norm": 3.125,
      "learning_rate": 1.914602300797516e-05,
      "loss": 1.2858,
      "step": 3130
    },
    {
      "epoch": 0.07165350737072704,
      "grad_norm": 2.984375,
      "learning_rate": 1.914131790058108e-05,
      "loss": 1.2985,
      "step": 3140
    },
    {
      "epoch": 0.0718817032540733,
      "grad_norm": 3.265625,
      "learning_rate": 1.9136612793187007e-05,
      "loss": 1.3154,
      "step": 3150
    },
    {
      "epoch": 0.07210989913741957,
      "grad_norm": 3.578125,
      "learning_rate": 1.913190768579293e-05,
      "loss": 1.2897,
      "step": 3160
    },
    {
      "epoch": 0.07233809502076582,
      "grad_norm": 3.015625,
      "learning_rate": 1.9127202578398855e-05,
      "loss": 1.2612,
      "step": 3170
    },
    {
      "epoch": 0.0725662909041121,
      "grad_norm": 2.90625,
      "learning_rate": 1.9122497471004777e-05,
      "loss": 1.2448,
      "step": 3180
    },
    {
      "epoch": 0.07279448678745835,
      "grad_norm": 2.96875,
      "learning_rate": 1.9117792363610702e-05,
      "loss": 1.3043,
      "step": 3190
    },
    {
      "epoch": 0.07302268267080463,
      "grad_norm": 2.875,
      "learning_rate": 1.9113087256216624e-05,
      "loss": 1.3078,
      "step": 3200
    },
    {
      "epoch": 0.07325087855415088,
      "grad_norm": 3.078125,
      "learning_rate": 1.910838214882255e-05,
      "loss": 1.2646,
      "step": 3210
    },
    {
      "epoch": 0.07347907443749714,
      "grad_norm": 3.015625,
      "learning_rate": 1.910367704142847e-05,
      "loss": 1.2309,
      "step": 3220
    },
    {
      "epoch": 0.07370727032084341,
      "grad_norm": 3.109375,
      "learning_rate": 1.9098971934034397e-05,
      "loss": 1.2709,
      "step": 3230
    },
    {
      "epoch": 0.07393546620418967,
      "grad_norm": 3.1875,
      "learning_rate": 1.909426682664032e-05,
      "loss": 1.2789,
      "step": 3240
    },
    {
      "epoch": 0.07416366208753594,
      "grad_norm": 2.765625,
      "learning_rate": 1.9089561719246244e-05,
      "loss": 1.2751,
      "step": 3250
    },
    {
      "epoch": 0.0743918579708822,
      "grad_norm": 3.71875,
      "learning_rate": 1.9084856611852166e-05,
      "loss": 1.2787,
      "step": 3260
    },
    {
      "epoch": 0.07462005385422847,
      "grad_norm": 3.015625,
      "learning_rate": 1.908015150445809e-05,
      "loss": 1.2515,
      "step": 3270
    },
    {
      "epoch": 0.07484824973757473,
      "grad_norm": 2.859375,
      "learning_rate": 1.9075446397064014e-05,
      "loss": 1.2512,
      "step": 3280
    },
    {
      "epoch": 0.075076445620921,
      "grad_norm": 2.921875,
      "learning_rate": 1.907074128966994e-05,
      "loss": 1.3457,
      "step": 3290
    },
    {
      "epoch": 0.07530464150426726,
      "grad_norm": 3.25,
      "learning_rate": 1.906603618227586e-05,
      "loss": 1.3079,
      "step": 3300
    },
    {
      "epoch": 0.07553283738761353,
      "grad_norm": 3.0625,
      "learning_rate": 1.9061331074881786e-05,
      "loss": 1.2903,
      "step": 3310
    },
    {
      "epoch": 0.07576103327095979,
      "grad_norm": 2.71875,
      "learning_rate": 1.905662596748771e-05,
      "loss": 1.2931,
      "step": 3320
    },
    {
      "epoch": 0.07598922915430606,
      "grad_norm": 3.015625,
      "learning_rate": 1.9051920860093634e-05,
      "loss": 1.3441,
      "step": 3330
    },
    {
      "epoch": 0.07621742503765232,
      "grad_norm": 2.5625,
      "learning_rate": 1.9047215752699556e-05,
      "loss": 1.266,
      "step": 3340
    },
    {
      "epoch": 0.07644562092099859,
      "grad_norm": 3.0625,
      "learning_rate": 1.904251064530548e-05,
      "loss": 1.3063,
      "step": 3350
    },
    {
      "epoch": 0.07667381680434485,
      "grad_norm": 3.015625,
      "learning_rate": 1.9037805537911403e-05,
      "loss": 1.3123,
      "step": 3360
    },
    {
      "epoch": 0.07690201268769112,
      "grad_norm": 3.15625,
      "learning_rate": 1.903310043051733e-05,
      "loss": 1.2629,
      "step": 3370
    },
    {
      "epoch": 0.07713020857103738,
      "grad_norm": 2.765625,
      "learning_rate": 1.902839532312325e-05,
      "loss": 1.3116,
      "step": 3380
    },
    {
      "epoch": 0.07735840445438365,
      "grad_norm": 2.9375,
      "learning_rate": 1.9023690215729176e-05,
      "loss": 1.2507,
      "step": 3390
    },
    {
      "epoch": 0.0775866003377299,
      "grad_norm": 2.890625,
      "learning_rate": 1.9018985108335098e-05,
      "loss": 1.2674,
      "step": 3400
    },
    {
      "epoch": 0.07781479622107618,
      "grad_norm": 3.578125,
      "learning_rate": 1.9014280000941024e-05,
      "loss": 1.288,
      "step": 3410
    },
    {
      "epoch": 0.07804299210442243,
      "grad_norm": 3.015625,
      "learning_rate": 1.9009574893546946e-05,
      "loss": 1.3085,
      "step": 3420
    },
    {
      "epoch": 0.0782711879877687,
      "grad_norm": 3.109375,
      "learning_rate": 1.900486978615287e-05,
      "loss": 1.271,
      "step": 3430
    },
    {
      "epoch": 0.07849938387111496,
      "grad_norm": 2.75,
      "learning_rate": 1.9000164678758793e-05,
      "loss": 1.2267,
      "step": 3440
    },
    {
      "epoch": 0.07872757975446124,
      "grad_norm": 2.640625,
      "learning_rate": 1.899545957136472e-05,
      "loss": 1.2884,
      "step": 3450
    },
    {
      "epoch": 0.0789557756378075,
      "grad_norm": 3.125,
      "learning_rate": 1.899075446397064e-05,
      "loss": 1.2701,
      "step": 3460
    },
    {
      "epoch": 0.07918397152115376,
      "grad_norm": 2.953125,
      "learning_rate": 1.8986049356576566e-05,
      "loss": 1.3125,
      "step": 3470
    },
    {
      "epoch": 0.07941216740450002,
      "grad_norm": 2.9375,
      "learning_rate": 1.8981344249182488e-05,
      "loss": 1.2674,
      "step": 3480
    },
    {
      "epoch": 0.0796403632878463,
      "grad_norm": 2.96875,
      "learning_rate": 1.8976639141788413e-05,
      "loss": 1.3193,
      "step": 3490
    },
    {
      "epoch": 0.07986855917119255,
      "grad_norm": 3.109375,
      "learning_rate": 1.8971934034394335e-05,
      "loss": 1.3263,
      "step": 3500
    },
    {
      "epoch": 0.08009675505453881,
      "grad_norm": 2.953125,
      "learning_rate": 1.896722892700026e-05,
      "loss": 1.3223,
      "step": 3510
    },
    {
      "epoch": 0.08032495093788508,
      "grad_norm": 3.140625,
      "learning_rate": 1.8962523819606186e-05,
      "loss": 1.2505,
      "step": 3520
    },
    {
      "epoch": 0.08055314682123134,
      "grad_norm": 2.890625,
      "learning_rate": 1.8957818712212108e-05,
      "loss": 1.2797,
      "step": 3530
    },
    {
      "epoch": 0.08078134270457761,
      "grad_norm": 2.984375,
      "learning_rate": 1.8953113604818034e-05,
      "loss": 1.2591,
      "step": 3540
    },
    {
      "epoch": 0.08100953858792387,
      "grad_norm": 2.90625,
      "learning_rate": 1.8948408497423956e-05,
      "loss": 1.259,
      "step": 3550
    },
    {
      "epoch": 0.08123773447127014,
      "grad_norm": 2.921875,
      "learning_rate": 1.894370339002988e-05,
      "loss": 1.2834,
      "step": 3560
    },
    {
      "epoch": 0.0814659303546164,
      "grad_norm": 2.8125,
      "learning_rate": 1.8938998282635803e-05,
      "loss": 1.2744,
      "step": 3570
    },
    {
      "epoch": 0.08169412623796267,
      "grad_norm": 2.578125,
      "learning_rate": 1.893429317524173e-05,
      "loss": 1.2332,
      "step": 3580
    },
    {
      "epoch": 0.08192232212130893,
      "grad_norm": 2.984375,
      "learning_rate": 1.892958806784765e-05,
      "loss": 1.2801,
      "step": 3590
    },
    {
      "epoch": 0.0821505180046552,
      "grad_norm": 3.234375,
      "learning_rate": 1.8924882960453576e-05,
      "loss": 1.2917,
      "step": 3600
    },
    {
      "epoch": 0.08237871388800146,
      "grad_norm": 3.0,
      "learning_rate": 1.8920177853059498e-05,
      "loss": 1.33,
      "step": 3610
    },
    {
      "epoch": 0.08260690977134773,
      "grad_norm": 3.359375,
      "learning_rate": 1.8915472745665423e-05,
      "loss": 1.315,
      "step": 3620
    },
    {
      "epoch": 0.08283510565469399,
      "grad_norm": 2.90625,
      "learning_rate": 1.8910767638271345e-05,
      "loss": 1.2019,
      "step": 3630
    },
    {
      "epoch": 0.08306330153804026,
      "grad_norm": 2.96875,
      "learning_rate": 1.890606253087727e-05,
      "loss": 1.3372,
      "step": 3640
    },
    {
      "epoch": 0.08329149742138652,
      "grad_norm": 3.046875,
      "learning_rate": 1.8901357423483193e-05,
      "loss": 1.2758,
      "step": 3650
    },
    {
      "epoch": 0.08351969330473279,
      "grad_norm": 3.015625,
      "learning_rate": 1.8896652316089118e-05,
      "loss": 1.2893,
      "step": 3660
    },
    {
      "epoch": 0.08374788918807904,
      "grad_norm": 2.875,
      "learning_rate": 1.889194720869504e-05,
      "loss": 1.2878,
      "step": 3670
    },
    {
      "epoch": 0.08397608507142532,
      "grad_norm": 2.609375,
      "learning_rate": 1.8887242101300966e-05,
      "loss": 1.2628,
      "step": 3680
    },
    {
      "epoch": 0.08420428095477157,
      "grad_norm": 3.046875,
      "learning_rate": 1.8882536993906888e-05,
      "loss": 1.3282,
      "step": 3690
    },
    {
      "epoch": 0.08443247683811785,
      "grad_norm": 2.890625,
      "learning_rate": 1.8877831886512813e-05,
      "loss": 1.2558,
      "step": 3700
    },
    {
      "epoch": 0.0846606727214641,
      "grad_norm": 3.6875,
      "learning_rate": 1.8873126779118735e-05,
      "loss": 1.2594,
      "step": 3710
    },
    {
      "epoch": 0.08488886860481037,
      "grad_norm": 2.84375,
      "learning_rate": 1.886842167172466e-05,
      "loss": 1.2525,
      "step": 3720
    },
    {
      "epoch": 0.08511706448815663,
      "grad_norm": 2.90625,
      "learning_rate": 1.8863716564330582e-05,
      "loss": 1.2717,
      "step": 3730
    },
    {
      "epoch": 0.0853452603715029,
      "grad_norm": 3.09375,
      "learning_rate": 1.8859011456936508e-05,
      "loss": 1.2786,
      "step": 3740
    },
    {
      "epoch": 0.08557345625484916,
      "grad_norm": 2.859375,
      "learning_rate": 1.885430634954243e-05,
      "loss": 1.2976,
      "step": 3750
    },
    {
      "epoch": 0.08580165213819543,
      "grad_norm": 2.65625,
      "learning_rate": 1.8849601242148355e-05,
      "loss": 1.2451,
      "step": 3760
    },
    {
      "epoch": 0.08602984802154169,
      "grad_norm": 2.6875,
      "learning_rate": 1.8844896134754277e-05,
      "loss": 1.2518,
      "step": 3770
    },
    {
      "epoch": 0.08625804390488796,
      "grad_norm": 2.84375,
      "learning_rate": 1.8840191027360203e-05,
      "loss": 1.2399,
      "step": 3780
    },
    {
      "epoch": 0.08648623978823422,
      "grad_norm": 3.09375,
      "learning_rate": 1.8835485919966125e-05,
      "loss": 1.2926,
      "step": 3790
    },
    {
      "epoch": 0.08671443567158048,
      "grad_norm": 3.078125,
      "learning_rate": 1.883078081257205e-05,
      "loss": 1.3284,
      "step": 3800
    },
    {
      "epoch": 0.08694263155492675,
      "grad_norm": 2.890625,
      "learning_rate": 1.8826075705177972e-05,
      "loss": 1.3465,
      "step": 3810
    },
    {
      "epoch": 0.08717082743827301,
      "grad_norm": 2.8125,
      "learning_rate": 1.8821370597783898e-05,
      "loss": 1.2742,
      "step": 3820
    },
    {
      "epoch": 0.08739902332161928,
      "grad_norm": 2.640625,
      "learning_rate": 1.881666549038982e-05,
      "loss": 1.2418,
      "step": 3830
    },
    {
      "epoch": 0.08762721920496554,
      "grad_norm": 2.890625,
      "learning_rate": 1.8811960382995745e-05,
      "loss": 1.2656,
      "step": 3840
    },
    {
      "epoch": 0.08785541508831181,
      "grad_norm": 2.96875,
      "learning_rate": 1.8807255275601667e-05,
      "loss": 1.3345,
      "step": 3850
    },
    {
      "epoch": 0.08808361097165807,
      "grad_norm": 2.796875,
      "learning_rate": 1.8802550168207592e-05,
      "loss": 1.3108,
      "step": 3860
    },
    {
      "epoch": 0.08831180685500434,
      "grad_norm": 3.0,
      "learning_rate": 1.8797845060813514e-05,
      "loss": 1.3534,
      "step": 3870
    },
    {
      "epoch": 0.0885400027383506,
      "grad_norm": 2.890625,
      "learning_rate": 1.879313995341944e-05,
      "loss": 1.2493,
      "step": 3880
    },
    {
      "epoch": 0.08876819862169687,
      "grad_norm": 3.1875,
      "learning_rate": 1.8788434846025362e-05,
      "loss": 1.3026,
      "step": 3890
    },
    {
      "epoch": 0.08899639450504312,
      "grad_norm": 2.84375,
      "learning_rate": 1.8783729738631287e-05,
      "loss": 1.2977,
      "step": 3900
    },
    {
      "epoch": 0.0892245903883894,
      "grad_norm": 2.890625,
      "learning_rate": 1.877902463123721e-05,
      "loss": 1.2378,
      "step": 3910
    },
    {
      "epoch": 0.08945278627173565,
      "grad_norm": 3.109375,
      "learning_rate": 1.8774319523843135e-05,
      "loss": 1.2688,
      "step": 3920
    },
    {
      "epoch": 0.08968098215508193,
      "grad_norm": 2.96875,
      "learning_rate": 1.8769614416449057e-05,
      "loss": 1.2846,
      "step": 3930
    },
    {
      "epoch": 0.08990917803842818,
      "grad_norm": 2.890625,
      "learning_rate": 1.8764909309054982e-05,
      "loss": 1.2867,
      "step": 3940
    },
    {
      "epoch": 0.09013737392177446,
      "grad_norm": 2.890625,
      "learning_rate": 1.8760204201660904e-05,
      "loss": 1.2363,
      "step": 3950
    },
    {
      "epoch": 0.09036556980512071,
      "grad_norm": 2.921875,
      "learning_rate": 1.875549909426683e-05,
      "loss": 1.3008,
      "step": 3960
    },
    {
      "epoch": 0.09059376568846698,
      "grad_norm": 2.796875,
      "learning_rate": 1.875079398687275e-05,
      "loss": 1.2138,
      "step": 3970
    },
    {
      "epoch": 0.09082196157181324,
      "grad_norm": 3.1875,
      "learning_rate": 1.8746088879478677e-05,
      "loss": 1.2049,
      "step": 3980
    },
    {
      "epoch": 0.09105015745515951,
      "grad_norm": 3.15625,
      "learning_rate": 1.87413837720846e-05,
      "loss": 1.2463,
      "step": 3990
    },
    {
      "epoch": 0.09127835333850577,
      "grad_norm": 2.84375,
      "learning_rate": 1.8736678664690524e-05,
      "loss": 1.2972,
      "step": 4000
    },
    {
      "epoch": 0.09150654922185204,
      "grad_norm": 3.03125,
      "learning_rate": 1.8731973557296446e-05,
      "loss": 1.2633,
      "step": 4010
    },
    {
      "epoch": 0.0917347451051983,
      "grad_norm": 3.03125,
      "learning_rate": 1.8727268449902372e-05,
      "loss": 1.247,
      "step": 4020
    },
    {
      "epoch": 0.09196294098854457,
      "grad_norm": 3.09375,
      "learning_rate": 1.8722563342508294e-05,
      "loss": 1.2925,
      "step": 4030
    },
    {
      "epoch": 0.09219113687189083,
      "grad_norm": 3.078125,
      "learning_rate": 1.871785823511422e-05,
      "loss": 1.2233,
      "step": 4040
    },
    {
      "epoch": 0.0924193327552371,
      "grad_norm": 3.0625,
      "learning_rate": 1.871315312772014e-05,
      "loss": 1.3218,
      "step": 4050
    },
    {
      "epoch": 0.09264752863858336,
      "grad_norm": 3.28125,
      "learning_rate": 1.8708448020326067e-05,
      "loss": 1.3449,
      "step": 4060
    },
    {
      "epoch": 0.09287572452192963,
      "grad_norm": 3.109375,
      "learning_rate": 1.870374291293199e-05,
      "loss": 1.2999,
      "step": 4070
    },
    {
      "epoch": 0.09310392040527589,
      "grad_norm": 2.75,
      "learning_rate": 1.8699037805537914e-05,
      "loss": 1.2917,
      "step": 4080
    },
    {
      "epoch": 0.09333211628862215,
      "grad_norm": 3.015625,
      "learning_rate": 1.8694332698143836e-05,
      "loss": 1.2619,
      "step": 4090
    },
    {
      "epoch": 0.09356031217196842,
      "grad_norm": 3.03125,
      "learning_rate": 1.8689627590749758e-05,
      "loss": 1.2881,
      "step": 4100
    },
    {
      "epoch": 0.09378850805531468,
      "grad_norm": 3.046875,
      "learning_rate": 1.8684922483355683e-05,
      "loss": 1.2401,
      "step": 4110
    },
    {
      "epoch": 0.09401670393866095,
      "grad_norm": 2.890625,
      "learning_rate": 1.8680217375961606e-05,
      "loss": 1.2744,
      "step": 4120
    },
    {
      "epoch": 0.0942448998220072,
      "grad_norm": 2.8125,
      "learning_rate": 1.867551226856753e-05,
      "loss": 1.256,
      "step": 4130
    },
    {
      "epoch": 0.09447309570535348,
      "grad_norm": 2.890625,
      "learning_rate": 1.8670807161173453e-05,
      "loss": 1.2605,
      "step": 4140
    },
    {
      "epoch": 0.09470129158869973,
      "grad_norm": 2.9375,
      "learning_rate": 1.866610205377938e-05,
      "loss": 1.255,
      "step": 4150
    },
    {
      "epoch": 0.094929487472046,
      "grad_norm": 2.78125,
      "learning_rate": 1.86613969463853e-05,
      "loss": 1.2723,
      "step": 4160
    },
    {
      "epoch": 0.09515768335539226,
      "grad_norm": 3.21875,
      "learning_rate": 1.8656691838991226e-05,
      "loss": 1.2614,
      "step": 4170
    },
    {
      "epoch": 0.09538587923873854,
      "grad_norm": 3.046875,
      "learning_rate": 1.8651986731597148e-05,
      "loss": 1.2515,
      "step": 4180
    },
    {
      "epoch": 0.0956140751220848,
      "grad_norm": 3.25,
      "learning_rate": 1.8647281624203073e-05,
      "loss": 1.2351,
      "step": 4190
    },
    {
      "epoch": 0.09584227100543107,
      "grad_norm": 2.953125,
      "learning_rate": 1.8642576516808995e-05,
      "loss": 1.2404,
      "step": 4200
    },
    {
      "epoch": 0.09607046688877732,
      "grad_norm": 2.765625,
      "learning_rate": 1.863787140941492e-05,
      "loss": 1.3401,
      "step": 4210
    },
    {
      "epoch": 0.0962986627721236,
      "grad_norm": 3.03125,
      "learning_rate": 1.8633166302020843e-05,
      "loss": 1.2879,
      "step": 4220
    },
    {
      "epoch": 0.09652685865546985,
      "grad_norm": 3.21875,
      "learning_rate": 1.8628461194626768e-05,
      "loss": 1.2982,
      "step": 4230
    },
    {
      "epoch": 0.09675505453881612,
      "grad_norm": 2.9375,
      "learning_rate": 1.862375608723269e-05,
      "loss": 1.2635,
      "step": 4240
    },
    {
      "epoch": 0.09698325042216238,
      "grad_norm": 2.875,
      "learning_rate": 1.8619050979838615e-05,
      "loss": 1.2737,
      "step": 4250
    },
    {
      "epoch": 0.09721144630550865,
      "grad_norm": 3.0,
      "learning_rate": 1.8614345872444537e-05,
      "loss": 1.2966,
      "step": 4260
    },
    {
      "epoch": 0.09743964218885491,
      "grad_norm": 2.890625,
      "learning_rate": 1.8609640765050463e-05,
      "loss": 1.2325,
      "step": 4270
    },
    {
      "epoch": 0.09766783807220118,
      "grad_norm": 2.71875,
      "learning_rate": 1.8604935657656388e-05,
      "loss": 1.2849,
      "step": 4280
    },
    {
      "epoch": 0.09789603395554744,
      "grad_norm": 3.125,
      "learning_rate": 1.860023055026231e-05,
      "loss": 1.284,
      "step": 4290
    },
    {
      "epoch": 0.09812422983889371,
      "grad_norm": 2.859375,
      "learning_rate": 1.8595525442868236e-05,
      "loss": 1.2073,
      "step": 4300
    },
    {
      "epoch": 0.09835242572223997,
      "grad_norm": 3.140625,
      "learning_rate": 1.8590820335474158e-05,
      "loss": 1.2362,
      "step": 4310
    },
    {
      "epoch": 0.09858062160558624,
      "grad_norm": 2.640625,
      "learning_rate": 1.8586115228080083e-05,
      "loss": 1.2705,
      "step": 4320
    },
    {
      "epoch": 0.0988088174889325,
      "grad_norm": 2.875,
      "learning_rate": 1.8581410120686005e-05,
      "loss": 1.3003,
      "step": 4330
    },
    {
      "epoch": 0.09903701337227877,
      "grad_norm": 2.953125,
      "learning_rate": 1.857670501329193e-05,
      "loss": 1.2598,
      "step": 4340
    },
    {
      "epoch": 0.09926520925562503,
      "grad_norm": 2.96875,
      "learning_rate": 1.8571999905897853e-05,
      "loss": 1.2231,
      "step": 4350
    },
    {
      "epoch": 0.0994934051389713,
      "grad_norm": 2.921875,
      "learning_rate": 1.8567294798503778e-05,
      "loss": 1.2559,
      "step": 4360
    },
    {
      "epoch": 0.09972160102231756,
      "grad_norm": 2.765625,
      "learning_rate": 1.85625896911097e-05,
      "loss": 1.2536,
      "step": 4370
    },
    {
      "epoch": 0.09994979690566382,
      "grad_norm": 2.875,
      "learning_rate": 1.8557884583715625e-05,
      "loss": 1.2588,
      "step": 4380
    },
    {
      "epoch": 0.10017799278901009,
      "grad_norm": 3.046875,
      "learning_rate": 1.8553179476321547e-05,
      "loss": 1.2592,
      "step": 4390
    },
    {
      "epoch": 0.10040618867235634,
      "grad_norm": 2.9375,
      "learning_rate": 1.8548474368927473e-05,
      "loss": 1.269,
      "step": 4400
    },
    {
      "epoch": 0.10063438455570262,
      "grad_norm": 2.953125,
      "learning_rate": 1.8543769261533395e-05,
      "loss": 1.2416,
      "step": 4410
    },
    {
      "epoch": 0.10086258043904887,
      "grad_norm": 2.8125,
      "learning_rate": 1.853906415413932e-05,
      "loss": 1.3027,
      "step": 4420
    },
    {
      "epoch": 0.10109077632239515,
      "grad_norm": 3.296875,
      "learning_rate": 1.8534359046745242e-05,
      "loss": 1.2604,
      "step": 4430
    },
    {
      "epoch": 0.1013189722057414,
      "grad_norm": 2.84375,
      "learning_rate": 1.8529653939351168e-05,
      "loss": 1.2807,
      "step": 4440
    },
    {
      "epoch": 0.10154716808908767,
      "grad_norm": 3.3125,
      "learning_rate": 1.852494883195709e-05,
      "loss": 1.2839,
      "step": 4450
    },
    {
      "epoch": 0.10177536397243393,
      "grad_norm": 3.046875,
      "learning_rate": 1.8520243724563015e-05,
      "loss": 1.2289,
      "step": 4460
    },
    {
      "epoch": 0.1020035598557802,
      "grad_norm": 3.03125,
      "learning_rate": 1.8515538617168937e-05,
      "loss": 1.2233,
      "step": 4470
    },
    {
      "epoch": 0.10223175573912646,
      "grad_norm": 2.859375,
      "learning_rate": 1.8510833509774863e-05,
      "loss": 1.2179,
      "step": 4480
    },
    {
      "epoch": 0.10245995162247273,
      "grad_norm": 2.796875,
      "learning_rate": 1.8506128402380785e-05,
      "loss": 1.3219,
      "step": 4490
    },
    {
      "epoch": 0.10268814750581899,
      "grad_norm": 2.78125,
      "learning_rate": 1.850142329498671e-05,
      "loss": 1.2578,
      "step": 4500
    },
    {
      "epoch": 0.10291634338916526,
      "grad_norm": 2.859375,
      "learning_rate": 1.8496718187592632e-05,
      "loss": 1.2995,
      "step": 4510
    },
    {
      "epoch": 0.10314453927251152,
      "grad_norm": 3.25,
      "learning_rate": 1.8492013080198557e-05,
      "loss": 1.2729,
      "step": 4520
    },
    {
      "epoch": 0.10337273515585779,
      "grad_norm": 2.875,
      "learning_rate": 1.848730797280448e-05,
      "loss": 1.2876,
      "step": 4530
    },
    {
      "epoch": 0.10360093103920405,
      "grad_norm": 2.84375,
      "learning_rate": 1.8482602865410405e-05,
      "loss": 1.2863,
      "step": 4540
    },
    {
      "epoch": 0.10382912692255032,
      "grad_norm": 3.015625,
      "learning_rate": 1.8477897758016327e-05,
      "loss": 1.2277,
      "step": 4550
    },
    {
      "epoch": 0.10405732280589658,
      "grad_norm": 2.96875,
      "learning_rate": 1.8473192650622252e-05,
      "loss": 1.2692,
      "step": 4560
    },
    {
      "epoch": 0.10428551868924285,
      "grad_norm": 2.890625,
      "learning_rate": 1.8468487543228174e-05,
      "loss": 1.2142,
      "step": 4570
    },
    {
      "epoch": 0.10451371457258911,
      "grad_norm": 3.03125,
      "learning_rate": 1.84637824358341e-05,
      "loss": 1.2877,
      "step": 4580
    },
    {
      "epoch": 0.10474191045593538,
      "grad_norm": 3.109375,
      "learning_rate": 1.845907732844002e-05,
      "loss": 1.2741,
      "step": 4590
    },
    {
      "epoch": 0.10497010633928164,
      "grad_norm": 2.9375,
      "learning_rate": 1.8454372221045947e-05,
      "loss": 1.2728,
      "step": 4600
    },
    {
      "epoch": 0.10519830222262791,
      "grad_norm": 2.828125,
      "learning_rate": 1.844966711365187e-05,
      "loss": 1.2688,
      "step": 4610
    },
    {
      "epoch": 0.10542649810597417,
      "grad_norm": 2.65625,
      "learning_rate": 1.8444962006257795e-05,
      "loss": 1.2641,
      "step": 4620
    },
    {
      "epoch": 0.10565469398932044,
      "grad_norm": 3.0,
      "learning_rate": 1.8440256898863717e-05,
      "loss": 1.2258,
      "step": 4630
    },
    {
      "epoch": 0.1058828898726667,
      "grad_norm": 3.15625,
      "learning_rate": 1.8435551791469642e-05,
      "loss": 1.2974,
      "step": 4640
    },
    {
      "epoch": 0.10611108575601297,
      "grad_norm": 2.953125,
      "learning_rate": 1.8430846684075564e-05,
      "loss": 1.2839,
      "step": 4650
    },
    {
      "epoch": 0.10633928163935923,
      "grad_norm": 3.078125,
      "learning_rate": 1.842614157668149e-05,
      "loss": 1.2838,
      "step": 4660
    },
    {
      "epoch": 0.10656747752270548,
      "grad_norm": 3.171875,
      "learning_rate": 1.842143646928741e-05,
      "loss": 1.2589,
      "step": 4670
    },
    {
      "epoch": 0.10679567340605176,
      "grad_norm": 2.796875,
      "learning_rate": 1.8416731361893337e-05,
      "loss": 1.2832,
      "step": 4680
    },
    {
      "epoch": 0.10702386928939801,
      "grad_norm": 3.078125,
      "learning_rate": 1.841202625449926e-05,
      "loss": 1.2472,
      "step": 4690
    },
    {
      "epoch": 0.10725206517274428,
      "grad_norm": 3.078125,
      "learning_rate": 1.8407321147105184e-05,
      "loss": 1.25,
      "step": 4700
    },
    {
      "epoch": 0.10748026105609054,
      "grad_norm": 2.90625,
      "learning_rate": 1.8402616039711106e-05,
      "loss": 1.292,
      "step": 4710
    },
    {
      "epoch": 0.10770845693943681,
      "grad_norm": 3.5,
      "learning_rate": 1.839791093231703e-05,
      "loss": 1.2658,
      "step": 4720
    },
    {
      "epoch": 0.10793665282278307,
      "grad_norm": 2.71875,
      "learning_rate": 1.8393205824922954e-05,
      "loss": 1.2651,
      "step": 4730
    },
    {
      "epoch": 0.10816484870612934,
      "grad_norm": 3.140625,
      "learning_rate": 1.838850071752888e-05,
      "loss": 1.3102,
      "step": 4740
    },
    {
      "epoch": 0.1083930445894756,
      "grad_norm": 3.0,
      "learning_rate": 1.83837956101348e-05,
      "loss": 1.2472,
      "step": 4750
    },
    {
      "epoch": 0.10862124047282187,
      "grad_norm": 2.84375,
      "learning_rate": 1.8379090502740726e-05,
      "loss": 1.2705,
      "step": 4760
    },
    {
      "epoch": 0.10884943635616813,
      "grad_norm": 3.15625,
      "learning_rate": 1.837438539534665e-05,
      "loss": 1.2625,
      "step": 4770
    },
    {
      "epoch": 0.1090776322395144,
      "grad_norm": 3.234375,
      "learning_rate": 1.8369680287952574e-05,
      "loss": 1.2495,
      "step": 4780
    },
    {
      "epoch": 0.10930582812286066,
      "grad_norm": 2.96875,
      "learning_rate": 1.8364975180558496e-05,
      "loss": 1.3189,
      "step": 4790
    },
    {
      "epoch": 0.10953402400620693,
      "grad_norm": 2.8125,
      "learning_rate": 1.836027007316442e-05,
      "loss": 1.2462,
      "step": 4800
    },
    {
      "epoch": 0.10976221988955319,
      "grad_norm": 2.9375,
      "learning_rate": 1.8355564965770343e-05,
      "loss": 1.2437,
      "step": 4810
    },
    {
      "epoch": 0.10999041577289946,
      "grad_norm": 3.109375,
      "learning_rate": 1.835085985837627e-05,
      "loss": 1.2827,
      "step": 4820
    },
    {
      "epoch": 0.11021861165624572,
      "grad_norm": 2.765625,
      "learning_rate": 1.834615475098219e-05,
      "loss": 1.2788,
      "step": 4830
    },
    {
      "epoch": 0.11044680753959199,
      "grad_norm": 2.765625,
      "learning_rate": 1.8341449643588116e-05,
      "loss": 1.3199,
      "step": 4840
    },
    {
      "epoch": 0.11067500342293825,
      "grad_norm": 2.890625,
      "learning_rate": 1.833674453619404e-05,
      "loss": 1.247,
      "step": 4850
    },
    {
      "epoch": 0.11090319930628452,
      "grad_norm": 2.875,
      "learning_rate": 1.8332039428799964e-05,
      "loss": 1.3147,
      "step": 4860
    },
    {
      "epoch": 0.11113139518963078,
      "grad_norm": 2.96875,
      "learning_rate": 1.832733432140589e-05,
      "loss": 1.2362,
      "step": 4870
    },
    {
      "epoch": 0.11135959107297705,
      "grad_norm": 3.21875,
      "learning_rate": 1.832262921401181e-05,
      "loss": 1.2855,
      "step": 4880
    },
    {
      "epoch": 0.1115877869563233,
      "grad_norm": 2.734375,
      "learning_rate": 1.8317924106617736e-05,
      "loss": 1.2943,
      "step": 4890
    },
    {
      "epoch": 0.11181598283966958,
      "grad_norm": 3.140625,
      "learning_rate": 1.831321899922366e-05,
      "loss": 1.2613,
      "step": 4900
    },
    {
      "epoch": 0.11204417872301584,
      "grad_norm": 2.703125,
      "learning_rate": 1.8308513891829584e-05,
      "loss": 1.2518,
      "step": 4910
    },
    {
      "epoch": 0.11227237460636211,
      "grad_norm": 2.953125,
      "learning_rate": 1.8303808784435506e-05,
      "loss": 1.236,
      "step": 4920
    },
    {
      "epoch": 0.11250057048970837,
      "grad_norm": 2.953125,
      "learning_rate": 1.829910367704143e-05,
      "loss": 1.246,
      "step": 4930
    },
    {
      "epoch": 0.11272876637305464,
      "grad_norm": 2.984375,
      "learning_rate": 1.8294398569647353e-05,
      "loss": 1.3012,
      "step": 4940
    },
    {
      "epoch": 0.1129569622564009,
      "grad_norm": 2.84375,
      "learning_rate": 1.828969346225328e-05,
      "loss": 1.2405,
      "step": 4950
    },
    {
      "epoch": 0.11318515813974715,
      "grad_norm": 2.828125,
      "learning_rate": 1.82849883548592e-05,
      "loss": 1.3221,
      "step": 4960
    },
    {
      "epoch": 0.11341335402309342,
      "grad_norm": 2.9375,
      "learning_rate": 1.8280283247465126e-05,
      "loss": 1.2593,
      "step": 4970
    },
    {
      "epoch": 0.11364154990643968,
      "grad_norm": 3.34375,
      "learning_rate": 1.8275578140071048e-05,
      "loss": 1.2521,
      "step": 4980
    },
    {
      "epoch": 0.11386974578978595,
      "grad_norm": 3.125,
      "learning_rate": 1.8270873032676974e-05,
      "loss": 1.2798,
      "step": 4990
    },
    {
      "epoch": 0.11409794167313221,
      "grad_norm": 2.890625,
      "learning_rate": 1.8266167925282896e-05,
      "loss": 1.2496,
      "step": 5000
    },
    {
      "epoch": 0.11432613755647848,
      "grad_norm": 2.84375,
      "learning_rate": 1.826146281788882e-05,
      "loss": 1.2854,
      "step": 5010
    },
    {
      "epoch": 0.11455433343982474,
      "grad_norm": 2.890625,
      "learning_rate": 1.8256757710494743e-05,
      "loss": 1.2502,
      "step": 5020
    },
    {
      "epoch": 0.11478252932317101,
      "grad_norm": 2.796875,
      "learning_rate": 1.825205260310067e-05,
      "loss": 1.3004,
      "step": 5030
    },
    {
      "epoch": 0.11501072520651727,
      "grad_norm": 2.96875,
      "learning_rate": 1.824734749570659e-05,
      "loss": 1.2751,
      "step": 5040
    },
    {
      "epoch": 0.11523892108986354,
      "grad_norm": 2.75,
      "learning_rate": 1.8242642388312516e-05,
      "loss": 1.2622,
      "step": 5050
    },
    {
      "epoch": 0.1154671169732098,
      "grad_norm": 3.296875,
      "learning_rate": 1.8237937280918438e-05,
      "loss": 1.2805,
      "step": 5060
    },
    {
      "epoch": 0.11569531285655607,
      "grad_norm": 3.046875,
      "learning_rate": 1.8233232173524363e-05,
      "loss": 1.3034,
      "step": 5070
    },
    {
      "epoch": 0.11592350873990233,
      "grad_norm": 3.0625,
      "learning_rate": 1.8228527066130285e-05,
      "loss": 1.2822,
      "step": 5080
    },
    {
      "epoch": 0.1161517046232486,
      "grad_norm": 2.796875,
      "learning_rate": 1.822382195873621e-05,
      "loss": 1.2507,
      "step": 5090
    },
    {
      "epoch": 0.11637990050659486,
      "grad_norm": 2.921875,
      "learning_rate": 1.8219116851342133e-05,
      "loss": 1.2798,
      "step": 5100
    },
    {
      "epoch": 0.11660809638994113,
      "grad_norm": 3.046875,
      "learning_rate": 1.8214411743948058e-05,
      "loss": 1.2438,
      "step": 5110
    },
    {
      "epoch": 0.11683629227328739,
      "grad_norm": 2.84375,
      "learning_rate": 1.820970663655398e-05,
      "loss": 1.2499,
      "step": 5120
    },
    {
      "epoch": 0.11706448815663366,
      "grad_norm": 2.984375,
      "learning_rate": 1.8205001529159906e-05,
      "loss": 1.2277,
      "step": 5130
    },
    {
      "epoch": 0.11729268403997992,
      "grad_norm": 2.953125,
      "learning_rate": 1.8200296421765828e-05,
      "loss": 1.2105,
      "step": 5140
    },
    {
      "epoch": 0.11752087992332619,
      "grad_norm": 3.0625,
      "learning_rate": 1.8195591314371753e-05,
      "loss": 1.2571,
      "step": 5150
    },
    {
      "epoch": 0.11774907580667245,
      "grad_norm": 3.171875,
      "learning_rate": 1.8190886206977675e-05,
      "loss": 1.281,
      "step": 5160
    },
    {
      "epoch": 0.11797727169001872,
      "grad_norm": 3.078125,
      "learning_rate": 1.81861810995836e-05,
      "loss": 1.2809,
      "step": 5170
    },
    {
      "epoch": 0.11820546757336498,
      "grad_norm": 2.84375,
      "learning_rate": 1.8181475992189522e-05,
      "loss": 1.2533,
      "step": 5180
    },
    {
      "epoch": 0.11843366345671125,
      "grad_norm": 2.921875,
      "learning_rate": 1.8176770884795448e-05,
      "loss": 1.2672,
      "step": 5190
    },
    {
      "epoch": 0.1186618593400575,
      "grad_norm": 3.203125,
      "learning_rate": 1.817206577740137e-05,
      "loss": 1.301,
      "step": 5200
    },
    {
      "epoch": 0.11889005522340378,
      "grad_norm": 2.921875,
      "learning_rate": 1.8167360670007295e-05,
      "loss": 1.2164,
      "step": 5210
    },
    {
      "epoch": 0.11911825110675003,
      "grad_norm": 3.125,
      "learning_rate": 1.8162655562613217e-05,
      "loss": 1.2581,
      "step": 5220
    },
    {
      "epoch": 0.1193464469900963,
      "grad_norm": 3.125,
      "learning_rate": 1.8157950455219143e-05,
      "loss": 1.2218,
      "step": 5230
    },
    {
      "epoch": 0.11957464287344256,
      "grad_norm": 3.03125,
      "learning_rate": 1.8153245347825065e-05,
      "loss": 1.2251,
      "step": 5240
    },
    {
      "epoch": 0.11980283875678882,
      "grad_norm": 2.890625,
      "learning_rate": 1.814854024043099e-05,
      "loss": 1.2544,
      "step": 5250
    },
    {
      "epoch": 0.12003103464013509,
      "grad_norm": 3.15625,
      "learning_rate": 1.8143835133036912e-05,
      "loss": 1.3162,
      "step": 5260
    },
    {
      "epoch": 0.12025923052348135,
      "grad_norm": 2.921875,
      "learning_rate": 1.8139130025642838e-05,
      "loss": 1.1967,
      "step": 5270
    },
    {
      "epoch": 0.12048742640682762,
      "grad_norm": 2.890625,
      "learning_rate": 1.813442491824876e-05,
      "loss": 1.2682,
      "step": 5280
    },
    {
      "epoch": 0.12071562229017388,
      "grad_norm": 3.359375,
      "learning_rate": 1.8129719810854685e-05,
      "loss": 1.2578,
      "step": 5290
    },
    {
      "epoch": 0.12094381817352015,
      "grad_norm": 3.03125,
      "learning_rate": 1.8125014703460607e-05,
      "loss": 1.2763,
      "step": 5300
    },
    {
      "epoch": 0.12117201405686641,
      "grad_norm": 2.734375,
      "learning_rate": 1.8120309596066532e-05,
      "loss": 1.2392,
      "step": 5310
    },
    {
      "epoch": 0.12140020994021268,
      "grad_norm": 3.1875,
      "learning_rate": 1.8115604488672454e-05,
      "loss": 1.2178,
      "step": 5320
    },
    {
      "epoch": 0.12162840582355894,
      "grad_norm": 3.3125,
      "learning_rate": 1.811089938127838e-05,
      "loss": 1.2655,
      "step": 5330
    },
    {
      "epoch": 0.12185660170690521,
      "grad_norm": 3.078125,
      "learning_rate": 1.8106194273884302e-05,
      "loss": 1.2879,
      "step": 5340
    },
    {
      "epoch": 0.12208479759025147,
      "grad_norm": 2.859375,
      "learning_rate": 1.8101489166490227e-05,
      "loss": 1.2712,
      "step": 5350
    },
    {
      "epoch": 0.12231299347359774,
      "grad_norm": 2.765625,
      "learning_rate": 1.809678405909615e-05,
      "loss": 1.2287,
      "step": 5360
    },
    {
      "epoch": 0.122541189356944,
      "grad_norm": 3.015625,
      "learning_rate": 1.8092078951702075e-05,
      "loss": 1.2933,
      "step": 5370
    },
    {
      "epoch": 0.12276938524029027,
      "grad_norm": 3.125,
      "learning_rate": 1.8087373844307997e-05,
      "loss": 1.2734,
      "step": 5380
    },
    {
      "epoch": 0.12299758112363653,
      "grad_norm": 3.140625,
      "learning_rate": 1.8082668736913922e-05,
      "loss": 1.265,
      "step": 5390
    },
    {
      "epoch": 0.1232257770069828,
      "grad_norm": 2.671875,
      "learning_rate": 1.8077963629519844e-05,
      "loss": 1.2476,
      "step": 5400
    },
    {
      "epoch": 0.12345397289032906,
      "grad_norm": 3.140625,
      "learning_rate": 1.807325852212577e-05,
      "loss": 1.2881,
      "step": 5410
    },
    {
      "epoch": 0.12368216877367533,
      "grad_norm": 2.828125,
      "learning_rate": 1.8068553414731695e-05,
      "loss": 1.2796,
      "step": 5420
    },
    {
      "epoch": 0.12391036465702158,
      "grad_norm": 2.8125,
      "learning_rate": 1.8063848307337617e-05,
      "loss": 1.2454,
      "step": 5430
    },
    {
      "epoch": 0.12413856054036786,
      "grad_norm": 2.84375,
      "learning_rate": 1.8059143199943542e-05,
      "loss": 1.2776,
      "step": 5440
    },
    {
      "epoch": 0.12436675642371411,
      "grad_norm": 2.8125,
      "learning_rate": 1.8054438092549464e-05,
      "loss": 1.2627,
      "step": 5450
    },
    {
      "epoch": 0.12459495230706039,
      "grad_norm": 2.96875,
      "learning_rate": 1.804973298515539e-05,
      "loss": 1.2423,
      "step": 5460
    },
    {
      "epoch": 0.12482314819040664,
      "grad_norm": 2.890625,
      "learning_rate": 1.8045027877761312e-05,
      "loss": 1.2439,
      "step": 5470
    },
    {
      "epoch": 0.1250513440737529,
      "grad_norm": 2.96875,
      "learning_rate": 1.8040322770367237e-05,
      "loss": 1.217,
      "step": 5480
    },
    {
      "epoch": 0.12527953995709917,
      "grad_norm": 2.796875,
      "learning_rate": 1.803561766297316e-05,
      "loss": 1.2653,
      "step": 5490
    },
    {
      "epoch": 0.12550773584044544,
      "grad_norm": 3.015625,
      "learning_rate": 1.8030912555579085e-05,
      "loss": 1.2117,
      "step": 5500
    },
    {
      "epoch": 0.12573593172379172,
      "grad_norm": 2.9375,
      "learning_rate": 1.8026207448185007e-05,
      "loss": 1.2218,
      "step": 5510
    },
    {
      "epoch": 0.12596412760713796,
      "grad_norm": 3.171875,
      "learning_rate": 1.8021502340790932e-05,
      "loss": 1.2856,
      "step": 5520
    },
    {
      "epoch": 0.12619232349048423,
      "grad_norm": 3.140625,
      "learning_rate": 1.8016797233396854e-05,
      "loss": 1.2522,
      "step": 5530
    },
    {
      "epoch": 0.1264205193738305,
      "grad_norm": 2.984375,
      "learning_rate": 1.801209212600278e-05,
      "loss": 1.2559,
      "step": 5540
    },
    {
      "epoch": 0.12664871525717675,
      "grad_norm": 2.796875,
      "learning_rate": 1.80073870186087e-05,
      "loss": 1.2996,
      "step": 5550
    },
    {
      "epoch": 0.12687691114052302,
      "grad_norm": 3.015625,
      "learning_rate": 1.8002681911214627e-05,
      "loss": 1.2467,
      "step": 5560
    },
    {
      "epoch": 0.1271051070238693,
      "grad_norm": 2.828125,
      "learning_rate": 1.799797680382055e-05,
      "loss": 1.2308,
      "step": 5570
    },
    {
      "epoch": 0.12733330290721556,
      "grad_norm": 3.234375,
      "learning_rate": 1.7993271696426474e-05,
      "loss": 1.2991,
      "step": 5580
    },
    {
      "epoch": 0.1275614987905618,
      "grad_norm": 3.21875,
      "learning_rate": 1.7988566589032396e-05,
      "loss": 1.2194,
      "step": 5590
    },
    {
      "epoch": 0.12778969467390808,
      "grad_norm": 2.890625,
      "learning_rate": 1.7983861481638322e-05,
      "loss": 1.2444,
      "step": 5600
    },
    {
      "epoch": 0.12801789055725435,
      "grad_norm": 2.96875,
      "learning_rate": 1.7979156374244244e-05,
      "loss": 1.2831,
      "step": 5610
    },
    {
      "epoch": 0.12824608644060062,
      "grad_norm": 2.828125,
      "learning_rate": 1.797445126685017e-05,
      "loss": 1.2437,
      "step": 5620
    },
    {
      "epoch": 0.12847428232394686,
      "grad_norm": 3.03125,
      "learning_rate": 1.796974615945609e-05,
      "loss": 1.239,
      "step": 5630
    },
    {
      "epoch": 0.12870247820729314,
      "grad_norm": 2.921875,
      "learning_rate": 1.7965041052062017e-05,
      "loss": 1.2702,
      "step": 5640
    },
    {
      "epoch": 0.1289306740906394,
      "grad_norm": 2.921875,
      "learning_rate": 1.796033594466794e-05,
      "loss": 1.2606,
      "step": 5650
    },
    {
      "epoch": 0.12915886997398568,
      "grad_norm": 3.015625,
      "learning_rate": 1.7955630837273864e-05,
      "loss": 1.3044,
      "step": 5660
    },
    {
      "epoch": 0.12938706585733192,
      "grad_norm": 2.9375,
      "learning_rate": 1.7950925729879786e-05,
      "loss": 1.2438,
      "step": 5670
    },
    {
      "epoch": 0.1296152617406782,
      "grad_norm": 2.84375,
      "learning_rate": 1.794622062248571e-05,
      "loss": 1.1546,
      "step": 5680
    },
    {
      "epoch": 0.12984345762402447,
      "grad_norm": 2.78125,
      "learning_rate": 1.7941515515091633e-05,
      "loss": 1.286,
      "step": 5690
    },
    {
      "epoch": 0.13007165350737074,
      "grad_norm": 3.5,
      "learning_rate": 1.793681040769756e-05,
      "loss": 1.219,
      "step": 5700
    },
    {
      "epoch": 0.13029984939071698,
      "grad_norm": 3.140625,
      "learning_rate": 1.793210530030348e-05,
      "loss": 1.2426,
      "step": 5710
    },
    {
      "epoch": 0.13052804527406325,
      "grad_norm": 3.296875,
      "learning_rate": 1.7927400192909406e-05,
      "loss": 1.2158,
      "step": 5720
    },
    {
      "epoch": 0.13075624115740953,
      "grad_norm": 2.75,
      "learning_rate": 1.7922695085515328e-05,
      "loss": 1.277,
      "step": 5730
    },
    {
      "epoch": 0.1309844370407558,
      "grad_norm": 2.734375,
      "learning_rate": 1.7917989978121254e-05,
      "loss": 1.2435,
      "step": 5740
    },
    {
      "epoch": 0.13121263292410204,
      "grad_norm": 2.921875,
      "learning_rate": 1.7913284870727176e-05,
      "loss": 1.2143,
      "step": 5750
    },
    {
      "epoch": 0.1314408288074483,
      "grad_norm": 2.890625,
      "learning_rate": 1.79085797633331e-05,
      "loss": 1.258,
      "step": 5760
    },
    {
      "epoch": 0.13166902469079458,
      "grad_norm": 2.640625,
      "learning_rate": 1.7903874655939023e-05,
      "loss": 1.2309,
      "step": 5770
    },
    {
      "epoch": 0.13189722057414086,
      "grad_norm": 2.9375,
      "learning_rate": 1.789916954854495e-05,
      "loss": 1.23,
      "step": 5780
    },
    {
      "epoch": 0.1321254164574871,
      "grad_norm": 3.140625,
      "learning_rate": 1.789446444115087e-05,
      "loss": 1.2269,
      "step": 5790
    },
    {
      "epoch": 0.13235361234083337,
      "grad_norm": 2.671875,
      "learning_rate": 1.7889759333756796e-05,
      "loss": 1.2321,
      "step": 5800
    },
    {
      "epoch": 0.13258180822417964,
      "grad_norm": 3.109375,
      "learning_rate": 1.7885054226362718e-05,
      "loss": 1.2601,
      "step": 5810
    },
    {
      "epoch": 0.1328100041075259,
      "grad_norm": 2.75,
      "learning_rate": 1.7880349118968643e-05,
      "loss": 1.2247,
      "step": 5820
    },
    {
      "epoch": 0.13303819999087216,
      "grad_norm": 3.03125,
      "learning_rate": 1.7875644011574565e-05,
      "loss": 1.2723,
      "step": 5830
    },
    {
      "epoch": 0.13326639587421843,
      "grad_norm": 3.921875,
      "learning_rate": 1.787093890418049e-05,
      "loss": 1.2394,
      "step": 5840
    },
    {
      "epoch": 0.1334945917575647,
      "grad_norm": 2.9375,
      "learning_rate": 1.7866233796786413e-05,
      "loss": 1.2017,
      "step": 5850
    },
    {
      "epoch": 0.13372278764091095,
      "grad_norm": 3.0625,
      "learning_rate": 1.7861528689392338e-05,
      "loss": 1.2101,
      "step": 5860
    },
    {
      "epoch": 0.13395098352425722,
      "grad_norm": 2.78125,
      "learning_rate": 1.785682358199826e-05,
      "loss": 1.2719,
      "step": 5870
    },
    {
      "epoch": 0.1341791794076035,
      "grad_norm": 2.828125,
      "learning_rate": 1.7852118474604186e-05,
      "loss": 1.2616,
      "step": 5880
    },
    {
      "epoch": 0.13440737529094976,
      "grad_norm": 3.09375,
      "learning_rate": 1.7847413367210108e-05,
      "loss": 1.2283,
      "step": 5890
    },
    {
      "epoch": 0.134635571174296,
      "grad_norm": 3.140625,
      "learning_rate": 1.7842708259816033e-05,
      "loss": 1.2967,
      "step": 5900
    },
    {
      "epoch": 0.13486376705764228,
      "grad_norm": 2.96875,
      "learning_rate": 1.7838003152421955e-05,
      "loss": 1.2652,
      "step": 5910
    },
    {
      "epoch": 0.13509196294098855,
      "grad_norm": 3.0,
      "learning_rate": 1.783329804502788e-05,
      "loss": 1.2421,
      "step": 5920
    },
    {
      "epoch": 0.13532015882433482,
      "grad_norm": 2.984375,
      "learning_rate": 1.7828592937633803e-05,
      "loss": 1.2841,
      "step": 5930
    },
    {
      "epoch": 0.13554835470768106,
      "grad_norm": 3.140625,
      "learning_rate": 1.7823887830239725e-05,
      "loss": 1.2951,
      "step": 5940
    },
    {
      "epoch": 0.13577655059102733,
      "grad_norm": 3.1875,
      "learning_rate": 1.781918272284565e-05,
      "loss": 1.2944,
      "step": 5950
    },
    {
      "epoch": 0.1360047464743736,
      "grad_norm": 2.90625,
      "learning_rate": 1.7814477615451572e-05,
      "loss": 1.2784,
      "step": 5960
    },
    {
      "epoch": 0.13623294235771988,
      "grad_norm": 3.03125,
      "learning_rate": 1.7809772508057497e-05,
      "loss": 1.2043,
      "step": 5970
    },
    {
      "epoch": 0.13646113824106612,
      "grad_norm": 3.078125,
      "learning_rate": 1.780506740066342e-05,
      "loss": 1.2594,
      "step": 5980
    },
    {
      "epoch": 0.1366893341244124,
      "grad_norm": 2.921875,
      "learning_rate": 1.7800362293269345e-05,
      "loss": 1.3073,
      "step": 5990
    },
    {
      "epoch": 0.13691753000775866,
      "grad_norm": 2.8125,
      "learning_rate": 1.7795657185875267e-05,
      "loss": 1.2917,
      "step": 6000
    },
    {
      "epoch": 0.13714572589110494,
      "grad_norm": 2.953125,
      "learning_rate": 1.7790952078481192e-05,
      "loss": 1.2648,
      "step": 6010
    },
    {
      "epoch": 0.13737392177445118,
      "grad_norm": 3.03125,
      "learning_rate": 1.7786246971087114e-05,
      "loss": 1.2818,
      "step": 6020
    },
    {
      "epoch": 0.13760211765779745,
      "grad_norm": 3.078125,
      "learning_rate": 1.778154186369304e-05,
      "loss": 1.2916,
      "step": 6030
    },
    {
      "epoch": 0.13783031354114372,
      "grad_norm": 3.484375,
      "learning_rate": 1.777683675629896e-05,
      "loss": 1.2765,
      "step": 6040
    },
    {
      "epoch": 0.13805850942449,
      "grad_norm": 3.03125,
      "learning_rate": 1.7772131648904887e-05,
      "loss": 1.1827,
      "step": 6050
    },
    {
      "epoch": 0.13828670530783624,
      "grad_norm": 2.8125,
      "learning_rate": 1.776742654151081e-05,
      "loss": 1.2283,
      "step": 6060
    },
    {
      "epoch": 0.1385149011911825,
      "grad_norm": 2.828125,
      "learning_rate": 1.7762721434116735e-05,
      "loss": 1.2519,
      "step": 6070
    },
    {
      "epoch": 0.13874309707452878,
      "grad_norm": 3.328125,
      "learning_rate": 1.7758016326722657e-05,
      "loss": 1.246,
      "step": 6080
    },
    {
      "epoch": 0.13897129295787505,
      "grad_norm": 2.78125,
      "learning_rate": 1.7753311219328582e-05,
      "loss": 1.237,
      "step": 6090
    },
    {
      "epoch": 0.1391994888412213,
      "grad_norm": 2.96875,
      "learning_rate": 1.7748606111934504e-05,
      "loss": 1.2032,
      "step": 6100
    },
    {
      "epoch": 0.13942768472456757,
      "grad_norm": 2.828125,
      "learning_rate": 1.774390100454043e-05,
      "loss": 1.2937,
      "step": 6110
    },
    {
      "epoch": 0.13965588060791384,
      "grad_norm": 3.078125,
      "learning_rate": 1.773919589714635e-05,
      "loss": 1.279,
      "step": 6120
    },
    {
      "epoch": 0.13988407649126008,
      "grad_norm": 3.140625,
      "learning_rate": 1.7734490789752277e-05,
      "loss": 1.2794,
      "step": 6130
    },
    {
      "epoch": 0.14011227237460636,
      "grad_norm": 3.078125,
      "learning_rate": 1.77297856823582e-05,
      "loss": 1.2253,
      "step": 6140
    },
    {
      "epoch": 0.14034046825795263,
      "grad_norm": 2.9375,
      "learning_rate": 1.7725080574964124e-05,
      "loss": 1.252,
      "step": 6150
    },
    {
      "epoch": 0.1405686641412989,
      "grad_norm": 3.046875,
      "learning_rate": 1.7720375467570046e-05,
      "loss": 1.2493,
      "step": 6160
    },
    {
      "epoch": 0.14079686002464514,
      "grad_norm": 2.921875,
      "learning_rate": 1.771567036017597e-05,
      "loss": 1.2164,
      "step": 6170
    },
    {
      "epoch": 0.14102505590799141,
      "grad_norm": 2.953125,
      "learning_rate": 1.7710965252781897e-05,
      "loss": 1.2976,
      "step": 6180
    },
    {
      "epoch": 0.1412532517913377,
      "grad_norm": 3.03125,
      "learning_rate": 1.770626014538782e-05,
      "loss": 1.2307,
      "step": 6190
    },
    {
      "epoch": 0.14148144767468396,
      "grad_norm": 3.09375,
      "learning_rate": 1.7701555037993744e-05,
      "loss": 1.2403,
      "step": 6200
    },
    {
      "epoch": 0.1417096435580302,
      "grad_norm": 2.875,
      "learning_rate": 1.7696849930599666e-05,
      "loss": 1.2503,
      "step": 6210
    },
    {
      "epoch": 0.14193783944137647,
      "grad_norm": 3.046875,
      "learning_rate": 1.7692144823205592e-05,
      "loss": 1.2889,
      "step": 6220
    },
    {
      "epoch": 0.14216603532472274,
      "grad_norm": 3.25,
      "learning_rate": 1.7687439715811514e-05,
      "loss": 1.2622,
      "step": 6230
    },
    {
      "epoch": 0.14239423120806902,
      "grad_norm": 3.046875,
      "learning_rate": 1.768273460841744e-05,
      "loss": 1.3461,
      "step": 6240
    },
    {
      "epoch": 0.14262242709141526,
      "grad_norm": 3.21875,
      "learning_rate": 1.767802950102336e-05,
      "loss": 1.2733,
      "step": 6250
    },
    {
      "epoch": 0.14285062297476153,
      "grad_norm": 2.8125,
      "learning_rate": 1.7673324393629287e-05,
      "loss": 1.2585,
      "step": 6260
    },
    {
      "epoch": 0.1430788188581078,
      "grad_norm": 2.859375,
      "learning_rate": 1.766861928623521e-05,
      "loss": 1.2415,
      "step": 6270
    },
    {
      "epoch": 0.14330701474145408,
      "grad_norm": 2.921875,
      "learning_rate": 1.7663914178841134e-05,
      "loss": 1.2524,
      "step": 6280
    },
    {
      "epoch": 0.14353521062480032,
      "grad_norm": 2.921875,
      "learning_rate": 1.7659209071447056e-05,
      "loss": 1.2286,
      "step": 6290
    },
    {
      "epoch": 0.1437634065081466,
      "grad_norm": 3.0,
      "learning_rate": 1.765450396405298e-05,
      "loss": 1.2617,
      "step": 6300
    },
    {
      "epoch": 0.14399160239149286,
      "grad_norm": 2.828125,
      "learning_rate": 1.7649798856658904e-05,
      "loss": 1.2547,
      "step": 6310
    },
    {
      "epoch": 0.14421979827483913,
      "grad_norm": 2.65625,
      "learning_rate": 1.764509374926483e-05,
      "loss": 1.2538,
      "step": 6320
    },
    {
      "epoch": 0.14444799415818538,
      "grad_norm": 3.0625,
      "learning_rate": 1.764038864187075e-05,
      "loss": 1.2811,
      "step": 6330
    },
    {
      "epoch": 0.14467619004153165,
      "grad_norm": 3.078125,
      "learning_rate": 1.7635683534476676e-05,
      "loss": 1.2877,
      "step": 6340
    },
    {
      "epoch": 0.14490438592487792,
      "grad_norm": 2.703125,
      "learning_rate": 1.76309784270826e-05,
      "loss": 1.2913,
      "step": 6350
    },
    {
      "epoch": 0.1451325818082242,
      "grad_norm": 2.90625,
      "learning_rate": 1.7626273319688524e-05,
      "loss": 1.2183,
      "step": 6360
    },
    {
      "epoch": 0.14536077769157044,
      "grad_norm": 3.0,
      "learning_rate": 1.7621568212294446e-05,
      "loss": 1.18,
      "step": 6370
    },
    {
      "epoch": 0.1455889735749167,
      "grad_norm": 2.796875,
      "learning_rate": 1.761686310490037e-05,
      "loss": 1.2378,
      "step": 6380
    },
    {
      "epoch": 0.14581716945826298,
      "grad_norm": 2.90625,
      "learning_rate": 1.7612157997506293e-05,
      "loss": 1.2825,
      "step": 6390
    },
    {
      "epoch": 0.14604536534160925,
      "grad_norm": 3.015625,
      "learning_rate": 1.760745289011222e-05,
      "loss": 1.2144,
      "step": 6400
    },
    {
      "epoch": 0.1462735612249555,
      "grad_norm": 2.984375,
      "learning_rate": 1.760274778271814e-05,
      "loss": 1.2253,
      "step": 6410
    },
    {
      "epoch": 0.14650175710830177,
      "grad_norm": 3.171875,
      "learning_rate": 1.7598042675324066e-05,
      "loss": 1.2837,
      "step": 6420
    },
    {
      "epoch": 0.14672995299164804,
      "grad_norm": 3.0,
      "learning_rate": 1.7593337567929988e-05,
      "loss": 1.2847,
      "step": 6430
    },
    {
      "epoch": 0.14695814887499428,
      "grad_norm": 3.109375,
      "learning_rate": 1.7588632460535914e-05,
      "loss": 1.195,
      "step": 6440
    },
    {
      "epoch": 0.14718634475834055,
      "grad_norm": 3.09375,
      "learning_rate": 1.7583927353141836e-05,
      "loss": 1.2393,
      "step": 6450
    },
    {
      "epoch": 0.14741454064168683,
      "grad_norm": 2.828125,
      "learning_rate": 1.757922224574776e-05,
      "loss": 1.2869,
      "step": 6460
    },
    {
      "epoch": 0.1476427365250331,
      "grad_norm": 2.921875,
      "learning_rate": 1.7574517138353683e-05,
      "loss": 1.2533,
      "step": 6470
    },
    {
      "epoch": 0.14787093240837934,
      "grad_norm": 2.890625,
      "learning_rate": 1.756981203095961e-05,
      "loss": 1.1952,
      "step": 6480
    },
    {
      "epoch": 0.1480991282917256,
      "grad_norm": 2.90625,
      "learning_rate": 1.756510692356553e-05,
      "loss": 1.2557,
      "step": 6490
    },
    {
      "epoch": 0.14832732417507188,
      "grad_norm": 3.15625,
      "learning_rate": 1.7560401816171456e-05,
      "loss": 1.1968,
      "step": 6500
    },
    {
      "epoch": 0.14855552005841816,
      "grad_norm": 3.109375,
      "learning_rate": 1.7555696708777378e-05,
      "loss": 1.2667,
      "step": 6510
    },
    {
      "epoch": 0.1487837159417644,
      "grad_norm": 2.96875,
      "learning_rate": 1.7550991601383303e-05,
      "loss": 1.2411,
      "step": 6520
    },
    {
      "epoch": 0.14901191182511067,
      "grad_norm": 2.8125,
      "learning_rate": 1.7546286493989225e-05,
      "loss": 1.2893,
      "step": 6530
    },
    {
      "epoch": 0.14924010770845694,
      "grad_norm": 2.96875,
      "learning_rate": 1.754158138659515e-05,
      "loss": 1.2387,
      "step": 6540
    },
    {
      "epoch": 0.14946830359180321,
      "grad_norm": 2.96875,
      "learning_rate": 1.7536876279201073e-05,
      "loss": 1.1935,
      "step": 6550
    },
    {
      "epoch": 0.14969649947514946,
      "grad_norm": 3.1875,
      "learning_rate": 1.7532171171806998e-05,
      "loss": 1.2669,
      "step": 6560
    },
    {
      "epoch": 0.14992469535849573,
      "grad_norm": 2.90625,
      "learning_rate": 1.752746606441292e-05,
      "loss": 1.2812,
      "step": 6570
    },
    {
      "epoch": 0.150152891241842,
      "grad_norm": 2.859375,
      "learning_rate": 1.7522760957018846e-05,
      "loss": 1.206,
      "step": 6580
    },
    {
      "epoch": 0.15038108712518827,
      "grad_norm": 3.0625,
      "learning_rate": 1.7518055849624768e-05,
      "loss": 1.2076,
      "step": 6590
    },
    {
      "epoch": 0.15060928300853452,
      "grad_norm": 3.1875,
      "learning_rate": 1.7513350742230693e-05,
      "loss": 1.1793,
      "step": 6600
    },
    {
      "epoch": 0.1508374788918808,
      "grad_norm": 3.171875,
      "learning_rate": 1.7508645634836615e-05,
      "loss": 1.2306,
      "step": 6610
    },
    {
      "epoch": 0.15106567477522706,
      "grad_norm": 3.265625,
      "learning_rate": 1.750394052744254e-05,
      "loss": 1.2069,
      "step": 6620
    },
    {
      "epoch": 0.15129387065857333,
      "grad_norm": 2.96875,
      "learning_rate": 1.7499235420048462e-05,
      "loss": 1.2143,
      "step": 6630
    },
    {
      "epoch": 0.15152206654191958,
      "grad_norm": 3.09375,
      "learning_rate": 1.7494530312654388e-05,
      "loss": 1.3262,
      "step": 6640
    },
    {
      "epoch": 0.15175026242526585,
      "grad_norm": 2.734375,
      "learning_rate": 1.748982520526031e-05,
      "loss": 1.2349,
      "step": 6650
    },
    {
      "epoch": 0.15197845830861212,
      "grad_norm": 3.046875,
      "learning_rate": 1.7485120097866235e-05,
      "loss": 1.247,
      "step": 6660
    },
    {
      "epoch": 0.1522066541919584,
      "grad_norm": 2.953125,
      "learning_rate": 1.7480414990472157e-05,
      "loss": 1.2259,
      "step": 6670
    },
    {
      "epoch": 0.15243485007530463,
      "grad_norm": 2.984375,
      "learning_rate": 1.7475709883078083e-05,
      "loss": 1.2278,
      "step": 6680
    },
    {
      "epoch": 0.1526630459586509,
      "grad_norm": 3.078125,
      "learning_rate": 1.7471004775684005e-05,
      "loss": 1.2491,
      "step": 6690
    },
    {
      "epoch": 0.15289124184199718,
      "grad_norm": 2.703125,
      "learning_rate": 1.746629966828993e-05,
      "loss": 1.2678,
      "step": 6700
    },
    {
      "epoch": 0.15311943772534342,
      "grad_norm": 3.265625,
      "learning_rate": 1.7461594560895852e-05,
      "loss": 1.2059,
      "step": 6710
    },
    {
      "epoch": 0.1533476336086897,
      "grad_norm": 3.0625,
      "learning_rate": 1.7456889453501778e-05,
      "loss": 1.2429,
      "step": 6720
    },
    {
      "epoch": 0.15357582949203596,
      "grad_norm": 2.890625,
      "learning_rate": 1.74521843461077e-05,
      "loss": 1.3174,
      "step": 6730
    },
    {
      "epoch": 0.15380402537538224,
      "grad_norm": 2.84375,
      "learning_rate": 1.7447479238713625e-05,
      "loss": 1.1911,
      "step": 6740
    },
    {
      "epoch": 0.15403222125872848,
      "grad_norm": 2.875,
      "learning_rate": 1.744277413131955e-05,
      "loss": 1.2386,
      "step": 6750
    },
    {
      "epoch": 0.15426041714207475,
      "grad_norm": 2.984375,
      "learning_rate": 1.7438069023925472e-05,
      "loss": 1.2324,
      "step": 6760
    },
    {
      "epoch": 0.15448861302542102,
      "grad_norm": 2.921875,
      "learning_rate": 1.7433363916531398e-05,
      "loss": 1.2242,
      "step": 6770
    },
    {
      "epoch": 0.1547168089087673,
      "grad_norm": 2.78125,
      "learning_rate": 1.742865880913732e-05,
      "loss": 1.2817,
      "step": 6780
    },
    {
      "epoch": 0.15494500479211354,
      "grad_norm": 2.703125,
      "learning_rate": 1.7423953701743245e-05,
      "loss": 1.1993,
      "step": 6790
    },
    {
      "epoch": 0.1551732006754598,
      "grad_norm": 3.046875,
      "learning_rate": 1.7419248594349167e-05,
      "loss": 1.2244,
      "step": 6800
    },
    {
      "epoch": 0.15540139655880608,
      "grad_norm": 2.8125,
      "learning_rate": 1.7414543486955093e-05,
      "loss": 1.2154,
      "step": 6810
    },
    {
      "epoch": 0.15562959244215235,
      "grad_norm": 3.078125,
      "learning_rate": 1.7409838379561015e-05,
      "loss": 1.2563,
      "step": 6820
    },
    {
      "epoch": 0.1558577883254986,
      "grad_norm": 3.0625,
      "learning_rate": 1.740513327216694e-05,
      "loss": 1.2159,
      "step": 6830
    },
    {
      "epoch": 0.15608598420884487,
      "grad_norm": 2.90625,
      "learning_rate": 1.7400428164772862e-05,
      "loss": 1.262,
      "step": 6840
    },
    {
      "epoch": 0.15631418009219114,
      "grad_norm": 3.15625,
      "learning_rate": 1.7395723057378787e-05,
      "loss": 1.2336,
      "step": 6850
    },
    {
      "epoch": 0.1565423759755374,
      "grad_norm": 3.21875,
      "learning_rate": 1.739101794998471e-05,
      "loss": 1.2873,
      "step": 6860
    },
    {
      "epoch": 0.15677057185888366,
      "grad_norm": 2.796875,
      "learning_rate": 1.7386312842590635e-05,
      "loss": 1.2407,
      "step": 6870
    },
    {
      "epoch": 0.15699876774222993,
      "grad_norm": 3.171875,
      "learning_rate": 1.7381607735196557e-05,
      "loss": 1.2528,
      "step": 6880
    },
    {
      "epoch": 0.1572269636255762,
      "grad_norm": 3.0,
      "learning_rate": 1.7376902627802482e-05,
      "loss": 1.2919,
      "step": 6890
    },
    {
      "epoch": 0.15745515950892247,
      "grad_norm": 3.5,
      "learning_rate": 1.7372197520408404e-05,
      "loss": 1.2635,
      "step": 6900
    },
    {
      "epoch": 0.15768335539226871,
      "grad_norm": 3.0625,
      "learning_rate": 1.736749241301433e-05,
      "loss": 1.2135,
      "step": 6910
    },
    {
      "epoch": 0.157911551275615,
      "grad_norm": 3.0625,
      "learning_rate": 1.7362787305620252e-05,
      "loss": 1.2121,
      "step": 6920
    },
    {
      "epoch": 0.15813974715896126,
      "grad_norm": 3.171875,
      "learning_rate": 1.7358082198226177e-05,
      "loss": 1.2251,
      "step": 6930
    },
    {
      "epoch": 0.15836794304230753,
      "grad_norm": 2.859375,
      "learning_rate": 1.73533770908321e-05,
      "loss": 1.2839,
      "step": 6940
    },
    {
      "epoch": 0.15859613892565377,
      "grad_norm": 2.90625,
      "learning_rate": 1.7348671983438025e-05,
      "loss": 1.2039,
      "step": 6950
    },
    {
      "epoch": 0.15882433480900005,
      "grad_norm": 2.90625,
      "learning_rate": 1.7343966876043947e-05,
      "loss": 1.2413,
      "step": 6960
    },
    {
      "epoch": 0.15905253069234632,
      "grad_norm": 3.078125,
      "learning_rate": 1.7339261768649872e-05,
      "loss": 1.2439,
      "step": 6970
    },
    {
      "epoch": 0.1592807265756926,
      "grad_norm": 2.8125,
      "learning_rate": 1.7334556661255794e-05,
      "loss": 1.232,
      "step": 6980
    },
    {
      "epoch": 0.15950892245903883,
      "grad_norm": 3.171875,
      "learning_rate": 1.732985155386172e-05,
      "loss": 1.2376,
      "step": 6990
    },
    {
      "epoch": 0.1597371183423851,
      "grad_norm": 2.9375,
      "learning_rate": 1.732514644646764e-05,
      "loss": 1.1671,
      "step": 7000
    },
    {
      "epoch": 0.15996531422573138,
      "grad_norm": 3.203125,
      "learning_rate": 1.7320441339073567e-05,
      "loss": 1.298,
      "step": 7010
    },
    {
      "epoch": 0.16019351010907762,
      "grad_norm": 2.8125,
      "learning_rate": 1.731573623167949e-05,
      "loss": 1.2112,
      "step": 7020
    },
    {
      "epoch": 0.1604217059924239,
      "grad_norm": 3.078125,
      "learning_rate": 1.7311031124285414e-05,
      "loss": 1.2599,
      "step": 7030
    },
    {
      "epoch": 0.16064990187577016,
      "grad_norm": 2.90625,
      "learning_rate": 1.7306326016891336e-05,
      "loss": 1.274,
      "step": 7040
    },
    {
      "epoch": 0.16087809775911643,
      "grad_norm": 3.046875,
      "learning_rate": 1.7301620909497262e-05,
      "loss": 1.2567,
      "step": 7050
    },
    {
      "epoch": 0.16110629364246268,
      "grad_norm": 2.953125,
      "learning_rate": 1.7296915802103184e-05,
      "loss": 1.2401,
      "step": 7060
    },
    {
      "epoch": 0.16133448952580895,
      "grad_norm": 2.84375,
      "learning_rate": 1.729221069470911e-05,
      "loss": 1.2482,
      "step": 7070
    },
    {
      "epoch": 0.16156268540915522,
      "grad_norm": 3.171875,
      "learning_rate": 1.728750558731503e-05,
      "loss": 1.2447,
      "step": 7080
    },
    {
      "epoch": 0.1617908812925015,
      "grad_norm": 2.875,
      "learning_rate": 1.7282800479920957e-05,
      "loss": 1.3049,
      "step": 7090
    },
    {
      "epoch": 0.16201907717584774,
      "grad_norm": 2.828125,
      "learning_rate": 1.727809537252688e-05,
      "loss": 1.2663,
      "step": 7100
    },
    {
      "epoch": 0.162247273059194,
      "grad_norm": 2.75,
      "learning_rate": 1.7273390265132804e-05,
      "loss": 1.2596,
      "step": 7110
    },
    {
      "epoch": 0.16247546894254028,
      "grad_norm": 2.84375,
      "learning_rate": 1.7268685157738726e-05,
      "loss": 1.2227,
      "step": 7120
    },
    {
      "epoch": 0.16270366482588655,
      "grad_norm": 3.21875,
      "learning_rate": 1.726398005034465e-05,
      "loss": 1.2698,
      "step": 7130
    },
    {
      "epoch": 0.1629318607092328,
      "grad_norm": 2.84375,
      "learning_rate": 1.7259274942950573e-05,
      "loss": 1.2448,
      "step": 7140
    },
    {
      "epoch": 0.16316005659257907,
      "grad_norm": 3.296875,
      "learning_rate": 1.72545698355565e-05,
      "loss": 1.2286,
      "step": 7150
    },
    {
      "epoch": 0.16338825247592534,
      "grad_norm": 3.078125,
      "learning_rate": 1.724986472816242e-05,
      "loss": 1.194,
      "step": 7160
    },
    {
      "epoch": 0.1636164483592716,
      "grad_norm": 3.046875,
      "learning_rate": 1.7245159620768346e-05,
      "loss": 1.2211,
      "step": 7170
    },
    {
      "epoch": 0.16384464424261785,
      "grad_norm": 2.9375,
      "learning_rate": 1.7240454513374268e-05,
      "loss": 1.2029,
      "step": 7180
    },
    {
      "epoch": 0.16407284012596413,
      "grad_norm": 2.96875,
      "learning_rate": 1.7235749405980194e-05,
      "loss": 1.2151,
      "step": 7190
    },
    {
      "epoch": 0.1643010360093104,
      "grad_norm": 3.15625,
      "learning_rate": 1.7231044298586116e-05,
      "loss": 1.2413,
      "step": 7200
    },
    {
      "epoch": 0.16452923189265667,
      "grad_norm": 2.71875,
      "learning_rate": 1.722633919119204e-05,
      "loss": 1.213,
      "step": 7210
    },
    {
      "epoch": 0.1647574277760029,
      "grad_norm": 2.59375,
      "learning_rate": 1.7221634083797963e-05,
      "loss": 1.2342,
      "step": 7220
    },
    {
      "epoch": 0.16498562365934918,
      "grad_norm": 3.0625,
      "learning_rate": 1.721692897640389e-05,
      "loss": 1.1909,
      "step": 7230
    },
    {
      "epoch": 0.16521381954269546,
      "grad_norm": 2.625,
      "learning_rate": 1.721222386900981e-05,
      "loss": 1.2339,
      "step": 7240
    },
    {
      "epoch": 0.16544201542604173,
      "grad_norm": 2.640625,
      "learning_rate": 1.7207518761615736e-05,
      "loss": 1.2173,
      "step": 7250
    },
    {
      "epoch": 0.16567021130938797,
      "grad_norm": 3.328125,
      "learning_rate": 1.7202813654221658e-05,
      "loss": 1.2121,
      "step": 7260
    },
    {
      "epoch": 0.16589840719273424,
      "grad_norm": 3.015625,
      "learning_rate": 1.7198108546827583e-05,
      "loss": 1.256,
      "step": 7270
    },
    {
      "epoch": 0.16612660307608051,
      "grad_norm": 3.3125,
      "learning_rate": 1.7193403439433505e-05,
      "loss": 1.2925,
      "step": 7280
    },
    {
      "epoch": 0.16635479895942676,
      "grad_norm": 2.984375,
      "learning_rate": 1.718869833203943e-05,
      "loss": 1.2107,
      "step": 7290
    },
    {
      "epoch": 0.16658299484277303,
      "grad_norm": 2.515625,
      "learning_rate": 1.7183993224645353e-05,
      "loss": 1.2052,
      "step": 7300
    },
    {
      "epoch": 0.1668111907261193,
      "grad_norm": 3.125,
      "learning_rate": 1.7179288117251278e-05,
      "loss": 1.2695,
      "step": 7310
    },
    {
      "epoch": 0.16703938660946557,
      "grad_norm": 3.015625,
      "learning_rate": 1.7174583009857204e-05,
      "loss": 1.2359,
      "step": 7320
    },
    {
      "epoch": 0.16726758249281182,
      "grad_norm": 3.046875,
      "learning_rate": 1.7169877902463126e-05,
      "loss": 1.2219,
      "step": 7330
    },
    {
      "epoch": 0.1674957783761581,
      "grad_norm": 2.984375,
      "learning_rate": 1.716517279506905e-05,
      "loss": 1.2281,
      "step": 7340
    },
    {
      "epoch": 0.16772397425950436,
      "grad_norm": 3.359375,
      "learning_rate": 1.7160467687674973e-05,
      "loss": 1.289,
      "step": 7350
    },
    {
      "epoch": 0.16795217014285063,
      "grad_norm": 3.015625,
      "learning_rate": 1.71557625802809e-05,
      "loss": 1.2174,
      "step": 7360
    },
    {
      "epoch": 0.16818036602619688,
      "grad_norm": 3.234375,
      "learning_rate": 1.715105747288682e-05,
      "loss": 1.2504,
      "step": 7370
    },
    {
      "epoch": 0.16840856190954315,
      "grad_norm": 3.15625,
      "learning_rate": 1.7146352365492746e-05,
      "loss": 1.2153,
      "step": 7380
    },
    {
      "epoch": 0.16863675779288942,
      "grad_norm": 2.796875,
      "learning_rate": 1.7141647258098668e-05,
      "loss": 1.2925,
      "step": 7390
    },
    {
      "epoch": 0.1688649536762357,
      "grad_norm": 3.140625,
      "learning_rate": 1.7136942150704593e-05,
      "loss": 1.2358,
      "step": 7400
    },
    {
      "epoch": 0.16909314955958193,
      "grad_norm": 2.90625,
      "learning_rate": 1.7132237043310515e-05,
      "loss": 1.275,
      "step": 7410
    },
    {
      "epoch": 0.1693213454429282,
      "grad_norm": 2.96875,
      "learning_rate": 1.712753193591644e-05,
      "loss": 1.1721,
      "step": 7420
    },
    {
      "epoch": 0.16954954132627448,
      "grad_norm": 2.796875,
      "learning_rate": 1.7122826828522363e-05,
      "loss": 1.2111,
      "step": 7430
    },
    {
      "epoch": 0.16977773720962075,
      "grad_norm": 2.921875,
      "learning_rate": 1.7118121721128288e-05,
      "loss": 1.2425,
      "step": 7440
    },
    {
      "epoch": 0.170005933092967,
      "grad_norm": 2.703125,
      "learning_rate": 1.711341661373421e-05,
      "loss": 1.1983,
      "step": 7450
    },
    {
      "epoch": 0.17023412897631326,
      "grad_norm": 2.890625,
      "learning_rate": 1.7108711506340136e-05,
      "loss": 1.2396,
      "step": 7460
    },
    {
      "epoch": 0.17046232485965954,
      "grad_norm": 2.8125,
      "learning_rate": 1.7104006398946058e-05,
      "loss": 1.2773,
      "step": 7470
    },
    {
      "epoch": 0.1706905207430058,
      "grad_norm": 3.046875,
      "learning_rate": 1.7099301291551983e-05,
      "loss": 1.216,
      "step": 7480
    },
    {
      "epoch": 0.17091871662635205,
      "grad_norm": 2.953125,
      "learning_rate": 1.7094596184157905e-05,
      "loss": 1.2754,
      "step": 7490
    },
    {
      "epoch": 0.17114691250969832,
      "grad_norm": 3.046875,
      "learning_rate": 1.708989107676383e-05,
      "loss": 1.2823,
      "step": 7500
    },
    {
      "epoch": 0.1713751083930446,
      "grad_norm": 2.984375,
      "learning_rate": 1.7085185969369752e-05,
      "loss": 1.2593,
      "step": 7510
    },
    {
      "epoch": 0.17160330427639087,
      "grad_norm": 3.078125,
      "learning_rate": 1.7080480861975678e-05,
      "loss": 1.2618,
      "step": 7520
    },
    {
      "epoch": 0.1718315001597371,
      "grad_norm": 2.921875,
      "learning_rate": 1.70757757545816e-05,
      "loss": 1.1969,
      "step": 7530
    },
    {
      "epoch": 0.17205969604308338,
      "grad_norm": 3.25,
      "learning_rate": 1.7071070647187525e-05,
      "loss": 1.2147,
      "step": 7540
    },
    {
      "epoch": 0.17228789192642965,
      "grad_norm": 3.25,
      "learning_rate": 1.7066365539793447e-05,
      "loss": 1.2507,
      "step": 7550
    },
    {
      "epoch": 0.17251608780977593,
      "grad_norm": 3.03125,
      "learning_rate": 1.7061660432399373e-05,
      "loss": 1.2206,
      "step": 7560
    },
    {
      "epoch": 0.17274428369312217,
      "grad_norm": 2.875,
      "learning_rate": 1.7056955325005295e-05,
      "loss": 1.3046,
      "step": 7570
    },
    {
      "epoch": 0.17297247957646844,
      "grad_norm": 2.875,
      "learning_rate": 1.705225021761122e-05,
      "loss": 1.2473,
      "step": 7580
    },
    {
      "epoch": 0.1732006754598147,
      "grad_norm": 2.96875,
      "learning_rate": 1.7047545110217142e-05,
      "loss": 1.2497,
      "step": 7590
    },
    {
      "epoch": 0.17342887134316096,
      "grad_norm": 2.84375,
      "learning_rate": 1.7042840002823068e-05,
      "loss": 1.2577,
      "step": 7600
    },
    {
      "epoch": 0.17365706722650723,
      "grad_norm": 3.203125,
      "learning_rate": 1.703813489542899e-05,
      "loss": 1.2982,
      "step": 7610
    },
    {
      "epoch": 0.1738852631098535,
      "grad_norm": 2.875,
      "learning_rate": 1.7033429788034915e-05,
      "loss": 1.1961,
      "step": 7620
    },
    {
      "epoch": 0.17411345899319977,
      "grad_norm": 2.921875,
      "learning_rate": 1.7028724680640837e-05,
      "loss": 1.2168,
      "step": 7630
    },
    {
      "epoch": 0.17434165487654602,
      "grad_norm": 3.078125,
      "learning_rate": 1.7024019573246762e-05,
      "loss": 1.2051,
      "step": 7640
    },
    {
      "epoch": 0.1745698507598923,
      "grad_norm": 2.921875,
      "learning_rate": 1.7019314465852684e-05,
      "loss": 1.231,
      "step": 7650
    },
    {
      "epoch": 0.17479804664323856,
      "grad_norm": 3.03125,
      "learning_rate": 1.701460935845861e-05,
      "loss": 1.1906,
      "step": 7660
    },
    {
      "epoch": 0.17502624252658483,
      "grad_norm": 3.015625,
      "learning_rate": 1.7009904251064532e-05,
      "loss": 1.2849,
      "step": 7670
    },
    {
      "epoch": 0.17525443840993107,
      "grad_norm": 3.140625,
      "learning_rate": 1.7005199143670457e-05,
      "loss": 1.2748,
      "step": 7680
    },
    {
      "epoch": 0.17548263429327735,
      "grad_norm": 3.109375,
      "learning_rate": 1.700049403627638e-05,
      "loss": 1.2084,
      "step": 7690
    },
    {
      "epoch": 0.17571083017662362,
      "grad_norm": 2.890625,
      "learning_rate": 1.6995788928882305e-05,
      "loss": 1.2322,
      "step": 7700
    },
    {
      "epoch": 0.1759390260599699,
      "grad_norm": 3.21875,
      "learning_rate": 1.6991083821488227e-05,
      "loss": 1.226,
      "step": 7710
    },
    {
      "epoch": 0.17616722194331613,
      "grad_norm": 2.75,
      "learning_rate": 1.6986378714094152e-05,
      "loss": 1.2844,
      "step": 7720
    },
    {
      "epoch": 0.1763954178266624,
      "grad_norm": 2.953125,
      "learning_rate": 1.6981673606700074e-05,
      "loss": 1.2484,
      "step": 7730
    },
    {
      "epoch": 0.17662361371000868,
      "grad_norm": 3.109375,
      "learning_rate": 1.6976968499306e-05,
      "loss": 1.2107,
      "step": 7740
    },
    {
      "epoch": 0.17685180959335495,
      "grad_norm": 3.0,
      "learning_rate": 1.697226339191192e-05,
      "loss": 1.2365,
      "step": 7750
    },
    {
      "epoch": 0.1770800054767012,
      "grad_norm": 2.953125,
      "learning_rate": 1.6967558284517847e-05,
      "loss": 1.3016,
      "step": 7760
    },
    {
      "epoch": 0.17730820136004746,
      "grad_norm": 2.90625,
      "learning_rate": 1.696285317712377e-05,
      "loss": 1.2608,
      "step": 7770
    },
    {
      "epoch": 0.17753639724339373,
      "grad_norm": 2.796875,
      "learning_rate": 1.6958148069729694e-05,
      "loss": 1.231,
      "step": 7780
    },
    {
      "epoch": 0.17776459312674,
      "grad_norm": 2.90625,
      "learning_rate": 1.6953442962335616e-05,
      "loss": 1.2222,
      "step": 7790
    },
    {
      "epoch": 0.17799278901008625,
      "grad_norm": 3.046875,
      "learning_rate": 1.694873785494154e-05,
      "loss": 1.2736,
      "step": 7800
    },
    {
      "epoch": 0.17822098489343252,
      "grad_norm": 2.890625,
      "learning_rate": 1.6944032747547464e-05,
      "loss": 1.2609,
      "step": 7810
    },
    {
      "epoch": 0.1784491807767788,
      "grad_norm": 3.09375,
      "learning_rate": 1.6939327640153386e-05,
      "loss": 1.2165,
      "step": 7820
    },
    {
      "epoch": 0.17867737666012506,
      "grad_norm": 3.046875,
      "learning_rate": 1.693462253275931e-05,
      "loss": 1.2544,
      "step": 7830
    },
    {
      "epoch": 0.1789055725434713,
      "grad_norm": 3.125,
      "learning_rate": 1.6929917425365233e-05,
      "loss": 1.189,
      "step": 7840
    },
    {
      "epoch": 0.17913376842681758,
      "grad_norm": 3.015625,
      "learning_rate": 1.692521231797116e-05,
      "loss": 1.2573,
      "step": 7850
    },
    {
      "epoch": 0.17936196431016385,
      "grad_norm": 2.703125,
      "learning_rate": 1.692050721057708e-05,
      "loss": 1.178,
      "step": 7860
    },
    {
      "epoch": 0.1795901601935101,
      "grad_norm": 2.921875,
      "learning_rate": 1.6915802103183006e-05,
      "loss": 1.2282,
      "step": 7870
    },
    {
      "epoch": 0.17981835607685637,
      "grad_norm": 3.1875,
      "learning_rate": 1.6911096995788928e-05,
      "loss": 1.2442,
      "step": 7880
    },
    {
      "epoch": 0.18004655196020264,
      "grad_norm": 3.1875,
      "learning_rate": 1.6906391888394854e-05,
      "loss": 1.2972,
      "step": 7890
    },
    {
      "epoch": 0.1802747478435489,
      "grad_norm": 3.046875,
      "learning_rate": 1.6901686781000776e-05,
      "loss": 1.2147,
      "step": 7900
    },
    {
      "epoch": 0.18050294372689515,
      "grad_norm": 2.84375,
      "learning_rate": 1.68969816736067e-05,
      "loss": 1.2531,
      "step": 7910
    },
    {
      "epoch": 0.18073113961024143,
      "grad_norm": 3.015625,
      "learning_rate": 1.6892276566212623e-05,
      "loss": 1.2508,
      "step": 7920
    },
    {
      "epoch": 0.1809593354935877,
      "grad_norm": 2.765625,
      "learning_rate": 1.688757145881855e-05,
      "loss": 1.2334,
      "step": 7930
    },
    {
      "epoch": 0.18118753137693397,
      "grad_norm": 2.953125,
      "learning_rate": 1.688286635142447e-05,
      "loss": 1.2346,
      "step": 7940
    },
    {
      "epoch": 0.1814157272602802,
      "grad_norm": 2.921875,
      "learning_rate": 1.6878161244030396e-05,
      "loss": 1.2435,
      "step": 7950
    },
    {
      "epoch": 0.18164392314362648,
      "grad_norm": 2.9375,
      "learning_rate": 1.6873456136636318e-05,
      "loss": 1.2238,
      "step": 7960
    },
    {
      "epoch": 0.18187211902697276,
      "grad_norm": 3.125,
      "learning_rate": 1.6868751029242243e-05,
      "loss": 1.2472,
      "step": 7970
    },
    {
      "epoch": 0.18210031491031903,
      "grad_norm": 2.875,
      "learning_rate": 1.6864045921848165e-05,
      "loss": 1.265,
      "step": 7980
    },
    {
      "epoch": 0.18232851079366527,
      "grad_norm": 2.828125,
      "learning_rate": 1.685934081445409e-05,
      "loss": 1.2329,
      "step": 7990
    },
    {
      "epoch": 0.18255670667701154,
      "grad_norm": 3.0,
      "learning_rate": 1.6854635707060013e-05,
      "loss": 1.2732,
      "step": 8000
    },
    {
      "epoch": 0.18278490256035781,
      "grad_norm": 2.90625,
      "learning_rate": 1.6849930599665938e-05,
      "loss": 1.2611,
      "step": 8010
    },
    {
      "epoch": 0.1830130984437041,
      "grad_norm": 2.75,
      "learning_rate": 1.684522549227186e-05,
      "loss": 1.2042,
      "step": 8020
    },
    {
      "epoch": 0.18324129432705033,
      "grad_norm": 3.140625,
      "learning_rate": 1.6840520384877786e-05,
      "loss": 1.2802,
      "step": 8030
    },
    {
      "epoch": 0.1834694902103966,
      "grad_norm": 2.953125,
      "learning_rate": 1.6835815277483708e-05,
      "loss": 1.2261,
      "step": 8040
    },
    {
      "epoch": 0.18369768609374287,
      "grad_norm": 2.890625,
      "learning_rate": 1.6831110170089633e-05,
      "loss": 1.2182,
      "step": 8050
    },
    {
      "epoch": 0.18392588197708915,
      "grad_norm": 3.203125,
      "learning_rate": 1.6826405062695555e-05,
      "loss": 1.2519,
      "step": 8060
    },
    {
      "epoch": 0.1841540778604354,
      "grad_norm": 2.8125,
      "learning_rate": 1.682169995530148e-05,
      "loss": 1.1387,
      "step": 8070
    },
    {
      "epoch": 0.18438227374378166,
      "grad_norm": 2.796875,
      "learning_rate": 1.6816994847907406e-05,
      "loss": 1.2409,
      "step": 8080
    },
    {
      "epoch": 0.18461046962712793,
      "grad_norm": 3.046875,
      "learning_rate": 1.6812289740513328e-05,
      "loss": 1.2371,
      "step": 8090
    },
    {
      "epoch": 0.1848386655104742,
      "grad_norm": 2.796875,
      "learning_rate": 1.6807584633119253e-05,
      "loss": 1.2651,
      "step": 8100
    },
    {
      "epoch": 0.18506686139382045,
      "grad_norm": 2.859375,
      "learning_rate": 1.6802879525725175e-05,
      "loss": 1.2355,
      "step": 8110
    },
    {
      "epoch": 0.18529505727716672,
      "grad_norm": 3.3125,
      "learning_rate": 1.67981744183311e-05,
      "loss": 1.2193,
      "step": 8120
    },
    {
      "epoch": 0.185523253160513,
      "grad_norm": 2.828125,
      "learning_rate": 1.6793469310937023e-05,
      "loss": 1.2621,
      "step": 8130
    },
    {
      "epoch": 0.18575144904385926,
      "grad_norm": 2.953125,
      "learning_rate": 1.6788764203542948e-05,
      "loss": 1.2112,
      "step": 8140
    },
    {
      "epoch": 0.1859796449272055,
      "grad_norm": 3.203125,
      "learning_rate": 1.678405909614887e-05,
      "loss": 1.2499,
      "step": 8150
    },
    {
      "epoch": 0.18620784081055178,
      "grad_norm": 3.125,
      "learning_rate": 1.6779353988754795e-05,
      "loss": 1.2114,
      "step": 8160
    },
    {
      "epoch": 0.18643603669389805,
      "grad_norm": 2.875,
      "learning_rate": 1.6774648881360718e-05,
      "loss": 1.2144,
      "step": 8170
    },
    {
      "epoch": 0.1866642325772443,
      "grad_norm": 2.859375,
      "learning_rate": 1.6769943773966643e-05,
      "loss": 1.2474,
      "step": 8180
    },
    {
      "epoch": 0.18689242846059057,
      "grad_norm": 2.84375,
      "learning_rate": 1.6765238666572565e-05,
      "loss": 1.2681,
      "step": 8190
    },
    {
      "epoch": 0.18712062434393684,
      "grad_norm": 2.953125,
      "learning_rate": 1.676053355917849e-05,
      "loss": 1.2765,
      "step": 8200
    },
    {
      "epoch": 0.1873488202272831,
      "grad_norm": 3.3125,
      "learning_rate": 1.6755828451784412e-05,
      "loss": 1.235,
      "step": 8210
    },
    {
      "epoch": 0.18757701611062935,
      "grad_norm": 2.671875,
      "learning_rate": 1.6751123344390338e-05,
      "loss": 1.2618,
      "step": 8220
    },
    {
      "epoch": 0.18780521199397562,
      "grad_norm": 3.0,
      "learning_rate": 1.674641823699626e-05,
      "loss": 1.2591,
      "step": 8230
    },
    {
      "epoch": 0.1880334078773219,
      "grad_norm": 2.90625,
      "learning_rate": 1.6741713129602185e-05,
      "loss": 1.217,
      "step": 8240
    },
    {
      "epoch": 0.18826160376066817,
      "grad_norm": 3.15625,
      "learning_rate": 1.6737008022208107e-05,
      "loss": 1.2133,
      "step": 8250
    },
    {
      "epoch": 0.1884897996440144,
      "grad_norm": 3.046875,
      "learning_rate": 1.6732302914814033e-05,
      "loss": 1.2446,
      "step": 8260
    },
    {
      "epoch": 0.18871799552736068,
      "grad_norm": 3.09375,
      "learning_rate": 1.6727597807419955e-05,
      "loss": 1.2114,
      "step": 8270
    },
    {
      "epoch": 0.18894619141070695,
      "grad_norm": 3.0,
      "learning_rate": 1.672289270002588e-05,
      "loss": 1.2521,
      "step": 8280
    },
    {
      "epoch": 0.18917438729405323,
      "grad_norm": 2.921875,
      "learning_rate": 1.6718187592631802e-05,
      "loss": 1.2455,
      "step": 8290
    },
    {
      "epoch": 0.18940258317739947,
      "grad_norm": 3.34375,
      "learning_rate": 1.6713482485237727e-05,
      "loss": 1.2078,
      "step": 8300
    },
    {
      "epoch": 0.18963077906074574,
      "grad_norm": 3.0,
      "learning_rate": 1.670877737784365e-05,
      "loss": 1.1979,
      "step": 8310
    },
    {
      "epoch": 0.189858974944092,
      "grad_norm": 3.046875,
      "learning_rate": 1.6704072270449575e-05,
      "loss": 1.3055,
      "step": 8320
    },
    {
      "epoch": 0.19008717082743828,
      "grad_norm": 3.0,
      "learning_rate": 1.6699367163055497e-05,
      "loss": 1.2601,
      "step": 8330
    },
    {
      "epoch": 0.19031536671078453,
      "grad_norm": 2.796875,
      "learning_rate": 1.6694662055661422e-05,
      "loss": 1.2484,
      "step": 8340
    },
    {
      "epoch": 0.1905435625941308,
      "grad_norm": 2.96875,
      "learning_rate": 1.6689956948267344e-05,
      "loss": 1.2764,
      "step": 8350
    },
    {
      "epoch": 0.19077175847747707,
      "grad_norm": 2.859375,
      "learning_rate": 1.668525184087327e-05,
      "loss": 1.3503,
      "step": 8360
    },
    {
      "epoch": 0.19099995436082334,
      "grad_norm": 2.84375,
      "learning_rate": 1.6680546733479192e-05,
      "loss": 1.2904,
      "step": 8370
    },
    {
      "epoch": 0.1912281502441696,
      "grad_norm": 2.765625,
      "learning_rate": 1.6675841626085117e-05,
      "loss": 1.1493,
      "step": 8380
    },
    {
      "epoch": 0.19145634612751586,
      "grad_norm": 3.421875,
      "learning_rate": 1.667113651869104e-05,
      "loss": 1.2052,
      "step": 8390
    },
    {
      "epoch": 0.19168454201086213,
      "grad_norm": 2.796875,
      "learning_rate": 1.6666431411296965e-05,
      "loss": 1.2106,
      "step": 8400
    },
    {
      "epoch": 0.1919127378942084,
      "grad_norm": 3.296875,
      "learning_rate": 1.6661726303902887e-05,
      "loss": 1.2471,
      "step": 8410
    },
    {
      "epoch": 0.19214093377755465,
      "grad_norm": 2.84375,
      "learning_rate": 1.6657021196508812e-05,
      "loss": 1.2609,
      "step": 8420
    },
    {
      "epoch": 0.19236912966090092,
      "grad_norm": 3.234375,
      "learning_rate": 1.6652316089114734e-05,
      "loss": 1.2476,
      "step": 8430
    },
    {
      "epoch": 0.1925973255442472,
      "grad_norm": 2.765625,
      "learning_rate": 1.664761098172066e-05,
      "loss": 1.2086,
      "step": 8440
    },
    {
      "epoch": 0.19282552142759343,
      "grad_norm": 2.9375,
      "learning_rate": 1.664290587432658e-05,
      "loss": 1.2825,
      "step": 8450
    },
    {
      "epoch": 0.1930537173109397,
      "grad_norm": 3.109375,
      "learning_rate": 1.6638200766932507e-05,
      "loss": 1.2512,
      "step": 8460
    },
    {
      "epoch": 0.19328191319428598,
      "grad_norm": 3.640625,
      "learning_rate": 1.663349565953843e-05,
      "loss": 1.2766,
      "step": 8470
    },
    {
      "epoch": 0.19351010907763225,
      "grad_norm": 2.890625,
      "learning_rate": 1.6628790552144354e-05,
      "loss": 1.1933,
      "step": 8480
    },
    {
      "epoch": 0.1937383049609785,
      "grad_norm": 3.125,
      "learning_rate": 1.6624085444750276e-05,
      "loss": 1.2565,
      "step": 8490
    },
    {
      "epoch": 0.19396650084432476,
      "grad_norm": 3.09375,
      "learning_rate": 1.6619380337356202e-05,
      "loss": 1.2277,
      "step": 8500
    },
    {
      "epoch": 0.19419469672767103,
      "grad_norm": 3.25,
      "learning_rate": 1.6614675229962124e-05,
      "loss": 1.2493,
      "step": 8510
    },
    {
      "epoch": 0.1944228926110173,
      "grad_norm": 2.890625,
      "learning_rate": 1.660997012256805e-05,
      "loss": 1.24,
      "step": 8520
    },
    {
      "epoch": 0.19465108849436355,
      "grad_norm": 3.671875,
      "learning_rate": 1.660526501517397e-05,
      "loss": 1.2416,
      "step": 8530
    },
    {
      "epoch": 0.19487928437770982,
      "grad_norm": 2.8125,
      "learning_rate": 1.6600559907779897e-05,
      "loss": 1.202,
      "step": 8540
    },
    {
      "epoch": 0.1951074802610561,
      "grad_norm": 3.421875,
      "learning_rate": 1.659585480038582e-05,
      "loss": 1.1873,
      "step": 8550
    },
    {
      "epoch": 0.19533567614440236,
      "grad_norm": 3.046875,
      "learning_rate": 1.6591149692991744e-05,
      "loss": 1.2359,
      "step": 8560
    },
    {
      "epoch": 0.1955638720277486,
      "grad_norm": 2.921875,
      "learning_rate": 1.6586444585597666e-05,
      "loss": 1.2217,
      "step": 8570
    },
    {
      "epoch": 0.19579206791109488,
      "grad_norm": 2.859375,
      "learning_rate": 1.658173947820359e-05,
      "loss": 1.2365,
      "step": 8580
    },
    {
      "epoch": 0.19602026379444115,
      "grad_norm": 2.796875,
      "learning_rate": 1.6577034370809513e-05,
      "loss": 1.2042,
      "step": 8590
    },
    {
      "epoch": 0.19624845967778742,
      "grad_norm": 3.0625,
      "learning_rate": 1.657232926341544e-05,
      "loss": 1.1977,
      "step": 8600
    },
    {
      "epoch": 0.19647665556113367,
      "grad_norm": 3.0625,
      "learning_rate": 1.656762415602136e-05,
      "loss": 1.1981,
      "step": 8610
    },
    {
      "epoch": 0.19670485144447994,
      "grad_norm": 3.15625,
      "learning_rate": 1.6562919048627286e-05,
      "loss": 1.252,
      "step": 8620
    },
    {
      "epoch": 0.1969330473278262,
      "grad_norm": 2.703125,
      "learning_rate": 1.6558213941233208e-05,
      "loss": 1.2504,
      "step": 8630
    },
    {
      "epoch": 0.19716124321117248,
      "grad_norm": 3.03125,
      "learning_rate": 1.6553508833839134e-05,
      "loss": 1.2624,
      "step": 8640
    },
    {
      "epoch": 0.19738943909451873,
      "grad_norm": 2.875,
      "learning_rate": 1.654880372644506e-05,
      "loss": 1.226,
      "step": 8650
    },
    {
      "epoch": 0.197617634977865,
      "grad_norm": 2.6875,
      "learning_rate": 1.654409861905098e-05,
      "loss": 1.2627,
      "step": 8660
    },
    {
      "epoch": 0.19784583086121127,
      "grad_norm": 3.140625,
      "learning_rate": 1.6539393511656907e-05,
      "loss": 1.194,
      "step": 8670
    },
    {
      "epoch": 0.19807402674455754,
      "grad_norm": 3.03125,
      "learning_rate": 1.653468840426283e-05,
      "loss": 1.1666,
      "step": 8680
    },
    {
      "epoch": 0.19830222262790378,
      "grad_norm": 2.796875,
      "learning_rate": 1.6529983296868754e-05,
      "loss": 1.224,
      "step": 8690
    },
    {
      "epoch": 0.19853041851125006,
      "grad_norm": 2.859375,
      "learning_rate": 1.6525278189474676e-05,
      "loss": 1.2479,
      "step": 8700
    },
    {
      "epoch": 0.19875861439459633,
      "grad_norm": 3.296875,
      "learning_rate": 1.65205730820806e-05,
      "loss": 1.2069,
      "step": 8710
    },
    {
      "epoch": 0.1989868102779426,
      "grad_norm": 3.1875,
      "learning_rate": 1.6515867974686523e-05,
      "loss": 1.2209,
      "step": 8720
    },
    {
      "epoch": 0.19921500616128884,
      "grad_norm": 2.828125,
      "learning_rate": 1.651116286729245e-05,
      "loss": 1.2077,
      "step": 8730
    },
    {
      "epoch": 0.19944320204463512,
      "grad_norm": 2.828125,
      "learning_rate": 1.650645775989837e-05,
      "loss": 1.2508,
      "step": 8740
    },
    {
      "epoch": 0.1996713979279814,
      "grad_norm": 3.21875,
      "learning_rate": 1.6501752652504296e-05,
      "loss": 1.2285,
      "step": 8750
    },
    {
      "epoch": 0.19989959381132763,
      "grad_norm": 2.671875,
      "learning_rate": 1.6497047545110218e-05,
      "loss": 1.1902,
      "step": 8760
    },
    {
      "epoch": 0.2001277896946739,
      "grad_norm": 3.0625,
      "learning_rate": 1.6492342437716144e-05,
      "loss": 1.2066,
      "step": 8770
    },
    {
      "epoch": 0.20035598557802017,
      "grad_norm": 3.296875,
      "learning_rate": 1.6487637330322066e-05,
      "loss": 1.2115,
      "step": 8780
    },
    {
      "epoch": 0.20058418146136645,
      "grad_norm": 2.828125,
      "learning_rate": 1.648293222292799e-05,
      "loss": 1.1974,
      "step": 8790
    },
    {
      "epoch": 0.2008123773447127,
      "grad_norm": 2.796875,
      "learning_rate": 1.6478227115533913e-05,
      "loss": 1.2178,
      "step": 8800
    },
    {
      "epoch": 0.20104057322805896,
      "grad_norm": 3.265625,
      "learning_rate": 1.647352200813984e-05,
      "loss": 1.2097,
      "step": 8810
    },
    {
      "epoch": 0.20126876911140523,
      "grad_norm": 2.84375,
      "learning_rate": 1.646881690074576e-05,
      "loss": 1.2313,
      "step": 8820
    },
    {
      "epoch": 0.2014969649947515,
      "grad_norm": 2.890625,
      "learning_rate": 1.6464111793351686e-05,
      "loss": 1.2914,
      "step": 8830
    },
    {
      "epoch": 0.20172516087809775,
      "grad_norm": 3.046875,
      "learning_rate": 1.6459406685957608e-05,
      "loss": 1.2425,
      "step": 8840
    },
    {
      "epoch": 0.20195335676144402,
      "grad_norm": 3.109375,
      "learning_rate": 1.6454701578563533e-05,
      "loss": 1.2101,
      "step": 8850
    },
    {
      "epoch": 0.2021815526447903,
      "grad_norm": 3.140625,
      "learning_rate": 1.6449996471169455e-05,
      "loss": 1.2225,
      "step": 8860
    },
    {
      "epoch": 0.20240974852813656,
      "grad_norm": 2.8125,
      "learning_rate": 1.644529136377538e-05,
      "loss": 1.2123,
      "step": 8870
    },
    {
      "epoch": 0.2026379444114828,
      "grad_norm": 3.0,
      "learning_rate": 1.6440586256381303e-05,
      "loss": 1.2377,
      "step": 8880
    },
    {
      "epoch": 0.20286614029482908,
      "grad_norm": 3.1875,
      "learning_rate": 1.6435881148987228e-05,
      "loss": 1.2362,
      "step": 8890
    },
    {
      "epoch": 0.20309433617817535,
      "grad_norm": 3.0,
      "learning_rate": 1.643117604159315e-05,
      "loss": 1.2281,
      "step": 8900
    },
    {
      "epoch": 0.20332253206152162,
      "grad_norm": 3.359375,
      "learning_rate": 1.6426470934199076e-05,
      "loss": 1.2131,
      "step": 8910
    },
    {
      "epoch": 0.20355072794486787,
      "grad_norm": 3.265625,
      "learning_rate": 1.6421765826804998e-05,
      "loss": 1.2275,
      "step": 8920
    },
    {
      "epoch": 0.20377892382821414,
      "grad_norm": 2.984375,
      "learning_rate": 1.6417060719410923e-05,
      "loss": 1.2489,
      "step": 8930
    },
    {
      "epoch": 0.2040071197115604,
      "grad_norm": 3.296875,
      "learning_rate": 1.6412355612016845e-05,
      "loss": 1.2617,
      "step": 8940
    },
    {
      "epoch": 0.20423531559490668,
      "grad_norm": 3.390625,
      "learning_rate": 1.640765050462277e-05,
      "loss": 1.1925,
      "step": 8950
    },
    {
      "epoch": 0.20446351147825292,
      "grad_norm": 2.734375,
      "learning_rate": 1.6402945397228692e-05,
      "loss": 1.2315,
      "step": 8960
    },
    {
      "epoch": 0.2046917073615992,
      "grad_norm": 2.9375,
      "learning_rate": 1.6398240289834618e-05,
      "loss": 1.2203,
      "step": 8970
    },
    {
      "epoch": 0.20491990324494547,
      "grad_norm": 3.40625,
      "learning_rate": 1.639353518244054e-05,
      "loss": 1.2104,
      "step": 8980
    },
    {
      "epoch": 0.20514809912829174,
      "grad_norm": 2.9375,
      "learning_rate": 1.6388830075046465e-05,
      "loss": 1.2316,
      "step": 8990
    },
    {
      "epoch": 0.20537629501163798,
      "grad_norm": 3.203125,
      "learning_rate": 1.6384124967652387e-05,
      "loss": 1.2321,
      "step": 9000
    },
    {
      "epoch": 0.20560449089498425,
      "grad_norm": 2.984375,
      "learning_rate": 1.6379419860258313e-05,
      "loss": 1.2253,
      "step": 9010
    },
    {
      "epoch": 0.20583268677833053,
      "grad_norm": 2.96875,
      "learning_rate": 1.6374714752864235e-05,
      "loss": 1.2575,
      "step": 9020
    },
    {
      "epoch": 0.20606088266167677,
      "grad_norm": 3.015625,
      "learning_rate": 1.637000964547016e-05,
      "loss": 1.1955,
      "step": 9030
    },
    {
      "epoch": 0.20628907854502304,
      "grad_norm": 3.25,
      "learning_rate": 1.6365304538076082e-05,
      "loss": 1.2619,
      "step": 9040
    },
    {
      "epoch": 0.2065172744283693,
      "grad_norm": 2.9375,
      "learning_rate": 1.6360599430682008e-05,
      "loss": 1.1997,
      "step": 9050
    },
    {
      "epoch": 0.20674547031171558,
      "grad_norm": 3.0625,
      "learning_rate": 1.635589432328793e-05,
      "loss": 1.2155,
      "step": 9060
    },
    {
      "epoch": 0.20697366619506183,
      "grad_norm": 2.578125,
      "learning_rate": 1.6351189215893855e-05,
      "loss": 1.2284,
      "step": 9070
    },
    {
      "epoch": 0.2072018620784081,
      "grad_norm": 2.921875,
      "learning_rate": 1.6346484108499777e-05,
      "loss": 1.2549,
      "step": 9080
    },
    {
      "epoch": 0.20743005796175437,
      "grad_norm": 2.921875,
      "learning_rate": 1.6341779001105702e-05,
      "loss": 1.2642,
      "step": 9090
    },
    {
      "epoch": 0.20765825384510064,
      "grad_norm": 3.0625,
      "learning_rate": 1.6337073893711624e-05,
      "loss": 1.252,
      "step": 9100
    },
    {
      "epoch": 0.2078864497284469,
      "grad_norm": 3.0,
      "learning_rate": 1.633236878631755e-05,
      "loss": 1.2527,
      "step": 9110
    },
    {
      "epoch": 0.20811464561179316,
      "grad_norm": 2.796875,
      "learning_rate": 1.6327663678923472e-05,
      "loss": 1.2323,
      "step": 9120
    },
    {
      "epoch": 0.20834284149513943,
      "grad_norm": 3.171875,
      "learning_rate": 1.6322958571529397e-05,
      "loss": 1.2304,
      "step": 9130
    },
    {
      "epoch": 0.2085710373784857,
      "grad_norm": 3.453125,
      "learning_rate": 1.631825346413532e-05,
      "loss": 1.2907,
      "step": 9140
    },
    {
      "epoch": 0.20879923326183195,
      "grad_norm": 2.90625,
      "learning_rate": 1.6313548356741245e-05,
      "loss": 1.2535,
      "step": 9150
    },
    {
      "epoch": 0.20902742914517822,
      "grad_norm": 2.859375,
      "learning_rate": 1.6308843249347167e-05,
      "loss": 1.2341,
      "step": 9160
    },
    {
      "epoch": 0.2092556250285245,
      "grad_norm": 2.796875,
      "learning_rate": 1.6304138141953092e-05,
      "loss": 1.2072,
      "step": 9170
    },
    {
      "epoch": 0.20948382091187076,
      "grad_norm": 3.046875,
      "learning_rate": 1.6299433034559014e-05,
      "loss": 1.2074,
      "step": 9180
    },
    {
      "epoch": 0.209712016795217,
      "grad_norm": 2.84375,
      "learning_rate": 1.629472792716494e-05,
      "loss": 1.2382,
      "step": 9190
    },
    {
      "epoch": 0.20994021267856328,
      "grad_norm": 2.984375,
      "learning_rate": 1.629002281977086e-05,
      "loss": 1.2143,
      "step": 9200
    },
    {
      "epoch": 0.21016840856190955,
      "grad_norm": 2.953125,
      "learning_rate": 1.6285317712376787e-05,
      "loss": 1.2377,
      "step": 9210
    },
    {
      "epoch": 0.21039660444525582,
      "grad_norm": 3.09375,
      "learning_rate": 1.628061260498271e-05,
      "loss": 1.2578,
      "step": 9220
    },
    {
      "epoch": 0.21062480032860206,
      "grad_norm": 3.296875,
      "learning_rate": 1.6275907497588634e-05,
      "loss": 1.2393,
      "step": 9230
    },
    {
      "epoch": 0.21085299621194833,
      "grad_norm": 3.015625,
      "learning_rate": 1.627120239019456e-05,
      "loss": 1.2501,
      "step": 9240
    },
    {
      "epoch": 0.2110811920952946,
      "grad_norm": 3.015625,
      "learning_rate": 1.6266497282800482e-05,
      "loss": 1.2043,
      "step": 9250
    },
    {
      "epoch": 0.21130938797864088,
      "grad_norm": 2.734375,
      "learning_rate": 1.6261792175406407e-05,
      "loss": 1.2088,
      "step": 9260
    },
    {
      "epoch": 0.21153758386198712,
      "grad_norm": 2.9375,
      "learning_rate": 1.625708706801233e-05,
      "loss": 1.1972,
      "step": 9270
    },
    {
      "epoch": 0.2117657797453334,
      "grad_norm": 3.03125,
      "learning_rate": 1.6252381960618255e-05,
      "loss": 1.2615,
      "step": 9280
    },
    {
      "epoch": 0.21199397562867967,
      "grad_norm": 3.140625,
      "learning_rate": 1.6247676853224177e-05,
      "loss": 1.2385,
      "step": 9290
    },
    {
      "epoch": 0.21222217151202594,
      "grad_norm": 3.0,
      "learning_rate": 1.6242971745830102e-05,
      "loss": 1.1936,
      "step": 9300
    },
    {
      "epoch": 0.21245036739537218,
      "grad_norm": 2.921875,
      "learning_rate": 1.6238266638436024e-05,
      "loss": 1.2059,
      "step": 9310
    },
    {
      "epoch": 0.21267856327871845,
      "grad_norm": 2.84375,
      "learning_rate": 1.623356153104195e-05,
      "loss": 1.1747,
      "step": 9320
    },
    {
      "epoch": 0.21290675916206472,
      "grad_norm": 2.953125,
      "learning_rate": 1.622885642364787e-05,
      "loss": 1.2798,
      "step": 9330
    },
    {
      "epoch": 0.21313495504541097,
      "grad_norm": 3.046875,
      "learning_rate": 1.6224151316253797e-05,
      "loss": 1.1945,
      "step": 9340
    },
    {
      "epoch": 0.21336315092875724,
      "grad_norm": 2.78125,
      "learning_rate": 1.621944620885972e-05,
      "loss": 1.2655,
      "step": 9350
    },
    {
      "epoch": 0.2135913468121035,
      "grad_norm": 2.96875,
      "learning_rate": 1.6214741101465644e-05,
      "loss": 1.2627,
      "step": 9360
    },
    {
      "epoch": 0.21381954269544978,
      "grad_norm": 3.109375,
      "learning_rate": 1.6210035994071566e-05,
      "loss": 1.2118,
      "step": 9370
    },
    {
      "epoch": 0.21404773857879603,
      "grad_norm": 3.046875,
      "learning_rate": 1.6205330886677492e-05,
      "loss": 1.2575,
      "step": 9380
    },
    {
      "epoch": 0.2142759344621423,
      "grad_norm": 3.15625,
      "learning_rate": 1.6200625779283414e-05,
      "loss": 1.2487,
      "step": 9390
    },
    {
      "epoch": 0.21450413034548857,
      "grad_norm": 3.0625,
      "learning_rate": 1.619592067188934e-05,
      "loss": 1.2209,
      "step": 9400
    },
    {
      "epoch": 0.21473232622883484,
      "grad_norm": 2.875,
      "learning_rate": 1.619121556449526e-05,
      "loss": 1.2439,
      "step": 9410
    },
    {
      "epoch": 0.21496052211218108,
      "grad_norm": 2.765625,
      "learning_rate": 1.6186510457101187e-05,
      "loss": 1.2373,
      "step": 9420
    },
    {
      "epoch": 0.21518871799552736,
      "grad_norm": 2.828125,
      "learning_rate": 1.618180534970711e-05,
      "loss": 1.2356,
      "step": 9430
    },
    {
      "epoch": 0.21541691387887363,
      "grad_norm": 2.828125,
      "learning_rate": 1.6177100242313034e-05,
      "loss": 1.2488,
      "step": 9440
    },
    {
      "epoch": 0.2156451097622199,
      "grad_norm": 2.84375,
      "learning_rate": 1.6172395134918956e-05,
      "loss": 1.2302,
      "step": 9450
    },
    {
      "epoch": 0.21587330564556614,
      "grad_norm": 3.0625,
      "learning_rate": 1.616769002752488e-05,
      "loss": 1.2619,
      "step": 9460
    },
    {
      "epoch": 0.21610150152891242,
      "grad_norm": 3.0,
      "learning_rate": 1.6162984920130804e-05,
      "loss": 1.2002,
      "step": 9470
    },
    {
      "epoch": 0.2163296974122587,
      "grad_norm": 3.4375,
      "learning_rate": 1.615827981273673e-05,
      "loss": 1.2303,
      "step": 9480
    },
    {
      "epoch": 0.21655789329560496,
      "grad_norm": 3.0625,
      "learning_rate": 1.615357470534265e-05,
      "loss": 1.261,
      "step": 9490
    },
    {
      "epoch": 0.2167860891789512,
      "grad_norm": 2.75,
      "learning_rate": 1.6148869597948576e-05,
      "loss": 1.2069,
      "step": 9500
    },
    {
      "epoch": 0.21701428506229747,
      "grad_norm": 3.1875,
      "learning_rate": 1.61441644905545e-05,
      "loss": 1.2473,
      "step": 9510
    },
    {
      "epoch": 0.21724248094564375,
      "grad_norm": 3.046875,
      "learning_rate": 1.6139459383160424e-05,
      "loss": 1.2196,
      "step": 9520
    },
    {
      "epoch": 0.21747067682899002,
      "grad_norm": 2.953125,
      "learning_rate": 1.6134754275766346e-05,
      "loss": 1.2077,
      "step": 9530
    },
    {
      "epoch": 0.21769887271233626,
      "grad_norm": 3.046875,
      "learning_rate": 1.613004916837227e-05,
      "loss": 1.1796,
      "step": 9540
    },
    {
      "epoch": 0.21792706859568253,
      "grad_norm": 3.296875,
      "learning_rate": 1.6125344060978193e-05,
      "loss": 1.2555,
      "step": 9550
    },
    {
      "epoch": 0.2181552644790288,
      "grad_norm": 3.140625,
      "learning_rate": 1.612063895358412e-05,
      "loss": 1.2043,
      "step": 9560
    },
    {
      "epoch": 0.21838346036237508,
      "grad_norm": 3.09375,
      "learning_rate": 1.611593384619004e-05,
      "loss": 1.2371,
      "step": 9570
    },
    {
      "epoch": 0.21861165624572132,
      "grad_norm": 3.015625,
      "learning_rate": 1.6111228738795966e-05,
      "loss": 1.2308,
      "step": 9580
    },
    {
      "epoch": 0.2188398521290676,
      "grad_norm": 2.8125,
      "learning_rate": 1.6106523631401888e-05,
      "loss": 1.2808,
      "step": 9590
    },
    {
      "epoch": 0.21906804801241386,
      "grad_norm": 2.8125,
      "learning_rate": 1.6101818524007813e-05,
      "loss": 1.1971,
      "step": 9600
    },
    {
      "epoch": 0.2192962438957601,
      "grad_norm": 3.0625,
      "learning_rate": 1.6097113416613735e-05,
      "loss": 1.2058,
      "step": 9610
    },
    {
      "epoch": 0.21952443977910638,
      "grad_norm": 3.046875,
      "learning_rate": 1.609240830921966e-05,
      "loss": 1.2024,
      "step": 9620
    },
    {
      "epoch": 0.21975263566245265,
      "grad_norm": 3.015625,
      "learning_rate": 1.6087703201825583e-05,
      "loss": 1.1794,
      "step": 9630
    },
    {
      "epoch": 0.21998083154579892,
      "grad_norm": 3.375,
      "learning_rate": 1.6082998094431505e-05,
      "loss": 1.2134,
      "step": 9640
    },
    {
      "epoch": 0.22020902742914517,
      "grad_norm": 3.03125,
      "learning_rate": 1.607829298703743e-05,
      "loss": 1.1885,
      "step": 9650
    },
    {
      "epoch": 0.22043722331249144,
      "grad_norm": 3.109375,
      "learning_rate": 1.6073587879643352e-05,
      "loss": 1.2228,
      "step": 9660
    },
    {
      "epoch": 0.2206654191958377,
      "grad_norm": 2.859375,
      "learning_rate": 1.6068882772249278e-05,
      "loss": 1.202,
      "step": 9670
    },
    {
      "epoch": 0.22089361507918398,
      "grad_norm": 2.9375,
      "learning_rate": 1.60641776648552e-05,
      "loss": 1.2108,
      "step": 9680
    },
    {
      "epoch": 0.22112181096253022,
      "grad_norm": 3.03125,
      "learning_rate": 1.6059472557461125e-05,
      "loss": 1.2458,
      "step": 9690
    },
    {
      "epoch": 0.2213500068458765,
      "grad_norm": 3.0,
      "learning_rate": 1.6054767450067047e-05,
      "loss": 1.2029,
      "step": 9700
    },
    {
      "epoch": 0.22157820272922277,
      "grad_norm": 3.015625,
      "learning_rate": 1.6050062342672973e-05,
      "loss": 1.1826,
      "step": 9710
    },
    {
      "epoch": 0.22180639861256904,
      "grad_norm": 2.9375,
      "learning_rate": 1.6045357235278895e-05,
      "loss": 1.2282,
      "step": 9720
    },
    {
      "epoch": 0.22203459449591528,
      "grad_norm": 3.28125,
      "learning_rate": 1.604065212788482e-05,
      "loss": 1.2441,
      "step": 9730
    },
    {
      "epoch": 0.22226279037926155,
      "grad_norm": 3.1875,
      "learning_rate": 1.6035947020490742e-05,
      "loss": 1.2339,
      "step": 9740
    },
    {
      "epoch": 0.22249098626260783,
      "grad_norm": 3.296875,
      "learning_rate": 1.6031241913096667e-05,
      "loss": 1.2271,
      "step": 9750
    },
    {
      "epoch": 0.2227191821459541,
      "grad_norm": 3.078125,
      "learning_rate": 1.602653680570259e-05,
      "loss": 1.2271,
      "step": 9760
    },
    {
      "epoch": 0.22294737802930034,
      "grad_norm": 3.1875,
      "learning_rate": 1.6021831698308515e-05,
      "loss": 1.2855,
      "step": 9770
    },
    {
      "epoch": 0.2231755739126466,
      "grad_norm": 3.0,
      "learning_rate": 1.6017126590914437e-05,
      "loss": 1.2563,
      "step": 9780
    },
    {
      "epoch": 0.22340376979599288,
      "grad_norm": 2.875,
      "learning_rate": 1.6012421483520362e-05,
      "loss": 1.2667,
      "step": 9790
    },
    {
      "epoch": 0.22363196567933916,
      "grad_norm": 2.953125,
      "learning_rate": 1.6007716376126284e-05,
      "loss": 1.2325,
      "step": 9800
    },
    {
      "epoch": 0.2238601615626854,
      "grad_norm": 3.140625,
      "learning_rate": 1.600301126873221e-05,
      "loss": 1.2873,
      "step": 9810
    },
    {
      "epoch": 0.22408835744603167,
      "grad_norm": 2.96875,
      "learning_rate": 1.5998306161338132e-05,
      "loss": 1.2253,
      "step": 9820
    },
    {
      "epoch": 0.22431655332937794,
      "grad_norm": 2.9375,
      "learning_rate": 1.5993601053944057e-05,
      "loss": 1.219,
      "step": 9830
    },
    {
      "epoch": 0.22454474921272422,
      "grad_norm": 3.0,
      "learning_rate": 1.598889594654998e-05,
      "loss": 1.2209,
      "step": 9840
    },
    {
      "epoch": 0.22477294509607046,
      "grad_norm": 3.15625,
      "learning_rate": 1.5984190839155905e-05,
      "loss": 1.2125,
      "step": 9850
    },
    {
      "epoch": 0.22500114097941673,
      "grad_norm": 3.03125,
      "learning_rate": 1.5979485731761827e-05,
      "loss": 1.2451,
      "step": 9860
    },
    {
      "epoch": 0.225229336862763,
      "grad_norm": 3.03125,
      "learning_rate": 1.5974780624367752e-05,
      "loss": 1.2371,
      "step": 9870
    },
    {
      "epoch": 0.22545753274610927,
      "grad_norm": 3.078125,
      "learning_rate": 1.5970075516973674e-05,
      "loss": 1.2315,
      "step": 9880
    },
    {
      "epoch": 0.22568572862945552,
      "grad_norm": 3.21875,
      "learning_rate": 1.59653704095796e-05,
      "loss": 1.2656,
      "step": 9890
    },
    {
      "epoch": 0.2259139245128018,
      "grad_norm": 3.015625,
      "learning_rate": 1.596066530218552e-05,
      "loss": 1.1967,
      "step": 9900
    },
    {
      "epoch": 0.22614212039614806,
      "grad_norm": 3.0,
      "learning_rate": 1.5955960194791447e-05,
      "loss": 1.1916,
      "step": 9910
    },
    {
      "epoch": 0.2263703162794943,
      "grad_norm": 3.015625,
      "learning_rate": 1.595125508739737e-05,
      "loss": 1.2228,
      "step": 9920
    },
    {
      "epoch": 0.22659851216284058,
      "grad_norm": 3.046875,
      "learning_rate": 1.5946549980003294e-05,
      "loss": 1.2501,
      "step": 9930
    },
    {
      "epoch": 0.22682670804618685,
      "grad_norm": 3.15625,
      "learning_rate": 1.5941844872609216e-05,
      "loss": 1.2497,
      "step": 9940
    },
    {
      "epoch": 0.22705490392953312,
      "grad_norm": 3.3125,
      "learning_rate": 1.5937139765215142e-05,
      "loss": 1.1685,
      "step": 9950
    },
    {
      "epoch": 0.22728309981287936,
      "grad_norm": 3.09375,
      "learning_rate": 1.5932434657821064e-05,
      "loss": 1.2034,
      "step": 9960
    },
    {
      "epoch": 0.22751129569622563,
      "grad_norm": 3.078125,
      "learning_rate": 1.592772955042699e-05,
      "loss": 1.1938,
      "step": 9970
    },
    {
      "epoch": 0.2277394915795719,
      "grad_norm": 3.015625,
      "learning_rate": 1.592302444303291e-05,
      "loss": 1.2094,
      "step": 9980
    },
    {
      "epoch": 0.22796768746291818,
      "grad_norm": 3.109375,
      "learning_rate": 1.5918319335638837e-05,
      "loss": 1.2194,
      "step": 9990
    },
    {
      "epoch": 0.22819588334626442,
      "grad_norm": 2.984375,
      "learning_rate": 1.5913614228244762e-05,
      "loss": 1.2077,
      "step": 10000
    },
    {
      "epoch": 0.2284240792296107,
      "grad_norm": 3.046875,
      "learning_rate": 1.5908909120850684e-05,
      "loss": 1.2208,
      "step": 10010
    },
    {
      "epoch": 0.22865227511295697,
      "grad_norm": 3.109375,
      "learning_rate": 1.590420401345661e-05,
      "loss": 1.2242,
      "step": 10020
    },
    {
      "epoch": 0.22888047099630324,
      "grad_norm": 3.125,
      "learning_rate": 1.589949890606253e-05,
      "loss": 1.2118,
      "step": 10030
    },
    {
      "epoch": 0.22910866687964948,
      "grad_norm": 3.046875,
      "learning_rate": 1.5894793798668457e-05,
      "loss": 1.1964,
      "step": 10040
    },
    {
      "epoch": 0.22933686276299575,
      "grad_norm": 2.859375,
      "learning_rate": 1.589008869127438e-05,
      "loss": 1.2671,
      "step": 10050
    },
    {
      "epoch": 0.22956505864634202,
      "grad_norm": 3.359375,
      "learning_rate": 1.5885383583880304e-05,
      "loss": 1.2592,
      "step": 10060
    },
    {
      "epoch": 0.2297932545296883,
      "grad_norm": 2.859375,
      "learning_rate": 1.5880678476486226e-05,
      "loss": 1.2516,
      "step": 10070
    },
    {
      "epoch": 0.23002145041303454,
      "grad_norm": 3.0625,
      "learning_rate": 1.587597336909215e-05,
      "loss": 1.2485,
      "step": 10080
    },
    {
      "epoch": 0.2302496462963808,
      "grad_norm": 3.1875,
      "learning_rate": 1.5871268261698074e-05,
      "loss": 1.1679,
      "step": 10090
    },
    {
      "epoch": 0.23047784217972708,
      "grad_norm": 2.890625,
      "learning_rate": 1.5866563154304e-05,
      "loss": 1.2396,
      "step": 10100
    },
    {
      "epoch": 0.23070603806307335,
      "grad_norm": 2.921875,
      "learning_rate": 1.586185804690992e-05,
      "loss": 1.1536,
      "step": 10110
    },
    {
      "epoch": 0.2309342339464196,
      "grad_norm": 3.09375,
      "learning_rate": 1.5857152939515847e-05,
      "loss": 1.1746,
      "step": 10120
    },
    {
      "epoch": 0.23116242982976587,
      "grad_norm": 3.09375,
      "learning_rate": 1.585244783212177e-05,
      "loss": 1.2041,
      "step": 10130
    },
    {
      "epoch": 0.23139062571311214,
      "grad_norm": 2.609375,
      "learning_rate": 1.5847742724727694e-05,
      "loss": 1.1882,
      "step": 10140
    },
    {
      "epoch": 0.2316188215964584,
      "grad_norm": 2.703125,
      "learning_rate": 1.5843037617333616e-05,
      "loss": 1.2003,
      "step": 10150
    },
    {
      "epoch": 0.23184701747980466,
      "grad_norm": 3.0,
      "learning_rate": 1.583833250993954e-05,
      "loss": 1.2207,
      "step": 10160
    },
    {
      "epoch": 0.23207521336315093,
      "grad_norm": 2.8125,
      "learning_rate": 1.5833627402545463e-05,
      "loss": 1.2495,
      "step": 10170
    },
    {
      "epoch": 0.2323034092464972,
      "grad_norm": 3.296875,
      "learning_rate": 1.582892229515139e-05,
      "loss": 1.2113,
      "step": 10180
    },
    {
      "epoch": 0.23253160512984344,
      "grad_norm": 3.125,
      "learning_rate": 1.582421718775731e-05,
      "loss": 1.2979,
      "step": 10190
    },
    {
      "epoch": 0.23275980101318972,
      "grad_norm": 3.03125,
      "learning_rate": 1.5819512080363236e-05,
      "loss": 1.2664,
      "step": 10200
    },
    {
      "epoch": 0.232987996896536,
      "grad_norm": 3.1875,
      "learning_rate": 1.5814806972969158e-05,
      "loss": 1.2043,
      "step": 10210
    },
    {
      "epoch": 0.23321619277988226,
      "grad_norm": 2.75,
      "learning_rate": 1.5810101865575084e-05,
      "loss": 1.2391,
      "step": 10220
    },
    {
      "epoch": 0.2334443886632285,
      "grad_norm": 3.109375,
      "learning_rate": 1.5805396758181006e-05,
      "loss": 1.2001,
      "step": 10230
    },
    {
      "epoch": 0.23367258454657477,
      "grad_norm": 2.796875,
      "learning_rate": 1.580069165078693e-05,
      "loss": 1.1994,
      "step": 10240
    },
    {
      "epoch": 0.23390078042992105,
      "grad_norm": 2.890625,
      "learning_rate": 1.5795986543392853e-05,
      "loss": 1.1867,
      "step": 10250
    },
    {
      "epoch": 0.23412897631326732,
      "grad_norm": 2.734375,
      "learning_rate": 1.579128143599878e-05,
      "loss": 1.2427,
      "step": 10260
    },
    {
      "epoch": 0.23435717219661356,
      "grad_norm": 2.859375,
      "learning_rate": 1.57865763286047e-05,
      "loss": 1.2284,
      "step": 10270
    },
    {
      "epoch": 0.23458536807995983,
      "grad_norm": 2.890625,
      "learning_rate": 1.5781871221210626e-05,
      "loss": 1.2205,
      "step": 10280
    },
    {
      "epoch": 0.2348135639633061,
      "grad_norm": 3.046875,
      "learning_rate": 1.5777166113816548e-05,
      "loss": 1.234,
      "step": 10290
    },
    {
      "epoch": 0.23504175984665238,
      "grad_norm": 3.046875,
      "learning_rate": 1.5772461006422473e-05,
      "loss": 1.2435,
      "step": 10300
    },
    {
      "epoch": 0.23526995572999862,
      "grad_norm": 2.84375,
      "learning_rate": 1.5767755899028395e-05,
      "loss": 1.2304,
      "step": 10310
    },
    {
      "epoch": 0.2354981516133449,
      "grad_norm": 2.984375,
      "learning_rate": 1.576305079163432e-05,
      "loss": 1.2045,
      "step": 10320
    },
    {
      "epoch": 0.23572634749669116,
      "grad_norm": 2.8125,
      "learning_rate": 1.5758345684240243e-05,
      "loss": 1.2271,
      "step": 10330
    },
    {
      "epoch": 0.23595454338003743,
      "grad_norm": 3.171875,
      "learning_rate": 1.5753640576846168e-05,
      "loss": 1.1787,
      "step": 10340
    },
    {
      "epoch": 0.23618273926338368,
      "grad_norm": 2.921875,
      "learning_rate": 1.574893546945209e-05,
      "loss": 1.2273,
      "step": 10350
    },
    {
      "epoch": 0.23641093514672995,
      "grad_norm": 2.9375,
      "learning_rate": 1.5744230362058016e-05,
      "loss": 1.2261,
      "step": 10360
    },
    {
      "epoch": 0.23663913103007622,
      "grad_norm": 3.0,
      "learning_rate": 1.5739525254663938e-05,
      "loss": 1.2406,
      "step": 10370
    },
    {
      "epoch": 0.2368673269134225,
      "grad_norm": 3.109375,
      "learning_rate": 1.5734820147269863e-05,
      "loss": 1.2096,
      "step": 10380
    },
    {
      "epoch": 0.23709552279676874,
      "grad_norm": 3.015625,
      "learning_rate": 1.5730115039875785e-05,
      "loss": 1.1715,
      "step": 10390
    },
    {
      "epoch": 0.237323718680115,
      "grad_norm": 2.8125,
      "learning_rate": 1.572540993248171e-05,
      "loss": 1.1993,
      "step": 10400
    },
    {
      "epoch": 0.23755191456346128,
      "grad_norm": 3.078125,
      "learning_rate": 1.5720704825087632e-05,
      "loss": 1.2547,
      "step": 10410
    },
    {
      "epoch": 0.23778011044680755,
      "grad_norm": 3.109375,
      "learning_rate": 1.5715999717693558e-05,
      "loss": 1.2577,
      "step": 10420
    },
    {
      "epoch": 0.2380083063301538,
      "grad_norm": 3.09375,
      "learning_rate": 1.571129461029948e-05,
      "loss": 1.2618,
      "step": 10430
    },
    {
      "epoch": 0.23823650221350007,
      "grad_norm": 2.796875,
      "learning_rate": 1.5706589502905405e-05,
      "loss": 1.1878,
      "step": 10440
    },
    {
      "epoch": 0.23846469809684634,
      "grad_norm": 3.046875,
      "learning_rate": 1.5701884395511327e-05,
      "loss": 1.2325,
      "step": 10450
    },
    {
      "epoch": 0.2386928939801926,
      "grad_norm": 2.875,
      "learning_rate": 1.5697179288117253e-05,
      "loss": 1.1809,
      "step": 10460
    },
    {
      "epoch": 0.23892108986353885,
      "grad_norm": 2.953125,
      "learning_rate": 1.5692474180723175e-05,
      "loss": 1.3004,
      "step": 10470
    },
    {
      "epoch": 0.23914928574688513,
      "grad_norm": 3.0,
      "learning_rate": 1.56877690733291e-05,
      "loss": 1.2388,
      "step": 10480
    },
    {
      "epoch": 0.2393774816302314,
      "grad_norm": 2.984375,
      "learning_rate": 1.5683063965935022e-05,
      "loss": 1.2123,
      "step": 10490
    },
    {
      "epoch": 0.23960567751357764,
      "grad_norm": 2.859375,
      "learning_rate": 1.5678358858540948e-05,
      "loss": 1.1964,
      "step": 10500
    },
    {
      "epoch": 0.2398338733969239,
      "grad_norm": 3.390625,
      "learning_rate": 1.567365375114687e-05,
      "loss": 1.2141,
      "step": 10510
    },
    {
      "epoch": 0.24006206928027018,
      "grad_norm": 2.796875,
      "learning_rate": 1.5668948643752795e-05,
      "loss": 1.17,
      "step": 10520
    },
    {
      "epoch": 0.24029026516361646,
      "grad_norm": 2.9375,
      "learning_rate": 1.5664243536358717e-05,
      "loss": 1.2269,
      "step": 10530
    },
    {
      "epoch": 0.2405184610469627,
      "grad_norm": 3.0625,
      "learning_rate": 1.5659538428964642e-05,
      "loss": 1.2257,
      "step": 10540
    },
    {
      "epoch": 0.24074665693030897,
      "grad_norm": 3.265625,
      "learning_rate": 1.5654833321570564e-05,
      "loss": 1.263,
      "step": 10550
    },
    {
      "epoch": 0.24097485281365524,
      "grad_norm": 2.96875,
      "learning_rate": 1.565012821417649e-05,
      "loss": 1.2121,
      "step": 10560
    },
    {
      "epoch": 0.24120304869700152,
      "grad_norm": 2.703125,
      "learning_rate": 1.5645423106782415e-05,
      "loss": 1.1657,
      "step": 10570
    },
    {
      "epoch": 0.24143124458034776,
      "grad_norm": 3.0625,
      "learning_rate": 1.5640717999388337e-05,
      "loss": 1.2295,
      "step": 10580
    },
    {
      "epoch": 0.24165944046369403,
      "grad_norm": 3.171875,
      "learning_rate": 1.5636012891994263e-05,
      "loss": 1.2416,
      "step": 10590
    },
    {
      "epoch": 0.2418876363470403,
      "grad_norm": 3.125,
      "learning_rate": 1.5631307784600185e-05,
      "loss": 1.1164,
      "step": 10600
    },
    {
      "epoch": 0.24211583223038657,
      "grad_norm": 2.9375,
      "learning_rate": 1.562660267720611e-05,
      "loss": 1.2456,
      "step": 10610
    },
    {
      "epoch": 0.24234402811373282,
      "grad_norm": 3.03125,
      "learning_rate": 1.5621897569812032e-05,
      "loss": 1.2137,
      "step": 10620
    },
    {
      "epoch": 0.2425722239970791,
      "grad_norm": 3.125,
      "learning_rate": 1.5617192462417958e-05,
      "loss": 1.2055,
      "step": 10630
    },
    {
      "epoch": 0.24280041988042536,
      "grad_norm": 2.828125,
      "learning_rate": 1.561248735502388e-05,
      "loss": 1.2155,
      "step": 10640
    },
    {
      "epoch": 0.24302861576377163,
      "grad_norm": 3.109375,
      "learning_rate": 1.5607782247629805e-05,
      "loss": 1.2164,
      "step": 10650
    },
    {
      "epoch": 0.24325681164711788,
      "grad_norm": 3.015625,
      "learning_rate": 1.5603077140235727e-05,
      "loss": 1.2582,
      "step": 10660
    },
    {
      "epoch": 0.24348500753046415,
      "grad_norm": 3.1875,
      "learning_rate": 1.5598372032841652e-05,
      "loss": 1.2111,
      "step": 10670
    },
    {
      "epoch": 0.24371320341381042,
      "grad_norm": 2.9375,
      "learning_rate": 1.5593666925447574e-05,
      "loss": 1.2469,
      "step": 10680
    },
    {
      "epoch": 0.2439413992971567,
      "grad_norm": 2.75,
      "learning_rate": 1.55889618180535e-05,
      "loss": 1.1767,
      "step": 10690
    },
    {
      "epoch": 0.24416959518050294,
      "grad_norm": 4.03125,
      "learning_rate": 1.5584256710659422e-05,
      "loss": 1.2617,
      "step": 10700
    },
    {
      "epoch": 0.2443977910638492,
      "grad_norm": 3.015625,
      "learning_rate": 1.5579551603265347e-05,
      "loss": 1.2542,
      "step": 10710
    },
    {
      "epoch": 0.24462598694719548,
      "grad_norm": 2.984375,
      "learning_rate": 1.557484649587127e-05,
      "loss": 1.2068,
      "step": 10720
    },
    {
      "epoch": 0.24485418283054175,
      "grad_norm": 3.109375,
      "learning_rate": 1.5570141388477195e-05,
      "loss": 1.1685,
      "step": 10730
    },
    {
      "epoch": 0.245082378713888,
      "grad_norm": 3.0,
      "learning_rate": 1.5565436281083117e-05,
      "loss": 1.1642,
      "step": 10740
    },
    {
      "epoch": 0.24531057459723427,
      "grad_norm": 2.921875,
      "learning_rate": 1.5560731173689042e-05,
      "loss": 1.1956,
      "step": 10750
    },
    {
      "epoch": 0.24553877048058054,
      "grad_norm": 2.84375,
      "learning_rate": 1.5556026066294964e-05,
      "loss": 1.1974,
      "step": 10760
    },
    {
      "epoch": 0.24576696636392678,
      "grad_norm": 2.875,
      "learning_rate": 1.555132095890089e-05,
      "loss": 1.1999,
      "step": 10770
    },
    {
      "epoch": 0.24599516224727305,
      "grad_norm": 3.265625,
      "learning_rate": 1.554661585150681e-05,
      "loss": 1.2033,
      "step": 10780
    },
    {
      "epoch": 0.24622335813061932,
      "grad_norm": 3.140625,
      "learning_rate": 1.5541910744112737e-05,
      "loss": 1.2514,
      "step": 10790
    },
    {
      "epoch": 0.2464515540139656,
      "grad_norm": 3.28125,
      "learning_rate": 1.553720563671866e-05,
      "loss": 1.2211,
      "step": 10800
    },
    {
      "epoch": 0.24667974989731184,
      "grad_norm": 3.0,
      "learning_rate": 1.5532500529324584e-05,
      "loss": 1.2156,
      "step": 10810
    },
    {
      "epoch": 0.2469079457806581,
      "grad_norm": 3.0625,
      "learning_rate": 1.5527795421930506e-05,
      "loss": 1.2073,
      "step": 10820
    },
    {
      "epoch": 0.24713614166400438,
      "grad_norm": 3.140625,
      "learning_rate": 1.5523090314536432e-05,
      "loss": 1.2383,
      "step": 10830
    },
    {
      "epoch": 0.24736433754735065,
      "grad_norm": 3.109375,
      "learning_rate": 1.5518385207142354e-05,
      "loss": 1.187,
      "step": 10840
    },
    {
      "epoch": 0.2475925334306969,
      "grad_norm": 3.078125,
      "learning_rate": 1.551368009974828e-05,
      "loss": 1.2681,
      "step": 10850
    },
    {
      "epoch": 0.24782072931404317,
      "grad_norm": 2.953125,
      "learning_rate": 1.55089749923542e-05,
      "loss": 1.2341,
      "step": 10860
    },
    {
      "epoch": 0.24804892519738944,
      "grad_norm": 3.0,
      "learning_rate": 1.5504269884960127e-05,
      "loss": 1.2122,
      "step": 10870
    },
    {
      "epoch": 0.2482771210807357,
      "grad_norm": 2.828125,
      "learning_rate": 1.549956477756605e-05,
      "loss": 1.2273,
      "step": 10880
    },
    {
      "epoch": 0.24850531696408196,
      "grad_norm": 2.84375,
      "learning_rate": 1.5494859670171974e-05,
      "loss": 1.2278,
      "step": 10890
    },
    {
      "epoch": 0.24873351284742823,
      "grad_norm": 3.0,
      "learning_rate": 1.5490154562777896e-05,
      "loss": 1.2284,
      "step": 10900
    },
    {
      "epoch": 0.2489617087307745,
      "grad_norm": 3.0,
      "learning_rate": 1.548544945538382e-05,
      "loss": 1.2208,
      "step": 10910
    },
    {
      "epoch": 0.24918990461412077,
      "grad_norm": 3.015625,
      "learning_rate": 1.5480744347989743e-05,
      "loss": 1.2392,
      "step": 10920
    },
    {
      "epoch": 0.24941810049746702,
      "grad_norm": 2.9375,
      "learning_rate": 1.547603924059567e-05,
      "loss": 1.1777,
      "step": 10930
    },
    {
      "epoch": 0.2496462963808133,
      "grad_norm": 3.109375,
      "learning_rate": 1.547133413320159e-05,
      "loss": 1.2712,
      "step": 10940
    },
    {
      "epoch": 0.24987449226415956,
      "grad_norm": 3.03125,
      "learning_rate": 1.5466629025807516e-05,
      "loss": 1.2382,
      "step": 10950
    },
    {
      "epoch": 0.2501026881475058,
      "grad_norm": 2.890625,
      "learning_rate": 1.546192391841344e-05,
      "loss": 1.2482,
      "step": 10960
    },
    {
      "epoch": 0.2503308840308521,
      "grad_norm": 3.25,
      "learning_rate": 1.5457218811019364e-05,
      "loss": 1.2621,
      "step": 10970
    },
    {
      "epoch": 0.25055907991419835,
      "grad_norm": 3.0,
      "learning_rate": 1.5452513703625286e-05,
      "loss": 1.1997,
      "step": 10980
    },
    {
      "epoch": 0.2507872757975446,
      "grad_norm": 3.140625,
      "learning_rate": 1.544780859623121e-05,
      "loss": 1.193,
      "step": 10990
    },
    {
      "epoch": 0.2510154716808909,
      "grad_norm": 3.234375,
      "learning_rate": 1.5443103488837133e-05,
      "loss": 1.1715,
      "step": 11000
    },
    {
      "epoch": 0.25124366756423716,
      "grad_norm": 3.265625,
      "learning_rate": 1.543839838144306e-05,
      "loss": 1.1873,
      "step": 11010
    },
    {
      "epoch": 0.25147186344758343,
      "grad_norm": 2.875,
      "learning_rate": 1.543369327404898e-05,
      "loss": 1.1859,
      "step": 11020
    },
    {
      "epoch": 0.25170005933092965,
      "grad_norm": 2.734375,
      "learning_rate": 1.5428988166654906e-05,
      "loss": 1.2123,
      "step": 11030
    },
    {
      "epoch": 0.2519282552142759,
      "grad_norm": 3.171875,
      "learning_rate": 1.5424283059260828e-05,
      "loss": 1.1889,
      "step": 11040
    },
    {
      "epoch": 0.2521564510976222,
      "grad_norm": 3.03125,
      "learning_rate": 1.5419577951866753e-05,
      "loss": 1.2163,
      "step": 11050
    },
    {
      "epoch": 0.25238464698096846,
      "grad_norm": 3.75,
      "learning_rate": 1.5414872844472675e-05,
      "loss": 1.1763,
      "step": 11060
    },
    {
      "epoch": 0.25261284286431474,
      "grad_norm": 3.09375,
      "learning_rate": 1.54101677370786e-05,
      "loss": 1.222,
      "step": 11070
    },
    {
      "epoch": 0.252841038747661,
      "grad_norm": 3.171875,
      "learning_rate": 1.5405462629684523e-05,
      "loss": 1.2512,
      "step": 11080
    },
    {
      "epoch": 0.2530692346310073,
      "grad_norm": 3.046875,
      "learning_rate": 1.5400757522290448e-05,
      "loss": 1.2386,
      "step": 11090
    },
    {
      "epoch": 0.2532974305143535,
      "grad_norm": 3.234375,
      "learning_rate": 1.539605241489637e-05,
      "loss": 1.2165,
      "step": 11100
    },
    {
      "epoch": 0.25352562639769977,
      "grad_norm": 3.171875,
      "learning_rate": 1.5391347307502296e-05,
      "loss": 1.239,
      "step": 11110
    },
    {
      "epoch": 0.25375382228104604,
      "grad_norm": 3.265625,
      "learning_rate": 1.5386642200108218e-05,
      "loss": 1.2068,
      "step": 11120
    },
    {
      "epoch": 0.2539820181643923,
      "grad_norm": 3.28125,
      "learning_rate": 1.5381937092714143e-05,
      "loss": 1.2373,
      "step": 11130
    },
    {
      "epoch": 0.2542102140477386,
      "grad_norm": 3.0625,
      "learning_rate": 1.537723198532007e-05,
      "loss": 1.1943,
      "step": 11140
    },
    {
      "epoch": 0.25443840993108485,
      "grad_norm": 3.203125,
      "learning_rate": 1.537252687792599e-05,
      "loss": 1.2667,
      "step": 11150
    },
    {
      "epoch": 0.2546666058144311,
      "grad_norm": 3.046875,
      "learning_rate": 1.5367821770531916e-05,
      "loss": 1.2869,
      "step": 11160
    },
    {
      "epoch": 0.2548948016977774,
      "grad_norm": 2.78125,
      "learning_rate": 1.5363116663137838e-05,
      "loss": 1.1895,
      "step": 11170
    },
    {
      "epoch": 0.2551229975811236,
      "grad_norm": 2.875,
      "learning_rate": 1.5358411555743763e-05,
      "loss": 1.2264,
      "step": 11180
    },
    {
      "epoch": 0.2553511934644699,
      "grad_norm": 2.921875,
      "learning_rate": 1.5353706448349685e-05,
      "loss": 1.2246,
      "step": 11190
    },
    {
      "epoch": 0.25557938934781615,
      "grad_norm": 2.890625,
      "learning_rate": 1.534900134095561e-05,
      "loss": 1.1979,
      "step": 11200
    },
    {
      "epoch": 0.2558075852311624,
      "grad_norm": 2.90625,
      "learning_rate": 1.5344296233561533e-05,
      "loss": 1.1761,
      "step": 11210
    },
    {
      "epoch": 0.2560357811145087,
      "grad_norm": 2.90625,
      "learning_rate": 1.5339591126167458e-05,
      "loss": 1.2083,
      "step": 11220
    },
    {
      "epoch": 0.25626397699785497,
      "grad_norm": 3.078125,
      "learning_rate": 1.533488601877338e-05,
      "loss": 1.1816,
      "step": 11230
    },
    {
      "epoch": 0.25649217288120124,
      "grad_norm": 3.125,
      "learning_rate": 1.5330180911379306e-05,
      "loss": 1.2263,
      "step": 11240
    },
    {
      "epoch": 0.2567203687645475,
      "grad_norm": 2.953125,
      "learning_rate": 1.5325475803985228e-05,
      "loss": 1.2242,
      "step": 11250
    },
    {
      "epoch": 0.25694856464789373,
      "grad_norm": 3.265625,
      "learning_rate": 1.5320770696591153e-05,
      "loss": 1.202,
      "step": 11260
    },
    {
      "epoch": 0.25717676053124,
      "grad_norm": 2.75,
      "learning_rate": 1.5316065589197075e-05,
      "loss": 1.2326,
      "step": 11270
    },
    {
      "epoch": 0.2574049564145863,
      "grad_norm": 3.21875,
      "learning_rate": 1.5311360481803e-05,
      "loss": 1.2124,
      "step": 11280
    },
    {
      "epoch": 0.25763315229793254,
      "grad_norm": 3.234375,
      "learning_rate": 1.5306655374408923e-05,
      "loss": 1.2686,
      "step": 11290
    },
    {
      "epoch": 0.2578613481812788,
      "grad_norm": 3.375,
      "learning_rate": 1.5301950267014848e-05,
      "loss": 1.197,
      "step": 11300
    },
    {
      "epoch": 0.2580895440646251,
      "grad_norm": 3.25,
      "learning_rate": 1.529724515962077e-05,
      "loss": 1.2404,
      "step": 11310
    },
    {
      "epoch": 0.25831773994797136,
      "grad_norm": 2.96875,
      "learning_rate": 1.5292540052226695e-05,
      "loss": 1.2452,
      "step": 11320
    },
    {
      "epoch": 0.25854593583131763,
      "grad_norm": 3.203125,
      "learning_rate": 1.5287834944832617e-05,
      "loss": 1.213,
      "step": 11330
    },
    {
      "epoch": 0.25877413171466385,
      "grad_norm": 3.078125,
      "learning_rate": 1.5283129837438543e-05,
      "loss": 1.2022,
      "step": 11340
    },
    {
      "epoch": 0.2590023275980101,
      "grad_norm": 3.1875,
      "learning_rate": 1.5278424730044465e-05,
      "loss": 1.2319,
      "step": 11350
    },
    {
      "epoch": 0.2592305234813564,
      "grad_norm": 3.171875,
      "learning_rate": 1.527371962265039e-05,
      "loss": 1.2391,
      "step": 11360
    },
    {
      "epoch": 0.25945871936470266,
      "grad_norm": 3.0625,
      "learning_rate": 1.5269014515256312e-05,
      "loss": 1.2063,
      "step": 11370
    },
    {
      "epoch": 0.25968691524804893,
      "grad_norm": 3.109375,
      "learning_rate": 1.5264309407862238e-05,
      "loss": 1.1833,
      "step": 11380
    },
    {
      "epoch": 0.2599151111313952,
      "grad_norm": 2.875,
      "learning_rate": 1.525960430046816e-05,
      "loss": 1.2197,
      "step": 11390
    },
    {
      "epoch": 0.2601433070147415,
      "grad_norm": 3.140625,
      "learning_rate": 1.5254899193074083e-05,
      "loss": 1.253,
      "step": 11400
    },
    {
      "epoch": 0.2603715028980877,
      "grad_norm": 2.8125,
      "learning_rate": 1.5250194085680007e-05,
      "loss": 1.2512,
      "step": 11410
    },
    {
      "epoch": 0.26059969878143396,
      "grad_norm": 3.34375,
      "learning_rate": 1.5245488978285933e-05,
      "loss": 1.2069,
      "step": 11420
    },
    {
      "epoch": 0.26082789466478024,
      "grad_norm": 3.3125,
      "learning_rate": 1.5240783870891856e-05,
      "loss": 1.2273,
      "step": 11430
    },
    {
      "epoch": 0.2610560905481265,
      "grad_norm": 3.078125,
      "learning_rate": 1.523607876349778e-05,
      "loss": 1.2534,
      "step": 11440
    },
    {
      "epoch": 0.2612842864314728,
      "grad_norm": 3.1875,
      "learning_rate": 1.5231373656103704e-05,
      "loss": 1.219,
      "step": 11450
    },
    {
      "epoch": 0.26151248231481905,
      "grad_norm": 3.0625,
      "learning_rate": 1.5226668548709627e-05,
      "loss": 1.2044,
      "step": 11460
    },
    {
      "epoch": 0.2617406781981653,
      "grad_norm": 2.8125,
      "learning_rate": 1.5221963441315551e-05,
      "loss": 1.2161,
      "step": 11470
    },
    {
      "epoch": 0.2619688740815116,
      "grad_norm": 3.296875,
      "learning_rate": 1.5217258333921471e-05,
      "loss": 1.1763,
      "step": 11480
    },
    {
      "epoch": 0.2621970699648578,
      "grad_norm": 3.09375,
      "learning_rate": 1.5212553226527395e-05,
      "loss": 1.2035,
      "step": 11490
    },
    {
      "epoch": 0.2624252658482041,
      "grad_norm": 3.046875,
      "learning_rate": 1.5207848119133319e-05,
      "loss": 1.2428,
      "step": 11500
    },
    {
      "epoch": 0.26265346173155035,
      "grad_norm": 3.21875,
      "learning_rate": 1.5203143011739243e-05,
      "loss": 1.2238,
      "step": 11510
    },
    {
      "epoch": 0.2628816576148966,
      "grad_norm": 3.03125,
      "learning_rate": 1.5198437904345166e-05,
      "loss": 1.1791,
      "step": 11520
    },
    {
      "epoch": 0.2631098534982429,
      "grad_norm": 2.953125,
      "learning_rate": 1.519373279695109e-05,
      "loss": 1.2491,
      "step": 11530
    },
    {
      "epoch": 0.26333804938158917,
      "grad_norm": 3.15625,
      "learning_rate": 1.5189027689557014e-05,
      "loss": 1.1972,
      "step": 11540
    },
    {
      "epoch": 0.26356624526493544,
      "grad_norm": 2.96875,
      "learning_rate": 1.5184322582162937e-05,
      "loss": 1.2072,
      "step": 11550
    },
    {
      "epoch": 0.2637944411482817,
      "grad_norm": 2.84375,
      "learning_rate": 1.5179617474768861e-05,
      "loss": 1.2304,
      "step": 11560
    },
    {
      "epoch": 0.2640226370316279,
      "grad_norm": 3.015625,
      "learning_rate": 1.5174912367374785e-05,
      "loss": 1.1815,
      "step": 11570
    },
    {
      "epoch": 0.2642508329149742,
      "grad_norm": 2.71875,
      "learning_rate": 1.5170207259980709e-05,
      "loss": 1.2328,
      "step": 11580
    },
    {
      "epoch": 0.26447902879832047,
      "grad_norm": 3.09375,
      "learning_rate": 1.5165502152586632e-05,
      "loss": 1.151,
      "step": 11590
    },
    {
      "epoch": 0.26470722468166674,
      "grad_norm": 2.9375,
      "learning_rate": 1.5160797045192558e-05,
      "loss": 1.2247,
      "step": 11600
    },
    {
      "epoch": 0.264935420565013,
      "grad_norm": 3.078125,
      "learning_rate": 1.5156091937798481e-05,
      "loss": 1.201,
      "step": 11610
    },
    {
      "epoch": 0.2651636164483593,
      "grad_norm": 2.96875,
      "learning_rate": 1.5151386830404405e-05,
      "loss": 1.1911,
      "step": 11620
    },
    {
      "epoch": 0.26539181233170556,
      "grad_norm": 3.046875,
      "learning_rate": 1.5146681723010329e-05,
      "loss": 1.245,
      "step": 11630
    },
    {
      "epoch": 0.2656200082150518,
      "grad_norm": 2.828125,
      "learning_rate": 1.5141976615616252e-05,
      "loss": 1.1856,
      "step": 11640
    },
    {
      "epoch": 0.26584820409839804,
      "grad_norm": 2.796875,
      "learning_rate": 1.5137271508222176e-05,
      "loss": 1.1885,
      "step": 11650
    },
    {
      "epoch": 0.2660763999817443,
      "grad_norm": 3.265625,
      "learning_rate": 1.51325664008281e-05,
      "loss": 1.228,
      "step": 11660
    },
    {
      "epoch": 0.2663045958650906,
      "grad_norm": 2.921875,
      "learning_rate": 1.5127861293434024e-05,
      "loss": 1.1744,
      "step": 11670
    },
    {
      "epoch": 0.26653279174843686,
      "grad_norm": 3.03125,
      "learning_rate": 1.5123156186039947e-05,
      "loss": 1.2298,
      "step": 11680
    },
    {
      "epoch": 0.26676098763178313,
      "grad_norm": 3.203125,
      "learning_rate": 1.5118451078645871e-05,
      "loss": 1.2289,
      "step": 11690
    },
    {
      "epoch": 0.2669891835151294,
      "grad_norm": 3.265625,
      "learning_rate": 1.5113745971251795e-05,
      "loss": 1.2239,
      "step": 11700
    },
    {
      "epoch": 0.2672173793984757,
      "grad_norm": 3.28125,
      "learning_rate": 1.5109040863857718e-05,
      "loss": 1.2313,
      "step": 11710
    },
    {
      "epoch": 0.2674455752818219,
      "grad_norm": 3.046875,
      "learning_rate": 1.5104335756463642e-05,
      "loss": 1.2247,
      "step": 11720
    },
    {
      "epoch": 0.26767377116516816,
      "grad_norm": 3.265625,
      "learning_rate": 1.5099630649069566e-05,
      "loss": 1.1539,
      "step": 11730
    },
    {
      "epoch": 0.26790196704851443,
      "grad_norm": 3.015625,
      "learning_rate": 1.509492554167549e-05,
      "loss": 1.2104,
      "step": 11740
    },
    {
      "epoch": 0.2681301629318607,
      "grad_norm": 2.96875,
      "learning_rate": 1.5090220434281413e-05,
      "loss": 1.2803,
      "step": 11750
    },
    {
      "epoch": 0.268358358815207,
      "grad_norm": 2.8125,
      "learning_rate": 1.5085515326887337e-05,
      "loss": 1.1974,
      "step": 11760
    },
    {
      "epoch": 0.26858655469855325,
      "grad_norm": 2.90625,
      "learning_rate": 1.508081021949326e-05,
      "loss": 1.2368,
      "step": 11770
    },
    {
      "epoch": 0.2688147505818995,
      "grad_norm": 3.015625,
      "learning_rate": 1.5076105112099184e-05,
      "loss": 1.2267,
      "step": 11780
    },
    {
      "epoch": 0.2690429464652458,
      "grad_norm": 3.28125,
      "learning_rate": 1.5071400004705108e-05,
      "loss": 1.2216,
      "step": 11790
    },
    {
      "epoch": 0.269271142348592,
      "grad_norm": 3.3125,
      "learning_rate": 1.5066694897311032e-05,
      "loss": 1.2014,
      "step": 11800
    },
    {
      "epoch": 0.2694993382319383,
      "grad_norm": 2.96875,
      "learning_rate": 1.5061989789916956e-05,
      "loss": 1.2407,
      "step": 11810
    },
    {
      "epoch": 0.26972753411528455,
      "grad_norm": 2.859375,
      "learning_rate": 1.505728468252288e-05,
      "loss": 1.1953,
      "step": 11820
    },
    {
      "epoch": 0.2699557299986308,
      "grad_norm": 3.1875,
      "learning_rate": 1.5052579575128803e-05,
      "loss": 1.2256,
      "step": 11830
    },
    {
      "epoch": 0.2701839258819771,
      "grad_norm": 2.921875,
      "learning_rate": 1.5047874467734727e-05,
      "loss": 1.2495,
      "step": 11840
    },
    {
      "epoch": 0.27041212176532337,
      "grad_norm": 3.046875,
      "learning_rate": 1.504316936034065e-05,
      "loss": 1.2216,
      "step": 11850
    },
    {
      "epoch": 0.27064031764866964,
      "grad_norm": 3.0,
      "learning_rate": 1.5038464252946574e-05,
      "loss": 1.188,
      "step": 11860
    },
    {
      "epoch": 0.2708685135320159,
      "grad_norm": 3.203125,
      "learning_rate": 1.5033759145552498e-05,
      "loss": 1.2099,
      "step": 11870
    },
    {
      "epoch": 0.2710967094153621,
      "grad_norm": 2.734375,
      "learning_rate": 1.5029054038158422e-05,
      "loss": 1.2611,
      "step": 11880
    },
    {
      "epoch": 0.2713249052987084,
      "grad_norm": 2.875,
      "learning_rate": 1.5024348930764345e-05,
      "loss": 1.1934,
      "step": 11890
    },
    {
      "epoch": 0.27155310118205467,
      "grad_norm": 2.8125,
      "learning_rate": 1.5019643823370269e-05,
      "loss": 1.164,
      "step": 11900
    },
    {
      "epoch": 0.27178129706540094,
      "grad_norm": 2.90625,
      "learning_rate": 1.5014938715976193e-05,
      "loss": 1.2064,
      "step": 11910
    },
    {
      "epoch": 0.2720094929487472,
      "grad_norm": 3.078125,
      "learning_rate": 1.5010233608582116e-05,
      "loss": 1.2058,
      "step": 11920
    },
    {
      "epoch": 0.2722376888320935,
      "grad_norm": 2.625,
      "learning_rate": 1.500552850118804e-05,
      "loss": 1.2112,
      "step": 11930
    },
    {
      "epoch": 0.27246588471543975,
      "grad_norm": 3.265625,
      "learning_rate": 1.5000823393793964e-05,
      "loss": 1.2141,
      "step": 11940
    },
    {
      "epoch": 0.27269408059878597,
      "grad_norm": 3.125,
      "learning_rate": 1.4996118286399888e-05,
      "loss": 1.1945,
      "step": 11950
    },
    {
      "epoch": 0.27292227648213224,
      "grad_norm": 3.0,
      "learning_rate": 1.4991413179005811e-05,
      "loss": 1.1971,
      "step": 11960
    },
    {
      "epoch": 0.2731504723654785,
      "grad_norm": 2.859375,
      "learning_rate": 1.4986708071611735e-05,
      "loss": 1.2377,
      "step": 11970
    },
    {
      "epoch": 0.2733786682488248,
      "grad_norm": 2.9375,
      "learning_rate": 1.4982002964217659e-05,
      "loss": 1.2027,
      "step": 11980
    },
    {
      "epoch": 0.27360686413217106,
      "grad_norm": 3.015625,
      "learning_rate": 1.4977297856823582e-05,
      "loss": 1.1486,
      "step": 11990
    },
    {
      "epoch": 0.27383506001551733,
      "grad_norm": 3.109375,
      "learning_rate": 1.4972592749429506e-05,
      "loss": 1.2413,
      "step": 12000
    },
    {
      "epoch": 0.2740632558988636,
      "grad_norm": 3.34375,
      "learning_rate": 1.496788764203543e-05,
      "loss": 1.2195,
      "step": 12010
    },
    {
      "epoch": 0.27429145178220987,
      "grad_norm": 3.25,
      "learning_rate": 1.4963182534641354e-05,
      "loss": 1.2176,
      "step": 12020
    },
    {
      "epoch": 0.2745196476655561,
      "grad_norm": 3.234375,
      "learning_rate": 1.4958477427247277e-05,
      "loss": 1.1863,
      "step": 12030
    },
    {
      "epoch": 0.27474784354890236,
      "grad_norm": 3.0625,
      "learning_rate": 1.4953772319853201e-05,
      "loss": 1.2275,
      "step": 12040
    },
    {
      "epoch": 0.27497603943224863,
      "grad_norm": 3.546875,
      "learning_rate": 1.4949067212459125e-05,
      "loss": 1.237,
      "step": 12050
    },
    {
      "epoch": 0.2752042353155949,
      "grad_norm": 2.90625,
      "learning_rate": 1.4944362105065048e-05,
      "loss": 1.2495,
      "step": 12060
    },
    {
      "epoch": 0.2754324311989412,
      "grad_norm": 2.96875,
      "learning_rate": 1.4939656997670972e-05,
      "loss": 1.2489,
      "step": 12070
    },
    {
      "epoch": 0.27566062708228745,
      "grad_norm": 2.984375,
      "learning_rate": 1.4934951890276896e-05,
      "loss": 1.2296,
      "step": 12080
    },
    {
      "epoch": 0.2758888229656337,
      "grad_norm": 3.203125,
      "learning_rate": 1.493024678288282e-05,
      "loss": 1.1958,
      "step": 12090
    },
    {
      "epoch": 0.27611701884898,
      "grad_norm": 3.25,
      "learning_rate": 1.4925541675488743e-05,
      "loss": 1.22,
      "step": 12100
    },
    {
      "epoch": 0.2763452147323262,
      "grad_norm": 2.84375,
      "learning_rate": 1.4920836568094667e-05,
      "loss": 1.2326,
      "step": 12110
    },
    {
      "epoch": 0.2765734106156725,
      "grad_norm": 3.265625,
      "learning_rate": 1.491613146070059e-05,
      "loss": 1.2199,
      "step": 12120
    },
    {
      "epoch": 0.27680160649901875,
      "grad_norm": 2.765625,
      "learning_rate": 1.4911426353306514e-05,
      "loss": 1.1627,
      "step": 12130
    },
    {
      "epoch": 0.277029802382365,
      "grad_norm": 3.09375,
      "learning_rate": 1.4906721245912438e-05,
      "loss": 1.1984,
      "step": 12140
    },
    {
      "epoch": 0.2772579982657113,
      "grad_norm": 2.984375,
      "learning_rate": 1.4902016138518362e-05,
      "loss": 1.2493,
      "step": 12150
    },
    {
      "epoch": 0.27748619414905756,
      "grad_norm": 3.046875,
      "learning_rate": 1.4897311031124286e-05,
      "loss": 1.2463,
      "step": 12160
    },
    {
      "epoch": 0.27771439003240384,
      "grad_norm": 3.265625,
      "learning_rate": 1.489260592373021e-05,
      "loss": 1.2337,
      "step": 12170
    },
    {
      "epoch": 0.2779425859157501,
      "grad_norm": 3.328125,
      "learning_rate": 1.4887900816336135e-05,
      "loss": 1.266,
      "step": 12180
    },
    {
      "epoch": 0.2781707817990963,
      "grad_norm": 3.296875,
      "learning_rate": 1.4883195708942058e-05,
      "loss": 1.2526,
      "step": 12190
    },
    {
      "epoch": 0.2783989776824426,
      "grad_norm": 2.90625,
      "learning_rate": 1.4878490601547982e-05,
      "loss": 1.1954,
      "step": 12200
    },
    {
      "epoch": 0.27862717356578887,
      "grad_norm": 3.234375,
      "learning_rate": 1.4873785494153906e-05,
      "loss": 1.1969,
      "step": 12210
    },
    {
      "epoch": 0.27885536944913514,
      "grad_norm": 2.78125,
      "learning_rate": 1.486908038675983e-05,
      "loss": 1.2085,
      "step": 12220
    },
    {
      "epoch": 0.2790835653324814,
      "grad_norm": 3.59375,
      "learning_rate": 1.4864375279365753e-05,
      "loss": 1.1777,
      "step": 12230
    },
    {
      "epoch": 0.2793117612158277,
      "grad_norm": 2.90625,
      "learning_rate": 1.4859670171971677e-05,
      "loss": 1.2416,
      "step": 12240
    },
    {
      "epoch": 0.27953995709917395,
      "grad_norm": 2.6875,
      "learning_rate": 1.48549650645776e-05,
      "loss": 1.184,
      "step": 12250
    },
    {
      "epoch": 0.27976815298252017,
      "grad_norm": 2.953125,
      "learning_rate": 1.4850259957183524e-05,
      "loss": 1.2507,
      "step": 12260
    },
    {
      "epoch": 0.27999634886586644,
      "grad_norm": 3.234375,
      "learning_rate": 1.4845554849789448e-05,
      "loss": 1.2061,
      "step": 12270
    },
    {
      "epoch": 0.2802245447492127,
      "grad_norm": 3.0625,
      "learning_rate": 1.4840849742395372e-05,
      "loss": 1.2106,
      "step": 12280
    },
    {
      "epoch": 0.280452740632559,
      "grad_norm": 3.328125,
      "learning_rate": 1.4836144635001295e-05,
      "loss": 1.1919,
      "step": 12290
    },
    {
      "epoch": 0.28068093651590525,
      "grad_norm": 3.03125,
      "learning_rate": 1.483143952760722e-05,
      "loss": 1.1927,
      "step": 12300
    },
    {
      "epoch": 0.2809091323992515,
      "grad_norm": 2.90625,
      "learning_rate": 1.4826734420213143e-05,
      "loss": 1.1973,
      "step": 12310
    },
    {
      "epoch": 0.2811373282825978,
      "grad_norm": 3.15625,
      "learning_rate": 1.4822029312819067e-05,
      "loss": 1.1916,
      "step": 12320
    },
    {
      "epoch": 0.28136552416594407,
      "grad_norm": 3.171875,
      "learning_rate": 1.481732420542499e-05,
      "loss": 1.2726,
      "step": 12330
    },
    {
      "epoch": 0.2815937200492903,
      "grad_norm": 2.921875,
      "learning_rate": 1.4812619098030914e-05,
      "loss": 1.2292,
      "step": 12340
    },
    {
      "epoch": 0.28182191593263656,
      "grad_norm": 2.75,
      "learning_rate": 1.4807913990636838e-05,
      "loss": 1.1863,
      "step": 12350
    },
    {
      "epoch": 0.28205011181598283,
      "grad_norm": 3.421875,
      "learning_rate": 1.4803208883242761e-05,
      "loss": 1.2465,
      "step": 12360
    },
    {
      "epoch": 0.2822783076993291,
      "grad_norm": 3.0,
      "learning_rate": 1.4798503775848685e-05,
      "loss": 1.2196,
      "step": 12370
    },
    {
      "epoch": 0.2825065035826754,
      "grad_norm": 2.8125,
      "learning_rate": 1.4793798668454609e-05,
      "loss": 1.2329,
      "step": 12380
    },
    {
      "epoch": 0.28273469946602164,
      "grad_norm": 2.890625,
      "learning_rate": 1.4789093561060533e-05,
      "loss": 1.1639,
      "step": 12390
    },
    {
      "epoch": 0.2829628953493679,
      "grad_norm": 3.28125,
      "learning_rate": 1.4784388453666456e-05,
      "loss": 1.1765,
      "step": 12400
    },
    {
      "epoch": 0.2831910912327142,
      "grad_norm": 2.875,
      "learning_rate": 1.477968334627238e-05,
      "loss": 1.1825,
      "step": 12410
    },
    {
      "epoch": 0.2834192871160604,
      "grad_norm": 3.015625,
      "learning_rate": 1.4774978238878304e-05,
      "loss": 1.1715,
      "step": 12420
    },
    {
      "epoch": 0.2836474829994067,
      "grad_norm": 2.796875,
      "learning_rate": 1.4770273131484227e-05,
      "loss": 1.1939,
      "step": 12430
    },
    {
      "epoch": 0.28387567888275295,
      "grad_norm": 2.96875,
      "learning_rate": 1.4765568024090151e-05,
      "loss": 1.2408,
      "step": 12440
    },
    {
      "epoch": 0.2841038747660992,
      "grad_norm": 3.234375,
      "learning_rate": 1.4760862916696075e-05,
      "loss": 1.2935,
      "step": 12450
    },
    {
      "epoch": 0.2843320706494455,
      "grad_norm": 3.09375,
      "learning_rate": 1.4756157809301999e-05,
      "loss": 1.2226,
      "step": 12460
    },
    {
      "epoch": 0.28456026653279176,
      "grad_norm": 2.84375,
      "learning_rate": 1.4751452701907922e-05,
      "loss": 1.1737,
      "step": 12470
    },
    {
      "epoch": 0.28478846241613803,
      "grad_norm": 3.15625,
      "learning_rate": 1.4746747594513846e-05,
      "loss": 1.2135,
      "step": 12480
    },
    {
      "epoch": 0.2850166582994843,
      "grad_norm": 3.328125,
      "learning_rate": 1.474204248711977e-05,
      "loss": 1.2333,
      "step": 12490
    },
    {
      "epoch": 0.2852448541828305,
      "grad_norm": 2.796875,
      "learning_rate": 1.4737337379725693e-05,
      "loss": 1.2297,
      "step": 12500
    },
    {
      "epoch": 0.2854730500661768,
      "grad_norm": 3.390625,
      "learning_rate": 1.4732632272331617e-05,
      "loss": 1.2076,
      "step": 12510
    },
    {
      "epoch": 0.28570124594952306,
      "grad_norm": 2.84375,
      "learning_rate": 1.4727927164937541e-05,
      "loss": 1.2363,
      "step": 12520
    },
    {
      "epoch": 0.28592944183286934,
      "grad_norm": 2.953125,
      "learning_rate": 1.4723222057543465e-05,
      "loss": 1.1981,
      "step": 12530
    },
    {
      "epoch": 0.2861576377162156,
      "grad_norm": 3.015625,
      "learning_rate": 1.4718516950149388e-05,
      "loss": 1.2356,
      "step": 12540
    },
    {
      "epoch": 0.2863858335995619,
      "grad_norm": 3.265625,
      "learning_rate": 1.4713811842755312e-05,
      "loss": 1.2562,
      "step": 12550
    },
    {
      "epoch": 0.28661402948290815,
      "grad_norm": 2.90625,
      "learning_rate": 1.4709106735361236e-05,
      "loss": 1.1886,
      "step": 12560
    },
    {
      "epoch": 0.28684222536625437,
      "grad_norm": 3.171875,
      "learning_rate": 1.470440162796716e-05,
      "loss": 1.2359,
      "step": 12570
    },
    {
      "epoch": 0.28707042124960064,
      "grad_norm": 3.4375,
      "learning_rate": 1.4699696520573083e-05,
      "loss": 1.1782,
      "step": 12580
    },
    {
      "epoch": 0.2872986171329469,
      "grad_norm": 2.96875,
      "learning_rate": 1.4694991413179007e-05,
      "loss": 1.1736,
      "step": 12590
    },
    {
      "epoch": 0.2875268130162932,
      "grad_norm": 2.734375,
      "learning_rate": 1.469028630578493e-05,
      "loss": 1.2025,
      "step": 12600
    },
    {
      "epoch": 0.28775500889963945,
      "grad_norm": 3.109375,
      "learning_rate": 1.4685581198390854e-05,
      "loss": 1.1978,
      "step": 12610
    },
    {
      "epoch": 0.2879832047829857,
      "grad_norm": 3.046875,
      "learning_rate": 1.4680876090996778e-05,
      "loss": 1.1907,
      "step": 12620
    },
    {
      "epoch": 0.288211400666332,
      "grad_norm": 3.0,
      "learning_rate": 1.4676170983602702e-05,
      "loss": 1.2286,
      "step": 12630
    },
    {
      "epoch": 0.28843959654967827,
      "grad_norm": 3.15625,
      "learning_rate": 1.4671465876208625e-05,
      "loss": 1.2031,
      "step": 12640
    },
    {
      "epoch": 0.2886677924330245,
      "grad_norm": 3.046875,
      "learning_rate": 1.4666760768814549e-05,
      "loss": 1.2198,
      "step": 12650
    },
    {
      "epoch": 0.28889598831637076,
      "grad_norm": 2.890625,
      "learning_rate": 1.4662055661420473e-05,
      "loss": 1.1942,
      "step": 12660
    },
    {
      "epoch": 0.289124184199717,
      "grad_norm": 3.140625,
      "learning_rate": 1.4657350554026397e-05,
      "loss": 1.2404,
      "step": 12670
    },
    {
      "epoch": 0.2893523800830633,
      "grad_norm": 3.1875,
      "learning_rate": 1.465264544663232e-05,
      "loss": 1.2416,
      "step": 12680
    },
    {
      "epoch": 0.28958057596640957,
      "grad_norm": 3.109375,
      "learning_rate": 1.4647940339238244e-05,
      "loss": 1.2627,
      "step": 12690
    },
    {
      "epoch": 0.28980877184975584,
      "grad_norm": 2.5625,
      "learning_rate": 1.4643235231844168e-05,
      "loss": 1.1792,
      "step": 12700
    },
    {
      "epoch": 0.2900369677331021,
      "grad_norm": 2.796875,
      "learning_rate": 1.4638530124450091e-05,
      "loss": 1.1731,
      "step": 12710
    },
    {
      "epoch": 0.2902651636164484,
      "grad_norm": 3.1875,
      "learning_rate": 1.4633825017056015e-05,
      "loss": 1.2295,
      "step": 12720
    },
    {
      "epoch": 0.2904933594997946,
      "grad_norm": 3.03125,
      "learning_rate": 1.4629119909661939e-05,
      "loss": 1.259,
      "step": 12730
    },
    {
      "epoch": 0.2907215553831409,
      "grad_norm": 2.8125,
      "learning_rate": 1.4624414802267863e-05,
      "loss": 1.1758,
      "step": 12740
    },
    {
      "epoch": 0.29094975126648714,
      "grad_norm": 3.0625,
      "learning_rate": 1.4619709694873788e-05,
      "loss": 1.1624,
      "step": 12750
    },
    {
      "epoch": 0.2911779471498334,
      "grad_norm": 2.984375,
      "learning_rate": 1.4615004587479712e-05,
      "loss": 1.2543,
      "step": 12760
    },
    {
      "epoch": 0.2914061430331797,
      "grad_norm": 2.875,
      "learning_rate": 1.4610299480085635e-05,
      "loss": 1.2456,
      "step": 12770
    },
    {
      "epoch": 0.29163433891652596,
      "grad_norm": 2.96875,
      "learning_rate": 1.4605594372691559e-05,
      "loss": 1.166,
      "step": 12780
    },
    {
      "epoch": 0.29186253479987223,
      "grad_norm": 3.078125,
      "learning_rate": 1.4600889265297483e-05,
      "loss": 1.2313,
      "step": 12790
    },
    {
      "epoch": 0.2920907306832185,
      "grad_norm": 3.28125,
      "learning_rate": 1.4596184157903407e-05,
      "loss": 1.1835,
      "step": 12800
    },
    {
      "epoch": 0.2923189265665647,
      "grad_norm": 3.15625,
      "learning_rate": 1.459147905050933e-05,
      "loss": 1.2022,
      "step": 12810
    },
    {
      "epoch": 0.292547122449911,
      "grad_norm": 3.0625,
      "learning_rate": 1.4586773943115254e-05,
      "loss": 1.2939,
      "step": 12820
    },
    {
      "epoch": 0.29277531833325726,
      "grad_norm": 2.625,
      "learning_rate": 1.4582068835721178e-05,
      "loss": 1.1899,
      "step": 12830
    },
    {
      "epoch": 0.29300351421660353,
      "grad_norm": 3.15625,
      "learning_rate": 1.4577363728327101e-05,
      "loss": 1.206,
      "step": 12840
    },
    {
      "epoch": 0.2932317100999498,
      "grad_norm": 3.078125,
      "learning_rate": 1.4572658620933025e-05,
      "loss": 1.1752,
      "step": 12850
    },
    {
      "epoch": 0.2934599059832961,
      "grad_norm": 3.09375,
      "learning_rate": 1.4567953513538949e-05,
      "loss": 1.2518,
      "step": 12860
    },
    {
      "epoch": 0.29368810186664235,
      "grad_norm": 3.125,
      "learning_rate": 1.4563248406144872e-05,
      "loss": 1.1751,
      "step": 12870
    },
    {
      "epoch": 0.29391629774998856,
      "grad_norm": 2.96875,
      "learning_rate": 1.4558543298750796e-05,
      "loss": 1.1864,
      "step": 12880
    },
    {
      "epoch": 0.29414449363333484,
      "grad_norm": 3.234375,
      "learning_rate": 1.455383819135672e-05,
      "loss": 1.2198,
      "step": 12890
    },
    {
      "epoch": 0.2943726895166811,
      "grad_norm": 3.140625,
      "learning_rate": 1.4549133083962644e-05,
      "loss": 1.1407,
      "step": 12900
    },
    {
      "epoch": 0.2946008854000274,
      "grad_norm": 3.484375,
      "learning_rate": 1.4544427976568567e-05,
      "loss": 1.2192,
      "step": 12910
    },
    {
      "epoch": 0.29482908128337365,
      "grad_norm": 3.421875,
      "learning_rate": 1.4539722869174491e-05,
      "loss": 1.2513,
      "step": 12920
    },
    {
      "epoch": 0.2950572771667199,
      "grad_norm": 2.71875,
      "learning_rate": 1.4535017761780415e-05,
      "loss": 1.2101,
      "step": 12930
    },
    {
      "epoch": 0.2952854730500662,
      "grad_norm": 3.09375,
      "learning_rate": 1.4530312654386338e-05,
      "loss": 1.2363,
      "step": 12940
    },
    {
      "epoch": 0.29551366893341247,
      "grad_norm": 3.125,
      "learning_rate": 1.4525607546992262e-05,
      "loss": 1.2475,
      "step": 12950
    },
    {
      "epoch": 0.2957418648167587,
      "grad_norm": 3.125,
      "learning_rate": 1.4520902439598186e-05,
      "loss": 1.2431,
      "step": 12960
    },
    {
      "epoch": 0.29597006070010495,
      "grad_norm": 2.671875,
      "learning_rate": 1.451619733220411e-05,
      "loss": 1.1809,
      "step": 12970
    },
    {
      "epoch": 0.2961982565834512,
      "grad_norm": 3.125,
      "learning_rate": 1.4511492224810033e-05,
      "loss": 1.2512,
      "step": 12980
    },
    {
      "epoch": 0.2964264524667975,
      "grad_norm": 2.796875,
      "learning_rate": 1.4506787117415957e-05,
      "loss": 1.2085,
      "step": 12990
    },
    {
      "epoch": 0.29665464835014377,
      "grad_norm": 3.09375,
      "learning_rate": 1.450208201002188e-05,
      "loss": 1.1726,
      "step": 13000
    },
    {
      "epoch": 0.29688284423349004,
      "grad_norm": 3.15625,
      "learning_rate": 1.4497376902627804e-05,
      "loss": 1.2154,
      "step": 13010
    },
    {
      "epoch": 0.2971110401168363,
      "grad_norm": 3.09375,
      "learning_rate": 1.4492671795233728e-05,
      "loss": 1.2204,
      "step": 13020
    },
    {
      "epoch": 0.2973392360001826,
      "grad_norm": 3.015625,
      "learning_rate": 1.4487966687839652e-05,
      "loss": 1.1929,
      "step": 13030
    },
    {
      "epoch": 0.2975674318835288,
      "grad_norm": 2.75,
      "learning_rate": 1.4483261580445576e-05,
      "loss": 1.221,
      "step": 13040
    },
    {
      "epoch": 0.29779562776687507,
      "grad_norm": 3.15625,
      "learning_rate": 1.44785564730515e-05,
      "loss": 1.2341,
      "step": 13050
    },
    {
      "epoch": 0.29802382365022134,
      "grad_norm": 2.984375,
      "learning_rate": 1.4473851365657423e-05,
      "loss": 1.1667,
      "step": 13060
    },
    {
      "epoch": 0.2982520195335676,
      "grad_norm": 2.984375,
      "learning_rate": 1.4469146258263347e-05,
      "loss": 1.1911,
      "step": 13070
    },
    {
      "epoch": 0.2984802154169139,
      "grad_norm": 3.390625,
      "learning_rate": 1.446444115086927e-05,
      "loss": 1.1971,
      "step": 13080
    },
    {
      "epoch": 0.29870841130026016,
      "grad_norm": 3.3125,
      "learning_rate": 1.4459736043475194e-05,
      "loss": 1.1981,
      "step": 13090
    },
    {
      "epoch": 0.29893660718360643,
      "grad_norm": 3.078125,
      "learning_rate": 1.4455030936081118e-05,
      "loss": 1.2091,
      "step": 13100
    },
    {
      "epoch": 0.29916480306695264,
      "grad_norm": 2.84375,
      "learning_rate": 1.4450325828687042e-05,
      "loss": 1.2005,
      "step": 13110
    },
    {
      "epoch": 0.2993929989502989,
      "grad_norm": 3.046875,
      "learning_rate": 1.4445620721292965e-05,
      "loss": 1.1926,
      "step": 13120
    },
    {
      "epoch": 0.2996211948336452,
      "grad_norm": 3.171875,
      "learning_rate": 1.4440915613898889e-05,
      "loss": 1.2368,
      "step": 13130
    },
    {
      "epoch": 0.29984939071699146,
      "grad_norm": 3.09375,
      "learning_rate": 1.4436210506504813e-05,
      "loss": 1.1925,
      "step": 13140
    },
    {
      "epoch": 0.30007758660033773,
      "grad_norm": 3.296875,
      "learning_rate": 1.4431505399110736e-05,
      "loss": 1.2562,
      "step": 13150
    },
    {
      "epoch": 0.300305782483684,
      "grad_norm": 2.953125,
      "learning_rate": 1.442680029171666e-05,
      "loss": 1.2001,
      "step": 13160
    },
    {
      "epoch": 0.3005339783670303,
      "grad_norm": 2.984375,
      "learning_rate": 1.4422095184322584e-05,
      "loss": 1.1501,
      "step": 13170
    },
    {
      "epoch": 0.30076217425037655,
      "grad_norm": 3.0625,
      "learning_rate": 1.4417390076928508e-05,
      "loss": 1.181,
      "step": 13180
    },
    {
      "epoch": 0.30099037013372276,
      "grad_norm": 2.96875,
      "learning_rate": 1.4412684969534431e-05,
      "loss": 1.1905,
      "step": 13190
    },
    {
      "epoch": 0.30121856601706903,
      "grad_norm": 3.015625,
      "learning_rate": 1.4407979862140355e-05,
      "loss": 1.2101,
      "step": 13200
    },
    {
      "epoch": 0.3014467619004153,
      "grad_norm": 3.1875,
      "learning_rate": 1.4403274754746279e-05,
      "loss": 1.162,
      "step": 13210
    },
    {
      "epoch": 0.3016749577837616,
      "grad_norm": 2.953125,
      "learning_rate": 1.4398569647352202e-05,
      "loss": 1.1661,
      "step": 13220
    },
    {
      "epoch": 0.30190315366710785,
      "grad_norm": 2.703125,
      "learning_rate": 1.4393864539958126e-05,
      "loss": 1.1984,
      "step": 13230
    },
    {
      "epoch": 0.3021313495504541,
      "grad_norm": 2.78125,
      "learning_rate": 1.438915943256405e-05,
      "loss": 1.148,
      "step": 13240
    },
    {
      "epoch": 0.3023595454338004,
      "grad_norm": 2.734375,
      "learning_rate": 1.4384454325169974e-05,
      "loss": 1.2644,
      "step": 13250
    },
    {
      "epoch": 0.30258774131714666,
      "grad_norm": 3.015625,
      "learning_rate": 1.4379749217775897e-05,
      "loss": 1.2437,
      "step": 13260
    },
    {
      "epoch": 0.3028159372004929,
      "grad_norm": 3.453125,
      "learning_rate": 1.4375044110381821e-05,
      "loss": 1.2274,
      "step": 13270
    },
    {
      "epoch": 0.30304413308383915,
      "grad_norm": 2.84375,
      "learning_rate": 1.4370339002987745e-05,
      "loss": 1.1997,
      "step": 13280
    },
    {
      "epoch": 0.3032723289671854,
      "grad_norm": 2.953125,
      "learning_rate": 1.4365633895593668e-05,
      "loss": 1.1997,
      "step": 13290
    },
    {
      "epoch": 0.3035005248505317,
      "grad_norm": 3.078125,
      "learning_rate": 1.4360928788199592e-05,
      "loss": 1.1628,
      "step": 13300
    },
    {
      "epoch": 0.30372872073387797,
      "grad_norm": 3.15625,
      "learning_rate": 1.4356223680805516e-05,
      "loss": 1.2071,
      "step": 13310
    },
    {
      "epoch": 0.30395691661722424,
      "grad_norm": 3.359375,
      "learning_rate": 1.435151857341144e-05,
      "loss": 1.1921,
      "step": 13320
    },
    {
      "epoch": 0.3041851125005705,
      "grad_norm": 3.03125,
      "learning_rate": 1.4346813466017362e-05,
      "loss": 1.2044,
      "step": 13330
    },
    {
      "epoch": 0.3044133083839168,
      "grad_norm": 3.171875,
      "learning_rate": 1.4342108358623285e-05,
      "loss": 1.2484,
      "step": 13340
    },
    {
      "epoch": 0.304641504267263,
      "grad_norm": 3.25,
      "learning_rate": 1.4337403251229209e-05,
      "loss": 1.187,
      "step": 13350
    },
    {
      "epoch": 0.30486970015060927,
      "grad_norm": 2.953125,
      "learning_rate": 1.4332698143835133e-05,
      "loss": 1.1798,
      "step": 13360
    },
    {
      "epoch": 0.30509789603395554,
      "grad_norm": 3.171875,
      "learning_rate": 1.4327993036441056e-05,
      "loss": 1.2057,
      "step": 13370
    },
    {
      "epoch": 0.3053260919173018,
      "grad_norm": 3.046875,
      "learning_rate": 1.432328792904698e-05,
      "loss": 1.2144,
      "step": 13380
    },
    {
      "epoch": 0.3055542878006481,
      "grad_norm": 3.71875,
      "learning_rate": 1.4318582821652904e-05,
      "loss": 1.2433,
      "step": 13390
    },
    {
      "epoch": 0.30578248368399435,
      "grad_norm": 2.96875,
      "learning_rate": 1.4313877714258828e-05,
      "loss": 1.2052,
      "step": 13400
    },
    {
      "epoch": 0.3060106795673406,
      "grad_norm": 3.109375,
      "learning_rate": 1.4309172606864751e-05,
      "loss": 1.1657,
      "step": 13410
    },
    {
      "epoch": 0.30623887545068684,
      "grad_norm": 3.203125,
      "learning_rate": 1.4304467499470675e-05,
      "loss": 1.2071,
      "step": 13420
    },
    {
      "epoch": 0.3064670713340331,
      "grad_norm": 3.125,
      "learning_rate": 1.4299762392076599e-05,
      "loss": 1.1904,
      "step": 13430
    },
    {
      "epoch": 0.3066952672173794,
      "grad_norm": 2.953125,
      "learning_rate": 1.4295057284682522e-05,
      "loss": 1.2732,
      "step": 13440
    },
    {
      "epoch": 0.30692346310072566,
      "grad_norm": 3.03125,
      "learning_rate": 1.4290352177288446e-05,
      "loss": 1.1841,
      "step": 13450
    },
    {
      "epoch": 0.30715165898407193,
      "grad_norm": 2.765625,
      "learning_rate": 1.428564706989437e-05,
      "loss": 1.2394,
      "step": 13460
    },
    {
      "epoch": 0.3073798548674182,
      "grad_norm": 3.09375,
      "learning_rate": 1.4280941962500294e-05,
      "loss": 1.2529,
      "step": 13470
    },
    {
      "epoch": 0.3076080507507645,
      "grad_norm": 2.96875,
      "learning_rate": 1.4276236855106217e-05,
      "loss": 1.2395,
      "step": 13480
    },
    {
      "epoch": 0.30783624663411074,
      "grad_norm": 2.90625,
      "learning_rate": 1.4271531747712141e-05,
      "loss": 1.1967,
      "step": 13490
    },
    {
      "epoch": 0.30806444251745696,
      "grad_norm": 2.890625,
      "learning_rate": 1.4266826640318065e-05,
      "loss": 1.1972,
      "step": 13500
    },
    {
      "epoch": 0.30829263840080323,
      "grad_norm": 3.078125,
      "learning_rate": 1.426212153292399e-05,
      "loss": 1.258,
      "step": 13510
    },
    {
      "epoch": 0.3085208342841495,
      "grad_norm": 3.25,
      "learning_rate": 1.4257416425529914e-05,
      "loss": 1.1771,
      "step": 13520
    },
    {
      "epoch": 0.3087490301674958,
      "grad_norm": 3.25,
      "learning_rate": 1.4252711318135838e-05,
      "loss": 1.2233,
      "step": 13530
    },
    {
      "epoch": 0.30897722605084205,
      "grad_norm": 2.890625,
      "learning_rate": 1.4248006210741761e-05,
      "loss": 1.2073,
      "step": 13540
    },
    {
      "epoch": 0.3092054219341883,
      "grad_norm": 2.9375,
      "learning_rate": 1.4243301103347685e-05,
      "loss": 1.2742,
      "step": 13550
    },
    {
      "epoch": 0.3094336178175346,
      "grad_norm": 2.921875,
      "learning_rate": 1.4238595995953609e-05,
      "loss": 1.2068,
      "step": 13560
    },
    {
      "epoch": 0.30966181370088086,
      "grad_norm": 2.859375,
      "learning_rate": 1.4233890888559532e-05,
      "loss": 1.2147,
      "step": 13570
    },
    {
      "epoch": 0.3098900095842271,
      "grad_norm": 3.171875,
      "learning_rate": 1.4229185781165456e-05,
      "loss": 1.184,
      "step": 13580
    },
    {
      "epoch": 0.31011820546757335,
      "grad_norm": 2.765625,
      "learning_rate": 1.422448067377138e-05,
      "loss": 1.2224,
      "step": 13590
    },
    {
      "epoch": 0.3103464013509196,
      "grad_norm": 3.21875,
      "learning_rate": 1.4219775566377304e-05,
      "loss": 1.1943,
      "step": 13600
    },
    {
      "epoch": 0.3105745972342659,
      "grad_norm": 2.828125,
      "learning_rate": 1.4215070458983227e-05,
      "loss": 1.1859,
      "step": 13610
    },
    {
      "epoch": 0.31080279311761216,
      "grad_norm": 3.015625,
      "learning_rate": 1.4210365351589151e-05,
      "loss": 1.1881,
      "step": 13620
    },
    {
      "epoch": 0.31103098900095844,
      "grad_norm": 2.921875,
      "learning_rate": 1.4205660244195075e-05,
      "loss": 1.2548,
      "step": 13630
    },
    {
      "epoch": 0.3112591848843047,
      "grad_norm": 3.296875,
      "learning_rate": 1.4200955136800998e-05,
      "loss": 1.2202,
      "step": 13640
    },
    {
      "epoch": 0.311487380767651,
      "grad_norm": 2.609375,
      "learning_rate": 1.4196250029406922e-05,
      "loss": 1.1774,
      "step": 13650
    },
    {
      "epoch": 0.3117155766509972,
      "grad_norm": 3.078125,
      "learning_rate": 1.4191544922012846e-05,
      "loss": 1.213,
      "step": 13660
    },
    {
      "epoch": 0.31194377253434347,
      "grad_norm": 3.265625,
      "learning_rate": 1.418683981461877e-05,
      "loss": 1.2132,
      "step": 13670
    },
    {
      "epoch": 0.31217196841768974,
      "grad_norm": 2.84375,
      "learning_rate": 1.4182134707224693e-05,
      "loss": 1.2325,
      "step": 13680
    },
    {
      "epoch": 0.312400164301036,
      "grad_norm": 3.0,
      "learning_rate": 1.4177429599830617e-05,
      "loss": 1.2184,
      "step": 13690
    },
    {
      "epoch": 0.3126283601843823,
      "grad_norm": 3.1875,
      "learning_rate": 1.417272449243654e-05,
      "loss": 1.1969,
      "step": 13700
    },
    {
      "epoch": 0.31285655606772855,
      "grad_norm": 2.9375,
      "learning_rate": 1.4168019385042464e-05,
      "loss": 1.2356,
      "step": 13710
    },
    {
      "epoch": 0.3130847519510748,
      "grad_norm": 2.9375,
      "learning_rate": 1.4163314277648388e-05,
      "loss": 1.1178,
      "step": 13720
    },
    {
      "epoch": 0.31331294783442104,
      "grad_norm": 3.109375,
      "learning_rate": 1.4158609170254312e-05,
      "loss": 1.2411,
      "step": 13730
    },
    {
      "epoch": 0.3135411437177673,
      "grad_norm": 3.125,
      "learning_rate": 1.4153904062860235e-05,
      "loss": 1.22,
      "step": 13740
    },
    {
      "epoch": 0.3137693396011136,
      "grad_norm": 2.796875,
      "learning_rate": 1.414919895546616e-05,
      "loss": 1.1801,
      "step": 13750
    },
    {
      "epoch": 0.31399753548445986,
      "grad_norm": 2.84375,
      "learning_rate": 1.4144493848072083e-05,
      "loss": 1.1971,
      "step": 13760
    },
    {
      "epoch": 0.3142257313678061,
      "grad_norm": 3.28125,
      "learning_rate": 1.4139788740678007e-05,
      "loss": 1.231,
      "step": 13770
    },
    {
      "epoch": 0.3144539272511524,
      "grad_norm": 2.96875,
      "learning_rate": 1.413508363328393e-05,
      "loss": 1.2077,
      "step": 13780
    },
    {
      "epoch": 0.31468212313449867,
      "grad_norm": 3.140625,
      "learning_rate": 1.4130378525889854e-05,
      "loss": 1.1675,
      "step": 13790
    },
    {
      "epoch": 0.31491031901784494,
      "grad_norm": 3.109375,
      "learning_rate": 1.4125673418495778e-05,
      "loss": 1.2069,
      "step": 13800
    },
    {
      "epoch": 0.31513851490119116,
      "grad_norm": 2.859375,
      "learning_rate": 1.4120968311101701e-05,
      "loss": 1.1784,
      "step": 13810
    },
    {
      "epoch": 0.31536671078453743,
      "grad_norm": 2.96875,
      "learning_rate": 1.4116263203707625e-05,
      "loss": 1.2077,
      "step": 13820
    },
    {
      "epoch": 0.3155949066678837,
      "grad_norm": 3.0625,
      "learning_rate": 1.4111558096313549e-05,
      "loss": 1.2111,
      "step": 13830
    },
    {
      "epoch": 0.31582310255123,
      "grad_norm": 2.96875,
      "learning_rate": 1.4106852988919473e-05,
      "loss": 1.1521,
      "step": 13840
    },
    {
      "epoch": 0.31605129843457624,
      "grad_norm": 2.90625,
      "learning_rate": 1.4102147881525396e-05,
      "loss": 1.2157,
      "step": 13850
    },
    {
      "epoch": 0.3162794943179225,
      "grad_norm": 3.09375,
      "learning_rate": 1.409744277413132e-05,
      "loss": 1.2575,
      "step": 13860
    },
    {
      "epoch": 0.3165076902012688,
      "grad_norm": 3.015625,
      "learning_rate": 1.4092737666737244e-05,
      "loss": 1.2166,
      "step": 13870
    },
    {
      "epoch": 0.31673588608461506,
      "grad_norm": 3.140625,
      "learning_rate": 1.4088032559343167e-05,
      "loss": 1.1903,
      "step": 13880
    },
    {
      "epoch": 0.3169640819679613,
      "grad_norm": 3.25,
      "learning_rate": 1.4083327451949091e-05,
      "loss": 1.2131,
      "step": 13890
    },
    {
      "epoch": 0.31719227785130755,
      "grad_norm": 2.9375,
      "learning_rate": 1.4078622344555015e-05,
      "loss": 1.2064,
      "step": 13900
    },
    {
      "epoch": 0.3174204737346538,
      "grad_norm": 3.046875,
      "learning_rate": 1.4073917237160939e-05,
      "loss": 1.2113,
      "step": 13910
    },
    {
      "epoch": 0.3176486696180001,
      "grad_norm": 3.25,
      "learning_rate": 1.4069212129766862e-05,
      "loss": 1.1819,
      "step": 13920
    },
    {
      "epoch": 0.31787686550134636,
      "grad_norm": 3.0625,
      "learning_rate": 1.4064507022372786e-05,
      "loss": 1.215,
      "step": 13930
    },
    {
      "epoch": 0.31810506138469263,
      "grad_norm": 3.125,
      "learning_rate": 1.405980191497871e-05,
      "loss": 1.2316,
      "step": 13940
    },
    {
      "epoch": 0.3183332572680389,
      "grad_norm": 2.75,
      "learning_rate": 1.4055096807584633e-05,
      "loss": 1.1833,
      "step": 13950
    },
    {
      "epoch": 0.3185614531513852,
      "grad_norm": 2.90625,
      "learning_rate": 1.4050391700190557e-05,
      "loss": 1.1783,
      "step": 13960
    },
    {
      "epoch": 0.3187896490347314,
      "grad_norm": 2.984375,
      "learning_rate": 1.4045686592796481e-05,
      "loss": 1.2239,
      "step": 13970
    },
    {
      "epoch": 0.31901784491807766,
      "grad_norm": 2.84375,
      "learning_rate": 1.4040981485402405e-05,
      "loss": 1.1857,
      "step": 13980
    },
    {
      "epoch": 0.31924604080142394,
      "grad_norm": 3.046875,
      "learning_rate": 1.4036276378008328e-05,
      "loss": 1.1989,
      "step": 13990
    },
    {
      "epoch": 0.3194742366847702,
      "grad_norm": 2.90625,
      "learning_rate": 1.4031571270614252e-05,
      "loss": 1.242,
      "step": 14000
    },
    {
      "epoch": 0.3197024325681165,
      "grad_norm": 2.875,
      "learning_rate": 1.4026866163220176e-05,
      "loss": 1.1945,
      "step": 14010
    },
    {
      "epoch": 0.31993062845146275,
      "grad_norm": 3.515625,
      "learning_rate": 1.40221610558261e-05,
      "loss": 1.2098,
      "step": 14020
    },
    {
      "epoch": 0.320158824334809,
      "grad_norm": 2.859375,
      "learning_rate": 1.4017455948432023e-05,
      "loss": 1.1866,
      "step": 14030
    },
    {
      "epoch": 0.32038702021815524,
      "grad_norm": 3.140625,
      "learning_rate": 1.4012750841037947e-05,
      "loss": 1.2233,
      "step": 14040
    },
    {
      "epoch": 0.3206152161015015,
      "grad_norm": 3.046875,
      "learning_rate": 1.400804573364387e-05,
      "loss": 1.2092,
      "step": 14050
    },
    {
      "epoch": 0.3208434119848478,
      "grad_norm": 3.28125,
      "learning_rate": 1.4003340626249794e-05,
      "loss": 1.1528,
      "step": 14060
    },
    {
      "epoch": 0.32107160786819405,
      "grad_norm": 3.15625,
      "learning_rate": 1.3998635518855718e-05,
      "loss": 1.1615,
      "step": 14070
    },
    {
      "epoch": 0.3212998037515403,
      "grad_norm": 3.359375,
      "learning_rate": 1.3993930411461643e-05,
      "loss": 1.1604,
      "step": 14080
    },
    {
      "epoch": 0.3215279996348866,
      "grad_norm": 3.09375,
      "learning_rate": 1.3989225304067567e-05,
      "loss": 1.2487,
      "step": 14090
    },
    {
      "epoch": 0.32175619551823287,
      "grad_norm": 2.984375,
      "learning_rate": 1.398452019667349e-05,
      "loss": 1.1556,
      "step": 14100
    },
    {
      "epoch": 0.32198439140157914,
      "grad_norm": 2.921875,
      "learning_rate": 1.3979815089279415e-05,
      "loss": 1.1736,
      "step": 14110
    },
    {
      "epoch": 0.32221258728492536,
      "grad_norm": 2.90625,
      "learning_rate": 1.3975109981885338e-05,
      "loss": 1.2091,
      "step": 14120
    },
    {
      "epoch": 0.3224407831682716,
      "grad_norm": 3.328125,
      "learning_rate": 1.3970404874491262e-05,
      "loss": 1.2154,
      "step": 14130
    },
    {
      "epoch": 0.3226689790516179,
      "grad_norm": 2.953125,
      "learning_rate": 1.3965699767097186e-05,
      "loss": 1.2194,
      "step": 14140
    },
    {
      "epoch": 0.32289717493496417,
      "grad_norm": 3.078125,
      "learning_rate": 1.396099465970311e-05,
      "loss": 1.204,
      "step": 14150
    },
    {
      "epoch": 0.32312537081831044,
      "grad_norm": 3.046875,
      "learning_rate": 1.3956289552309033e-05,
      "loss": 1.2108,
      "step": 14160
    },
    {
      "epoch": 0.3233535667016567,
      "grad_norm": 3.09375,
      "learning_rate": 1.3951584444914957e-05,
      "loss": 1.2283,
      "step": 14170
    },
    {
      "epoch": 0.323581762585003,
      "grad_norm": 3.5,
      "learning_rate": 1.394687933752088e-05,
      "loss": 1.1934,
      "step": 14180
    },
    {
      "epoch": 0.32380995846834926,
      "grad_norm": 3.34375,
      "learning_rate": 1.3942174230126804e-05,
      "loss": 1.2406,
      "step": 14190
    },
    {
      "epoch": 0.3240381543516955,
      "grad_norm": 3.4375,
      "learning_rate": 1.3937469122732728e-05,
      "loss": 1.2024,
      "step": 14200
    },
    {
      "epoch": 0.32426635023504174,
      "grad_norm": 3.03125,
      "learning_rate": 1.3932764015338652e-05,
      "loss": 1.207,
      "step": 14210
    },
    {
      "epoch": 0.324494546118388,
      "grad_norm": 2.9375,
      "learning_rate": 1.3928058907944575e-05,
      "loss": 1.2454,
      "step": 14220
    },
    {
      "epoch": 0.3247227420017343,
      "grad_norm": 3.125,
      "learning_rate": 1.3923353800550499e-05,
      "loss": 1.1808,
      "step": 14230
    },
    {
      "epoch": 0.32495093788508056,
      "grad_norm": 3.03125,
      "learning_rate": 1.3918648693156423e-05,
      "loss": 1.1872,
      "step": 14240
    },
    {
      "epoch": 0.32517913376842683,
      "grad_norm": 3.1875,
      "learning_rate": 1.3913943585762347e-05,
      "loss": 1.2113,
      "step": 14250
    },
    {
      "epoch": 0.3254073296517731,
      "grad_norm": 3.21875,
      "learning_rate": 1.390923847836827e-05,
      "loss": 1.228,
      "step": 14260
    },
    {
      "epoch": 0.3256355255351193,
      "grad_norm": 3.125,
      "learning_rate": 1.3904533370974194e-05,
      "loss": 1.2244,
      "step": 14270
    },
    {
      "epoch": 0.3258637214184656,
      "grad_norm": 3.734375,
      "learning_rate": 1.3899828263580118e-05,
      "loss": 1.2005,
      "step": 14280
    },
    {
      "epoch": 0.32609191730181186,
      "grad_norm": 2.796875,
      "learning_rate": 1.3895123156186041e-05,
      "loss": 1.2011,
      "step": 14290
    },
    {
      "epoch": 0.32632011318515813,
      "grad_norm": 3.0625,
      "learning_rate": 1.3890418048791965e-05,
      "loss": 1.1483,
      "step": 14300
    },
    {
      "epoch": 0.3265483090685044,
      "grad_norm": 4.59375,
      "learning_rate": 1.3885712941397889e-05,
      "loss": 1.2069,
      "step": 14310
    },
    {
      "epoch": 0.3267765049518507,
      "grad_norm": 2.953125,
      "learning_rate": 1.3881007834003812e-05,
      "loss": 1.2382,
      "step": 14320
    },
    {
      "epoch": 0.32700470083519695,
      "grad_norm": 3.1875,
      "learning_rate": 1.3876302726609736e-05,
      "loss": 1.1998,
      "step": 14330
    },
    {
      "epoch": 0.3272328967185432,
      "grad_norm": 3.046875,
      "learning_rate": 1.387159761921566e-05,
      "loss": 1.2174,
      "step": 14340
    },
    {
      "epoch": 0.32746109260188944,
      "grad_norm": 2.984375,
      "learning_rate": 1.3866892511821584e-05,
      "loss": 1.2178,
      "step": 14350
    },
    {
      "epoch": 0.3276892884852357,
      "grad_norm": 2.96875,
      "learning_rate": 1.3862187404427507e-05,
      "loss": 1.2196,
      "step": 14360
    },
    {
      "epoch": 0.327917484368582,
      "grad_norm": 3.5,
      "learning_rate": 1.3857482297033431e-05,
      "loss": 1.1992,
      "step": 14370
    },
    {
      "epoch": 0.32814568025192825,
      "grad_norm": 3.078125,
      "learning_rate": 1.3852777189639355e-05,
      "loss": 1.2199,
      "step": 14380
    },
    {
      "epoch": 0.3283738761352745,
      "grad_norm": 3.21875,
      "learning_rate": 1.3848072082245278e-05,
      "loss": 1.2153,
      "step": 14390
    },
    {
      "epoch": 0.3286020720186208,
      "grad_norm": 3.0,
      "learning_rate": 1.3843366974851202e-05,
      "loss": 1.2228,
      "step": 14400
    },
    {
      "epoch": 0.32883026790196707,
      "grad_norm": 2.890625,
      "learning_rate": 1.3838661867457126e-05,
      "loss": 1.2367,
      "step": 14410
    },
    {
      "epoch": 0.32905846378531334,
      "grad_norm": 3.03125,
      "learning_rate": 1.383395676006305e-05,
      "loss": 1.1341,
      "step": 14420
    },
    {
      "epoch": 0.32928665966865955,
      "grad_norm": 3.140625,
      "learning_rate": 1.3829251652668973e-05,
      "loss": 1.2304,
      "step": 14430
    },
    {
      "epoch": 0.3295148555520058,
      "grad_norm": 2.890625,
      "learning_rate": 1.3824546545274897e-05,
      "loss": 1.2176,
      "step": 14440
    },
    {
      "epoch": 0.3297430514353521,
      "grad_norm": 3.171875,
      "learning_rate": 1.381984143788082e-05,
      "loss": 1.1924,
      "step": 14450
    },
    {
      "epoch": 0.32997124731869837,
      "grad_norm": 3.234375,
      "learning_rate": 1.3815136330486744e-05,
      "loss": 1.1911,
      "step": 14460
    },
    {
      "epoch": 0.33019944320204464,
      "grad_norm": 2.8125,
      "learning_rate": 1.3810431223092668e-05,
      "loss": 1.2127,
      "step": 14470
    },
    {
      "epoch": 0.3304276390853909,
      "grad_norm": 3.0,
      "learning_rate": 1.3805726115698592e-05,
      "loss": 1.2456,
      "step": 14480
    },
    {
      "epoch": 0.3306558349687372,
      "grad_norm": 2.8125,
      "learning_rate": 1.3801021008304516e-05,
      "loss": 1.1711,
      "step": 14490
    },
    {
      "epoch": 0.33088403085208345,
      "grad_norm": 2.84375,
      "learning_rate": 1.379631590091044e-05,
      "loss": 1.1721,
      "step": 14500
    },
    {
      "epoch": 0.33111222673542967,
      "grad_norm": 2.890625,
      "learning_rate": 1.3791610793516363e-05,
      "loss": 1.198,
      "step": 14510
    },
    {
      "epoch": 0.33134042261877594,
      "grad_norm": 2.78125,
      "learning_rate": 1.3786905686122287e-05,
      "loss": 1.2497,
      "step": 14520
    },
    {
      "epoch": 0.3315686185021222,
      "grad_norm": 3.0,
      "learning_rate": 1.378220057872821e-05,
      "loss": 1.2749,
      "step": 14530
    },
    {
      "epoch": 0.3317968143854685,
      "grad_norm": 2.734375,
      "learning_rate": 1.3777495471334134e-05,
      "loss": 1.1486,
      "step": 14540
    },
    {
      "epoch": 0.33202501026881476,
      "grad_norm": 3.265625,
      "learning_rate": 1.3772790363940058e-05,
      "loss": 1.2663,
      "step": 14550
    },
    {
      "epoch": 0.33225320615216103,
      "grad_norm": 3.5625,
      "learning_rate": 1.3768085256545982e-05,
      "loss": 1.1865,
      "step": 14560
    },
    {
      "epoch": 0.3324814020355073,
      "grad_norm": 3.0,
      "learning_rate": 1.3763380149151905e-05,
      "loss": 1.1811,
      "step": 14570
    },
    {
      "epoch": 0.3327095979188535,
      "grad_norm": 3.09375,
      "learning_rate": 1.3758675041757829e-05,
      "loss": 1.2021,
      "step": 14580
    },
    {
      "epoch": 0.3329377938021998,
      "grad_norm": 3.25,
      "learning_rate": 1.3753969934363753e-05,
      "loss": 1.1548,
      "step": 14590
    },
    {
      "epoch": 0.33316598968554606,
      "grad_norm": 3.21875,
      "learning_rate": 1.3749264826969676e-05,
      "loss": 1.1791,
      "step": 14600
    },
    {
      "epoch": 0.33339418556889233,
      "grad_norm": 2.9375,
      "learning_rate": 1.37445597195756e-05,
      "loss": 1.2549,
      "step": 14610
    },
    {
      "epoch": 0.3336223814522386,
      "grad_norm": 3.140625,
      "learning_rate": 1.3739854612181524e-05,
      "loss": 1.1835,
      "step": 14620
    },
    {
      "epoch": 0.3338505773355849,
      "grad_norm": 3.15625,
      "learning_rate": 1.3735149504787448e-05,
      "loss": 1.1639,
      "step": 14630
    },
    {
      "epoch": 0.33407877321893115,
      "grad_norm": 3.03125,
      "learning_rate": 1.3730444397393371e-05,
      "loss": 1.152,
      "step": 14640
    },
    {
      "epoch": 0.3343069691022774,
      "grad_norm": 3.359375,
      "learning_rate": 1.3725739289999295e-05,
      "loss": 1.1824,
      "step": 14650
    },
    {
      "epoch": 0.33453516498562363,
      "grad_norm": 2.8125,
      "learning_rate": 1.372103418260522e-05,
      "loss": 1.191,
      "step": 14660
    },
    {
      "epoch": 0.3347633608689699,
      "grad_norm": 2.84375,
      "learning_rate": 1.3716329075211144e-05,
      "loss": 1.2164,
      "step": 14670
    },
    {
      "epoch": 0.3349915567523162,
      "grad_norm": 2.96875,
      "learning_rate": 1.3711623967817068e-05,
      "loss": 1.2077,
      "step": 14680
    },
    {
      "epoch": 0.33521975263566245,
      "grad_norm": 3.296875,
      "learning_rate": 1.3706918860422992e-05,
      "loss": 1.1623,
      "step": 14690
    },
    {
      "epoch": 0.3354479485190087,
      "grad_norm": 3.0,
      "learning_rate": 1.3702213753028915e-05,
      "loss": 1.1797,
      "step": 14700
    },
    {
      "epoch": 0.335676144402355,
      "grad_norm": 3.046875,
      "learning_rate": 1.3697508645634839e-05,
      "loss": 1.1978,
      "step": 14710
    },
    {
      "epoch": 0.33590434028570126,
      "grad_norm": 3.1875,
      "learning_rate": 1.3692803538240763e-05,
      "loss": 1.1489,
      "step": 14720
    },
    {
      "epoch": 0.33613253616904754,
      "grad_norm": 2.703125,
      "learning_rate": 1.3688098430846686e-05,
      "loss": 1.2125,
      "step": 14730
    },
    {
      "epoch": 0.33636073205239375,
      "grad_norm": 3.09375,
      "learning_rate": 1.368339332345261e-05,
      "loss": 1.1587,
      "step": 14740
    },
    {
      "epoch": 0.33658892793574,
      "grad_norm": 2.96875,
      "learning_rate": 1.3678688216058534e-05,
      "loss": 1.1852,
      "step": 14750
    },
    {
      "epoch": 0.3368171238190863,
      "grad_norm": 3.28125,
      "learning_rate": 1.3673983108664458e-05,
      "loss": 1.2045,
      "step": 14760
    },
    {
      "epoch": 0.33704531970243257,
      "grad_norm": 3.09375,
      "learning_rate": 1.3669278001270381e-05,
      "loss": 1.1876,
      "step": 14770
    },
    {
      "epoch": 0.33727351558577884,
      "grad_norm": 3.125,
      "learning_rate": 1.3664572893876305e-05,
      "loss": 1.2033,
      "step": 14780
    },
    {
      "epoch": 0.3375017114691251,
      "grad_norm": 3.015625,
      "learning_rate": 1.3659867786482229e-05,
      "loss": 1.1502,
      "step": 14790
    },
    {
      "epoch": 0.3377299073524714,
      "grad_norm": 3.234375,
      "learning_rate": 1.3655162679088152e-05,
      "loss": 1.2277,
      "step": 14800
    },
    {
      "epoch": 0.33795810323581765,
      "grad_norm": 2.96875,
      "learning_rate": 1.3650457571694076e-05,
      "loss": 1.2208,
      "step": 14810
    },
    {
      "epoch": 0.33818629911916387,
      "grad_norm": 3.0625,
      "learning_rate": 1.36457524643e-05,
      "loss": 1.1929,
      "step": 14820
    },
    {
      "epoch": 0.33841449500251014,
      "grad_norm": 2.96875,
      "learning_rate": 1.3641047356905924e-05,
      "loss": 1.1762,
      "step": 14830
    },
    {
      "epoch": 0.3386426908858564,
      "grad_norm": 3.421875,
      "learning_rate": 1.3636342249511847e-05,
      "loss": 1.195,
      "step": 14840
    },
    {
      "epoch": 0.3388708867692027,
      "grad_norm": 2.9375,
      "learning_rate": 1.3631637142117771e-05,
      "loss": 1.1806,
      "step": 14850
    },
    {
      "epoch": 0.33909908265254896,
      "grad_norm": 3.1875,
      "learning_rate": 1.3626932034723695e-05,
      "loss": 1.2045,
      "step": 14860
    },
    {
      "epoch": 0.3393272785358952,
      "grad_norm": 3.453125,
      "learning_rate": 1.3622226927329618e-05,
      "loss": 1.2404,
      "step": 14870
    },
    {
      "epoch": 0.3395554744192415,
      "grad_norm": 3.34375,
      "learning_rate": 1.3617521819935542e-05,
      "loss": 1.2099,
      "step": 14880
    },
    {
      "epoch": 0.3397836703025877,
      "grad_norm": 3.171875,
      "learning_rate": 1.3612816712541466e-05,
      "loss": 1.198,
      "step": 14890
    },
    {
      "epoch": 0.340011866185934,
      "grad_norm": 3.375,
      "learning_rate": 1.360811160514739e-05,
      "loss": 1.2063,
      "step": 14900
    },
    {
      "epoch": 0.34024006206928026,
      "grad_norm": 3.03125,
      "learning_rate": 1.3603406497753313e-05,
      "loss": 1.1706,
      "step": 14910
    },
    {
      "epoch": 0.34046825795262653,
      "grad_norm": 3.1875,
      "learning_rate": 1.3598701390359237e-05,
      "loss": 1.2237,
      "step": 14920
    },
    {
      "epoch": 0.3406964538359728,
      "grad_norm": 3.09375,
      "learning_rate": 1.359399628296516e-05,
      "loss": 1.1967,
      "step": 14930
    },
    {
      "epoch": 0.3409246497193191,
      "grad_norm": 3.09375,
      "learning_rate": 1.3589291175571084e-05,
      "loss": 1.2034,
      "step": 14940
    },
    {
      "epoch": 0.34115284560266534,
      "grad_norm": 2.6875,
      "learning_rate": 1.3584586068177008e-05,
      "loss": 1.175,
      "step": 14950
    },
    {
      "epoch": 0.3413810414860116,
      "grad_norm": 2.890625,
      "learning_rate": 1.3579880960782932e-05,
      "loss": 1.1716,
      "step": 14960
    },
    {
      "epoch": 0.34160923736935783,
      "grad_norm": 3.03125,
      "learning_rate": 1.3575175853388855e-05,
      "loss": 1.2291,
      "step": 14970
    },
    {
      "epoch": 0.3418374332527041,
      "grad_norm": 3.03125,
      "learning_rate": 1.357047074599478e-05,
      "loss": 1.1691,
      "step": 14980
    },
    {
      "epoch": 0.3420656291360504,
      "grad_norm": 3.59375,
      "learning_rate": 1.3565765638600703e-05,
      "loss": 1.2115,
      "step": 14990
    },
    {
      "epoch": 0.34229382501939665,
      "grad_norm": 3.0,
      "learning_rate": 1.3561060531206627e-05,
      "loss": 1.1885,
      "step": 15000
    },
    {
      "epoch": 0.3425220209027429,
      "grad_norm": 3.96875,
      "learning_rate": 1.355635542381255e-05,
      "loss": 1.2239,
      "step": 15010
    },
    {
      "epoch": 0.3427502167860892,
      "grad_norm": 3.0625,
      "learning_rate": 1.3551650316418474e-05,
      "loss": 1.1962,
      "step": 15020
    },
    {
      "epoch": 0.34297841266943546,
      "grad_norm": 3.046875,
      "learning_rate": 1.3546945209024398e-05,
      "loss": 1.161,
      "step": 15030
    },
    {
      "epoch": 0.34320660855278173,
      "grad_norm": 3.15625,
      "learning_rate": 1.3542240101630321e-05,
      "loss": 1.2244,
      "step": 15040
    },
    {
      "epoch": 0.34343480443612795,
      "grad_norm": 3.296875,
      "learning_rate": 1.3537534994236245e-05,
      "loss": 1.1894,
      "step": 15050
    },
    {
      "epoch": 0.3436630003194742,
      "grad_norm": 2.75,
      "learning_rate": 1.3532829886842169e-05,
      "loss": 1.2354,
      "step": 15060
    },
    {
      "epoch": 0.3438911962028205,
      "grad_norm": 3.40625,
      "learning_rate": 1.3528124779448093e-05,
      "loss": 1.2192,
      "step": 15070
    },
    {
      "epoch": 0.34411939208616676,
      "grad_norm": 3.0,
      "learning_rate": 1.3523419672054016e-05,
      "loss": 1.2057,
      "step": 15080
    },
    {
      "epoch": 0.34434758796951304,
      "grad_norm": 2.984375,
      "learning_rate": 1.351871456465994e-05,
      "loss": 1.2015,
      "step": 15090
    },
    {
      "epoch": 0.3445757838528593,
      "grad_norm": 3.328125,
      "learning_rate": 1.3514009457265864e-05,
      "loss": 1.1839,
      "step": 15100
    },
    {
      "epoch": 0.3448039797362056,
      "grad_norm": 3.0,
      "learning_rate": 1.3509304349871787e-05,
      "loss": 1.1921,
      "step": 15110
    },
    {
      "epoch": 0.34503217561955185,
      "grad_norm": 3.234375,
      "learning_rate": 1.3504599242477711e-05,
      "loss": 1.1987,
      "step": 15120
    },
    {
      "epoch": 0.34526037150289807,
      "grad_norm": 3.421875,
      "learning_rate": 1.3499894135083635e-05,
      "loss": 1.2054,
      "step": 15130
    },
    {
      "epoch": 0.34548856738624434,
      "grad_norm": 3.078125,
      "learning_rate": 1.3495189027689559e-05,
      "loss": 1.2059,
      "step": 15140
    },
    {
      "epoch": 0.3457167632695906,
      "grad_norm": 3.171875,
      "learning_rate": 1.3490483920295482e-05,
      "loss": 1.2405,
      "step": 15150
    },
    {
      "epoch": 0.3459449591529369,
      "grad_norm": 2.875,
      "learning_rate": 1.3485778812901406e-05,
      "loss": 1.1984,
      "step": 15160
    },
    {
      "epoch": 0.34617315503628315,
      "grad_norm": 2.953125,
      "learning_rate": 1.348107370550733e-05,
      "loss": 1.219,
      "step": 15170
    },
    {
      "epoch": 0.3464013509196294,
      "grad_norm": 3.109375,
      "learning_rate": 1.3476368598113252e-05,
      "loss": 1.2116,
      "step": 15180
    },
    {
      "epoch": 0.3466295468029757,
      "grad_norm": 2.890625,
      "learning_rate": 1.3471663490719175e-05,
      "loss": 1.2076,
      "step": 15190
    },
    {
      "epoch": 0.3468577426863219,
      "grad_norm": 3.3125,
      "learning_rate": 1.34669583833251e-05,
      "loss": 1.1932,
      "step": 15200
    },
    {
      "epoch": 0.3470859385696682,
      "grad_norm": 2.84375,
      "learning_rate": 1.3462253275931023e-05,
      "loss": 1.238,
      "step": 15210
    },
    {
      "epoch": 0.34731413445301446,
      "grad_norm": 3.046875,
      "learning_rate": 1.3457548168536947e-05,
      "loss": 1.2001,
      "step": 15220
    },
    {
      "epoch": 0.3475423303363607,
      "grad_norm": 3.03125,
      "learning_rate": 1.345284306114287e-05,
      "loss": 1.1904,
      "step": 15230
    },
    {
      "epoch": 0.347770526219707,
      "grad_norm": 3.1875,
      "learning_rate": 1.3448137953748794e-05,
      "loss": 1.2016,
      "step": 15240
    },
    {
      "epoch": 0.34799872210305327,
      "grad_norm": 3.359375,
      "learning_rate": 1.3443432846354718e-05,
      "loss": 1.2336,
      "step": 15250
    },
    {
      "epoch": 0.34822691798639954,
      "grad_norm": 3.125,
      "learning_rate": 1.3438727738960641e-05,
      "loss": 1.1785,
      "step": 15260
    },
    {
      "epoch": 0.3484551138697458,
      "grad_norm": 3.03125,
      "learning_rate": 1.3434022631566565e-05,
      "loss": 1.2528,
      "step": 15270
    },
    {
      "epoch": 0.34868330975309203,
      "grad_norm": 3.1875,
      "learning_rate": 1.3429317524172489e-05,
      "loss": 1.2294,
      "step": 15280
    },
    {
      "epoch": 0.3489115056364383,
      "grad_norm": 3.046875,
      "learning_rate": 1.3424612416778413e-05,
      "loss": 1.2062,
      "step": 15290
    },
    {
      "epoch": 0.3491397015197846,
      "grad_norm": 3.328125,
      "learning_rate": 1.3419907309384336e-05,
      "loss": 1.2352,
      "step": 15300
    },
    {
      "epoch": 0.34936789740313084,
      "grad_norm": 3.5,
      "learning_rate": 1.341520220199026e-05,
      "loss": 1.1888,
      "step": 15310
    },
    {
      "epoch": 0.3495960932864771,
      "grad_norm": 3.484375,
      "learning_rate": 1.3410497094596184e-05,
      "loss": 1.2365,
      "step": 15320
    },
    {
      "epoch": 0.3498242891698234,
      "grad_norm": 3.046875,
      "learning_rate": 1.3405791987202107e-05,
      "loss": 1.2024,
      "step": 15330
    },
    {
      "epoch": 0.35005248505316966,
      "grad_norm": 3.296875,
      "learning_rate": 1.3401086879808031e-05,
      "loss": 1.218,
      "step": 15340
    },
    {
      "epoch": 0.35028068093651593,
      "grad_norm": 2.71875,
      "learning_rate": 1.3396381772413955e-05,
      "loss": 1.1904,
      "step": 15350
    },
    {
      "epoch": 0.35050887681986215,
      "grad_norm": 3.21875,
      "learning_rate": 1.3391676665019879e-05,
      "loss": 1.2249,
      "step": 15360
    },
    {
      "epoch": 0.3507370727032084,
      "grad_norm": 3.265625,
      "learning_rate": 1.3386971557625802e-05,
      "loss": 1.2309,
      "step": 15370
    },
    {
      "epoch": 0.3509652685865547,
      "grad_norm": 3.390625,
      "learning_rate": 1.3382266450231726e-05,
      "loss": 1.1939,
      "step": 15380
    },
    {
      "epoch": 0.35119346446990096,
      "grad_norm": 3.3125,
      "learning_rate": 1.337756134283765e-05,
      "loss": 1.2075,
      "step": 15390
    },
    {
      "epoch": 0.35142166035324723,
      "grad_norm": 3.234375,
      "learning_rate": 1.3372856235443573e-05,
      "loss": 1.2009,
      "step": 15400
    },
    {
      "epoch": 0.3516498562365935,
      "grad_norm": 3.5,
      "learning_rate": 1.3368151128049499e-05,
      "loss": 1.2064,
      "step": 15410
    },
    {
      "epoch": 0.3518780521199398,
      "grad_norm": 2.890625,
      "learning_rate": 1.3363446020655423e-05,
      "loss": 1.1637,
      "step": 15420
    },
    {
      "epoch": 0.352106248003286,
      "grad_norm": 2.796875,
      "learning_rate": 1.3358740913261346e-05,
      "loss": 1.2412,
      "step": 15430
    },
    {
      "epoch": 0.35233444388663226,
      "grad_norm": 3.046875,
      "learning_rate": 1.335403580586727e-05,
      "loss": 1.2063,
      "step": 15440
    },
    {
      "epoch": 0.35256263976997854,
      "grad_norm": 3.203125,
      "learning_rate": 1.3349330698473194e-05,
      "loss": 1.2261,
      "step": 15450
    },
    {
      "epoch": 0.3527908356533248,
      "grad_norm": 2.921875,
      "learning_rate": 1.3344625591079117e-05,
      "loss": 1.2231,
      "step": 15460
    },
    {
      "epoch": 0.3530190315366711,
      "grad_norm": 3.015625,
      "learning_rate": 1.3339920483685041e-05,
      "loss": 1.1942,
      "step": 15470
    },
    {
      "epoch": 0.35324722742001735,
      "grad_norm": 3.265625,
      "learning_rate": 1.3335215376290965e-05,
      "loss": 1.1827,
      "step": 15480
    },
    {
      "epoch": 0.3534754233033636,
      "grad_norm": 3.078125,
      "learning_rate": 1.3330510268896889e-05,
      "loss": 1.2152,
      "step": 15490
    },
    {
      "epoch": 0.3537036191867099,
      "grad_norm": 2.96875,
      "learning_rate": 1.3325805161502812e-05,
      "loss": 1.2246,
      "step": 15500
    },
    {
      "epoch": 0.3539318150700561,
      "grad_norm": 3.09375,
      "learning_rate": 1.3321100054108736e-05,
      "loss": 1.2056,
      "step": 15510
    },
    {
      "epoch": 0.3541600109534024,
      "grad_norm": 3.234375,
      "learning_rate": 1.331639494671466e-05,
      "loss": 1.245,
      "step": 15520
    },
    {
      "epoch": 0.35438820683674865,
      "grad_norm": 2.828125,
      "learning_rate": 1.3311689839320583e-05,
      "loss": 1.2286,
      "step": 15530
    },
    {
      "epoch": 0.3546164027200949,
      "grad_norm": 3.4375,
      "learning_rate": 1.3306984731926507e-05,
      "loss": 1.2112,
      "step": 15540
    },
    {
      "epoch": 0.3548445986034412,
      "grad_norm": 3.40625,
      "learning_rate": 1.330227962453243e-05,
      "loss": 1.2159,
      "step": 15550
    },
    {
      "epoch": 0.35507279448678747,
      "grad_norm": 3.3125,
      "learning_rate": 1.3297574517138355e-05,
      "loss": 1.186,
      "step": 15560
    },
    {
      "epoch": 0.35530099037013374,
      "grad_norm": 2.984375,
      "learning_rate": 1.3292869409744278e-05,
      "loss": 1.1474,
      "step": 15570
    },
    {
      "epoch": 0.35552918625348,
      "grad_norm": 3.140625,
      "learning_rate": 1.3288164302350202e-05,
      "loss": 1.212,
      "step": 15580
    },
    {
      "epoch": 0.35575738213682623,
      "grad_norm": 3.046875,
      "learning_rate": 1.3283459194956126e-05,
      "loss": 1.2264,
      "step": 15590
    },
    {
      "epoch": 0.3559855780201725,
      "grad_norm": 2.9375,
      "learning_rate": 1.327875408756205e-05,
      "loss": 1.2205,
      "step": 15600
    },
    {
      "epoch": 0.35621377390351877,
      "grad_norm": 3.109375,
      "learning_rate": 1.3274048980167973e-05,
      "loss": 1.1627,
      "step": 15610
    },
    {
      "epoch": 0.35644196978686504,
      "grad_norm": 3.03125,
      "learning_rate": 1.3269343872773897e-05,
      "loss": 1.185,
      "step": 15620
    },
    {
      "epoch": 0.3566701656702113,
      "grad_norm": 3.203125,
      "learning_rate": 1.326463876537982e-05,
      "loss": 1.2618,
      "step": 15630
    },
    {
      "epoch": 0.3568983615535576,
      "grad_norm": 2.984375,
      "learning_rate": 1.3259933657985744e-05,
      "loss": 1.232,
      "step": 15640
    },
    {
      "epoch": 0.35712655743690386,
      "grad_norm": 3.0625,
      "learning_rate": 1.3255228550591668e-05,
      "loss": 1.2192,
      "step": 15650
    },
    {
      "epoch": 0.35735475332025013,
      "grad_norm": 3.234375,
      "learning_rate": 1.3250523443197592e-05,
      "loss": 1.1644,
      "step": 15660
    },
    {
      "epoch": 0.35758294920359635,
      "grad_norm": 3.28125,
      "learning_rate": 1.3245818335803515e-05,
      "loss": 1.1844,
      "step": 15670
    },
    {
      "epoch": 0.3578111450869426,
      "grad_norm": 3.15625,
      "learning_rate": 1.3241113228409439e-05,
      "loss": 1.1788,
      "step": 15680
    },
    {
      "epoch": 0.3580393409702889,
      "grad_norm": 3.078125,
      "learning_rate": 1.3236408121015363e-05,
      "loss": 1.2106,
      "step": 15690
    },
    {
      "epoch": 0.35826753685363516,
      "grad_norm": 4.875,
      "learning_rate": 1.3231703013621286e-05,
      "loss": 1.1309,
      "step": 15700
    },
    {
      "epoch": 0.35849573273698143,
      "grad_norm": 2.75,
      "learning_rate": 1.322699790622721e-05,
      "loss": 1.2195,
      "step": 15710
    },
    {
      "epoch": 0.3587239286203277,
      "grad_norm": 2.921875,
      "learning_rate": 1.3222292798833134e-05,
      "loss": 1.2395,
      "step": 15720
    },
    {
      "epoch": 0.358952124503674,
      "grad_norm": 3.15625,
      "learning_rate": 1.3217587691439058e-05,
      "loss": 1.1979,
      "step": 15730
    },
    {
      "epoch": 0.3591803203870202,
      "grad_norm": 3.28125,
      "learning_rate": 1.3212882584044981e-05,
      "loss": 1.1978,
      "step": 15740
    },
    {
      "epoch": 0.35940851627036646,
      "grad_norm": 3.046875,
      "learning_rate": 1.3208177476650905e-05,
      "loss": 1.1386,
      "step": 15750
    },
    {
      "epoch": 0.35963671215371273,
      "grad_norm": 3.125,
      "learning_rate": 1.3203472369256829e-05,
      "loss": 1.2753,
      "step": 15760
    },
    {
      "epoch": 0.359864908037059,
      "grad_norm": 2.984375,
      "learning_rate": 1.3198767261862752e-05,
      "loss": 1.1789,
      "step": 15770
    },
    {
      "epoch": 0.3600931039204053,
      "grad_norm": 3.09375,
      "learning_rate": 1.3194062154468676e-05,
      "loss": 1.2123,
      "step": 15780
    },
    {
      "epoch": 0.36032129980375155,
      "grad_norm": 3.078125,
      "learning_rate": 1.31893570470746e-05,
      "loss": 1.2119,
      "step": 15790
    },
    {
      "epoch": 0.3605494956870978,
      "grad_norm": 3.3125,
      "learning_rate": 1.3184651939680524e-05,
      "loss": 1.213,
      "step": 15800
    },
    {
      "epoch": 0.3607776915704441,
      "grad_norm": 3.03125,
      "learning_rate": 1.3179946832286447e-05,
      "loss": 1.2254,
      "step": 15810
    },
    {
      "epoch": 0.3610058874537903,
      "grad_norm": 3.171875,
      "learning_rate": 1.3175241724892371e-05,
      "loss": 1.2053,
      "step": 15820
    },
    {
      "epoch": 0.3612340833371366,
      "grad_norm": 2.953125,
      "learning_rate": 1.3170536617498295e-05,
      "loss": 1.2365,
      "step": 15830
    },
    {
      "epoch": 0.36146227922048285,
      "grad_norm": 3.140625,
      "learning_rate": 1.3165831510104218e-05,
      "loss": 1.1934,
      "step": 15840
    },
    {
      "epoch": 0.3616904751038291,
      "grad_norm": 3.125,
      "learning_rate": 1.3161126402710142e-05,
      "loss": 1.1408,
      "step": 15850
    },
    {
      "epoch": 0.3619186709871754,
      "grad_norm": 3.015625,
      "learning_rate": 1.3156421295316066e-05,
      "loss": 1.2465,
      "step": 15860
    },
    {
      "epoch": 0.36214686687052167,
      "grad_norm": 2.890625,
      "learning_rate": 1.315171618792199e-05,
      "loss": 1.2194,
      "step": 15870
    },
    {
      "epoch": 0.36237506275386794,
      "grad_norm": 3.28125,
      "learning_rate": 1.3147011080527913e-05,
      "loss": 1.2426,
      "step": 15880
    },
    {
      "epoch": 0.3626032586372142,
      "grad_norm": 3.15625,
      "learning_rate": 1.3142305973133837e-05,
      "loss": 1.2381,
      "step": 15890
    },
    {
      "epoch": 0.3628314545205604,
      "grad_norm": 3.1875,
      "learning_rate": 1.313760086573976e-05,
      "loss": 1.2476,
      "step": 15900
    },
    {
      "epoch": 0.3630596504039067,
      "grad_norm": 2.953125,
      "learning_rate": 1.3132895758345684e-05,
      "loss": 1.2361,
      "step": 15910
    },
    {
      "epoch": 0.36328784628725297,
      "grad_norm": 3.3125,
      "learning_rate": 1.3128190650951608e-05,
      "loss": 1.2208,
      "step": 15920
    },
    {
      "epoch": 0.36351604217059924,
      "grad_norm": 3.140625,
      "learning_rate": 1.3123485543557532e-05,
      "loss": 1.203,
      "step": 15930
    },
    {
      "epoch": 0.3637442380539455,
      "grad_norm": 3.046875,
      "learning_rate": 1.3118780436163456e-05,
      "loss": 1.1826,
      "step": 15940
    },
    {
      "epoch": 0.3639724339372918,
      "grad_norm": 2.84375,
      "learning_rate": 1.311407532876938e-05,
      "loss": 1.1589,
      "step": 15950
    },
    {
      "epoch": 0.36420062982063806,
      "grad_norm": 3.1875,
      "learning_rate": 1.3109370221375303e-05,
      "loss": 1.1741,
      "step": 15960
    },
    {
      "epoch": 0.3644288257039843,
      "grad_norm": 3.1875,
      "learning_rate": 1.3104665113981227e-05,
      "loss": 1.1573,
      "step": 15970
    },
    {
      "epoch": 0.36465702158733054,
      "grad_norm": 3.0625,
      "learning_rate": 1.309996000658715e-05,
      "loss": 1.1857,
      "step": 15980
    },
    {
      "epoch": 0.3648852174706768,
      "grad_norm": 3.015625,
      "learning_rate": 1.3095254899193076e-05,
      "loss": 1.2343,
      "step": 15990
    },
    {
      "epoch": 0.3651134133540231,
      "grad_norm": 3.015625,
      "learning_rate": 1.3090549791799e-05,
      "loss": 1.1575,
      "step": 16000
    },
    {
      "epoch": 0.36534160923736936,
      "grad_norm": 3.140625,
      "learning_rate": 1.3085844684404923e-05,
      "loss": 1.1787,
      "step": 16010
    },
    {
      "epoch": 0.36556980512071563,
      "grad_norm": 3.0625,
      "learning_rate": 1.3081139577010847e-05,
      "loss": 1.2434,
      "step": 16020
    },
    {
      "epoch": 0.3657980010040619,
      "grad_norm": 2.984375,
      "learning_rate": 1.307643446961677e-05,
      "loss": 1.1878,
      "step": 16030
    },
    {
      "epoch": 0.3660261968874082,
      "grad_norm": 3.390625,
      "learning_rate": 1.3071729362222694e-05,
      "loss": 1.1887,
      "step": 16040
    },
    {
      "epoch": 0.3662543927707544,
      "grad_norm": 3.0,
      "learning_rate": 1.3067024254828618e-05,
      "loss": 1.1707,
      "step": 16050
    },
    {
      "epoch": 0.36648258865410066,
      "grad_norm": 3.015625,
      "learning_rate": 1.3062319147434542e-05,
      "loss": 1.1593,
      "step": 16060
    },
    {
      "epoch": 0.36671078453744693,
      "grad_norm": 2.875,
      "learning_rate": 1.3057614040040466e-05,
      "loss": 1.2315,
      "step": 16070
    },
    {
      "epoch": 0.3669389804207932,
      "grad_norm": 3.34375,
      "learning_rate": 1.305290893264639e-05,
      "loss": 1.2078,
      "step": 16080
    },
    {
      "epoch": 0.3671671763041395,
      "grad_norm": 2.90625,
      "learning_rate": 1.3048203825252313e-05,
      "loss": 1.2791,
      "step": 16090
    },
    {
      "epoch": 0.36739537218748575,
      "grad_norm": 2.875,
      "learning_rate": 1.3043498717858237e-05,
      "loss": 1.1562,
      "step": 16100
    },
    {
      "epoch": 0.367623568070832,
      "grad_norm": 2.90625,
      "learning_rate": 1.303879361046416e-05,
      "loss": 1.1838,
      "step": 16110
    },
    {
      "epoch": 0.3678517639541783,
      "grad_norm": 3.078125,
      "learning_rate": 1.3034088503070084e-05,
      "loss": 1.1935,
      "step": 16120
    },
    {
      "epoch": 0.3680799598375245,
      "grad_norm": 3.3125,
      "learning_rate": 1.3029383395676008e-05,
      "loss": 1.2088,
      "step": 16130
    },
    {
      "epoch": 0.3683081557208708,
      "grad_norm": 3.328125,
      "learning_rate": 1.3024678288281932e-05,
      "loss": 1.2538,
      "step": 16140
    },
    {
      "epoch": 0.36853635160421705,
      "grad_norm": 2.96875,
      "learning_rate": 1.3019973180887855e-05,
      "loss": 1.1922,
      "step": 16150
    },
    {
      "epoch": 0.3687645474875633,
      "grad_norm": 2.890625,
      "learning_rate": 1.3015268073493779e-05,
      "loss": 1.2076,
      "step": 16160
    },
    {
      "epoch": 0.3689927433709096,
      "grad_norm": 3.109375,
      "learning_rate": 1.3010562966099703e-05,
      "loss": 1.1793,
      "step": 16170
    },
    {
      "epoch": 0.36922093925425586,
      "grad_norm": 3.25,
      "learning_rate": 1.3005857858705626e-05,
      "loss": 1.1524,
      "step": 16180
    },
    {
      "epoch": 0.36944913513760214,
      "grad_norm": 3.625,
      "learning_rate": 1.300115275131155e-05,
      "loss": 1.2094,
      "step": 16190
    },
    {
      "epoch": 0.3696773310209484,
      "grad_norm": 2.859375,
      "learning_rate": 1.2996447643917474e-05,
      "loss": 1.1858,
      "step": 16200
    },
    {
      "epoch": 0.3699055269042946,
      "grad_norm": 3.15625,
      "learning_rate": 1.2991742536523398e-05,
      "loss": 1.1747,
      "step": 16210
    },
    {
      "epoch": 0.3701337227876409,
      "grad_norm": 3.1875,
      "learning_rate": 1.2987037429129321e-05,
      "loss": 1.1965,
      "step": 16220
    },
    {
      "epoch": 0.37036191867098717,
      "grad_norm": 2.953125,
      "learning_rate": 1.2982332321735245e-05,
      "loss": 1.2143,
      "step": 16230
    },
    {
      "epoch": 0.37059011455433344,
      "grad_norm": 3.40625,
      "learning_rate": 1.2977627214341169e-05,
      "loss": 1.2177,
      "step": 16240
    },
    {
      "epoch": 0.3708183104376797,
      "grad_norm": 2.890625,
      "learning_rate": 1.2972922106947092e-05,
      "loss": 1.2227,
      "step": 16250
    },
    {
      "epoch": 0.371046506321026,
      "grad_norm": 3.03125,
      "learning_rate": 1.2968216999553016e-05,
      "loss": 1.2451,
      "step": 16260
    },
    {
      "epoch": 0.37127470220437225,
      "grad_norm": 2.9375,
      "learning_rate": 1.296351189215894e-05,
      "loss": 1.1926,
      "step": 16270
    },
    {
      "epoch": 0.3715028980877185,
      "grad_norm": 2.875,
      "learning_rate": 1.2958806784764864e-05,
      "loss": 1.2023,
      "step": 16280
    },
    {
      "epoch": 0.37173109397106474,
      "grad_norm": 2.890625,
      "learning_rate": 1.2954101677370787e-05,
      "loss": 1.209,
      "step": 16290
    },
    {
      "epoch": 0.371959289854411,
      "grad_norm": 3.265625,
      "learning_rate": 1.2949396569976711e-05,
      "loss": 1.2786,
      "step": 16300
    },
    {
      "epoch": 0.3721874857377573,
      "grad_norm": 3.21875,
      "learning_rate": 1.2944691462582635e-05,
      "loss": 1.1895,
      "step": 16310
    },
    {
      "epoch": 0.37241568162110356,
      "grad_norm": 3.3125,
      "learning_rate": 1.2939986355188558e-05,
      "loss": 1.1487,
      "step": 16320
    },
    {
      "epoch": 0.3726438775044498,
      "grad_norm": 3.046875,
      "learning_rate": 1.2935281247794482e-05,
      "loss": 1.2483,
      "step": 16330
    },
    {
      "epoch": 0.3728720733877961,
      "grad_norm": 2.984375,
      "learning_rate": 1.2930576140400406e-05,
      "loss": 1.2016,
      "step": 16340
    },
    {
      "epoch": 0.37310026927114237,
      "grad_norm": 2.90625,
      "learning_rate": 1.292587103300633e-05,
      "loss": 1.1341,
      "step": 16350
    },
    {
      "epoch": 0.3733284651544886,
      "grad_norm": 3.046875,
      "learning_rate": 1.2921165925612253e-05,
      "loss": 1.1641,
      "step": 16360
    },
    {
      "epoch": 0.37355666103783486,
      "grad_norm": 2.859375,
      "learning_rate": 1.2916460818218177e-05,
      "loss": 1.2063,
      "step": 16370
    },
    {
      "epoch": 0.37378485692118113,
      "grad_norm": 3.0,
      "learning_rate": 1.29117557108241e-05,
      "loss": 1.2167,
      "step": 16380
    },
    {
      "epoch": 0.3740130528045274,
      "grad_norm": 3.046875,
      "learning_rate": 1.2907050603430024e-05,
      "loss": 1.167,
      "step": 16390
    },
    {
      "epoch": 0.3742412486878737,
      "grad_norm": 3.109375,
      "learning_rate": 1.2902345496035948e-05,
      "loss": 1.1908,
      "step": 16400
    },
    {
      "epoch": 0.37446944457121994,
      "grad_norm": 2.9375,
      "learning_rate": 1.2897640388641872e-05,
      "loss": 1.2402,
      "step": 16410
    },
    {
      "epoch": 0.3746976404545662,
      "grad_norm": 3.0,
      "learning_rate": 1.2892935281247795e-05,
      "loss": 1.1716,
      "step": 16420
    },
    {
      "epoch": 0.3749258363379125,
      "grad_norm": 3.21875,
      "learning_rate": 1.288823017385372e-05,
      "loss": 1.1787,
      "step": 16430
    },
    {
      "epoch": 0.3751540322212587,
      "grad_norm": 2.96875,
      "learning_rate": 1.2883525066459643e-05,
      "loss": 1.1748,
      "step": 16440
    },
    {
      "epoch": 0.375382228104605,
      "grad_norm": 3.078125,
      "learning_rate": 1.2878819959065567e-05,
      "loss": 1.2004,
      "step": 16450
    },
    {
      "epoch": 0.37561042398795125,
      "grad_norm": 3.015625,
      "learning_rate": 1.287411485167149e-05,
      "loss": 1.1846,
      "step": 16460
    },
    {
      "epoch": 0.3758386198712975,
      "grad_norm": 2.796875,
      "learning_rate": 1.2869409744277414e-05,
      "loss": 1.1962,
      "step": 16470
    },
    {
      "epoch": 0.3760668157546438,
      "grad_norm": 3.0625,
      "learning_rate": 1.2864704636883338e-05,
      "loss": 1.2507,
      "step": 16480
    },
    {
      "epoch": 0.37629501163799006,
      "grad_norm": 3.15625,
      "learning_rate": 1.2859999529489261e-05,
      "loss": 1.1598,
      "step": 16490
    },
    {
      "epoch": 0.37652320752133633,
      "grad_norm": 3.15625,
      "learning_rate": 1.2855294422095185e-05,
      "loss": 1.1638,
      "step": 16500
    },
    {
      "epoch": 0.3767514034046826,
      "grad_norm": 2.875,
      "learning_rate": 1.2850589314701109e-05,
      "loss": 1.2184,
      "step": 16510
    },
    {
      "epoch": 0.3769795992880288,
      "grad_norm": 2.90625,
      "learning_rate": 1.2845884207307033e-05,
      "loss": 1.2678,
      "step": 16520
    },
    {
      "epoch": 0.3772077951713751,
      "grad_norm": 3.046875,
      "learning_rate": 1.2841179099912956e-05,
      "loss": 1.1935,
      "step": 16530
    },
    {
      "epoch": 0.37743599105472136,
      "grad_norm": 3.21875,
      "learning_rate": 1.283647399251888e-05,
      "loss": 1.2447,
      "step": 16540
    },
    {
      "epoch": 0.37766418693806764,
      "grad_norm": 3.15625,
      "learning_rate": 1.2831768885124804e-05,
      "loss": 1.1932,
      "step": 16550
    },
    {
      "epoch": 0.3778923828214139,
      "grad_norm": 3.21875,
      "learning_rate": 1.2827063777730727e-05,
      "loss": 1.1753,
      "step": 16560
    },
    {
      "epoch": 0.3781205787047602,
      "grad_norm": 3.078125,
      "learning_rate": 1.2822358670336653e-05,
      "loss": 1.224,
      "step": 16570
    },
    {
      "epoch": 0.37834877458810645,
      "grad_norm": 3.015625,
      "learning_rate": 1.2817653562942577e-05,
      "loss": 1.1551,
      "step": 16580
    },
    {
      "epoch": 0.37857697047145267,
      "grad_norm": 2.984375,
      "learning_rate": 1.28129484555485e-05,
      "loss": 1.1638,
      "step": 16590
    },
    {
      "epoch": 0.37880516635479894,
      "grad_norm": 3.078125,
      "learning_rate": 1.2808243348154424e-05,
      "loss": 1.1726,
      "step": 16600
    },
    {
      "epoch": 0.3790333622381452,
      "grad_norm": 3.109375,
      "learning_rate": 1.2803538240760348e-05,
      "loss": 1.1901,
      "step": 16610
    },
    {
      "epoch": 0.3792615581214915,
      "grad_norm": 3.0625,
      "learning_rate": 1.2798833133366271e-05,
      "loss": 1.2205,
      "step": 16620
    },
    {
      "epoch": 0.37948975400483775,
      "grad_norm": 3.09375,
      "learning_rate": 1.2794128025972195e-05,
      "loss": 1.1921,
      "step": 16630
    },
    {
      "epoch": 0.379717949888184,
      "grad_norm": 2.921875,
      "learning_rate": 1.2789422918578119e-05,
      "loss": 1.1738,
      "step": 16640
    },
    {
      "epoch": 0.3799461457715303,
      "grad_norm": 2.84375,
      "learning_rate": 1.2784717811184043e-05,
      "loss": 1.2036,
      "step": 16650
    },
    {
      "epoch": 0.38017434165487657,
      "grad_norm": 2.828125,
      "learning_rate": 1.2780012703789966e-05,
      "loss": 1.2112,
      "step": 16660
    },
    {
      "epoch": 0.3804025375382228,
      "grad_norm": 3.265625,
      "learning_rate": 1.277530759639589e-05,
      "loss": 1.2476,
      "step": 16670
    },
    {
      "epoch": 0.38063073342156906,
      "grad_norm": 3.375,
      "learning_rate": 1.2770602489001814e-05,
      "loss": 1.1245,
      "step": 16680
    },
    {
      "epoch": 0.38085892930491533,
      "grad_norm": 3.0625,
      "learning_rate": 1.2765897381607737e-05,
      "loss": 1.1732,
      "step": 16690
    },
    {
      "epoch": 0.3810871251882616,
      "grad_norm": 3.0,
      "learning_rate": 1.2761192274213661e-05,
      "loss": 1.1911,
      "step": 16700
    },
    {
      "epoch": 0.38131532107160787,
      "grad_norm": 3.0625,
      "learning_rate": 1.2756487166819585e-05,
      "loss": 1.1713,
      "step": 16710
    },
    {
      "epoch": 0.38154351695495414,
      "grad_norm": 3.15625,
      "learning_rate": 1.2751782059425509e-05,
      "loss": 1.2042,
      "step": 16720
    },
    {
      "epoch": 0.3817717128383004,
      "grad_norm": 2.90625,
      "learning_rate": 1.2747076952031432e-05,
      "loss": 1.1705,
      "step": 16730
    },
    {
      "epoch": 0.3819999087216467,
      "grad_norm": 3.265625,
      "learning_rate": 1.2742371844637356e-05,
      "loss": 1.2086,
      "step": 16740
    },
    {
      "epoch": 0.3822281046049929,
      "grad_norm": 2.765625,
      "learning_rate": 1.273766673724328e-05,
      "loss": 1.1534,
      "step": 16750
    },
    {
      "epoch": 0.3824563004883392,
      "grad_norm": 2.84375,
      "learning_rate": 1.2732961629849203e-05,
      "loss": 1.2093,
      "step": 16760
    },
    {
      "epoch": 0.38268449637168545,
      "grad_norm": 3.078125,
      "learning_rate": 1.2728256522455127e-05,
      "loss": 1.2324,
      "step": 16770
    },
    {
      "epoch": 0.3829126922550317,
      "grad_norm": 3.03125,
      "learning_rate": 1.272355141506105e-05,
      "loss": 1.1854,
      "step": 16780
    },
    {
      "epoch": 0.383140888138378,
      "grad_norm": 3.421875,
      "learning_rate": 1.2718846307666975e-05,
      "loss": 1.2221,
      "step": 16790
    },
    {
      "epoch": 0.38336908402172426,
      "grad_norm": 3.109375,
      "learning_rate": 1.2714141200272898e-05,
      "loss": 1.2632,
      "step": 16800
    },
    {
      "epoch": 0.38359727990507053,
      "grad_norm": 3.03125,
      "learning_rate": 1.2709436092878822e-05,
      "loss": 1.2019,
      "step": 16810
    },
    {
      "epoch": 0.3838254757884168,
      "grad_norm": 3.09375,
      "learning_rate": 1.2704730985484746e-05,
      "loss": 1.2069,
      "step": 16820
    },
    {
      "epoch": 0.384053671671763,
      "grad_norm": 3.21875,
      "learning_rate": 1.270002587809067e-05,
      "loss": 1.1525,
      "step": 16830
    },
    {
      "epoch": 0.3842818675551093,
      "grad_norm": 2.875,
      "learning_rate": 1.2695320770696593e-05,
      "loss": 1.1976,
      "step": 16840
    },
    {
      "epoch": 0.38451006343845556,
      "grad_norm": 3.046875,
      "learning_rate": 1.2690615663302517e-05,
      "loss": 1.2349,
      "step": 16850
    },
    {
      "epoch": 0.38473825932180183,
      "grad_norm": 2.921875,
      "learning_rate": 1.268591055590844e-05,
      "loss": 1.2034,
      "step": 16860
    },
    {
      "epoch": 0.3849664552051481,
      "grad_norm": 3.078125,
      "learning_rate": 1.2681205448514364e-05,
      "loss": 1.1914,
      "step": 16870
    },
    {
      "epoch": 0.3851946510884944,
      "grad_norm": 3.296875,
      "learning_rate": 1.2676500341120288e-05,
      "loss": 1.2068,
      "step": 16880
    },
    {
      "epoch": 0.38542284697184065,
      "grad_norm": 2.90625,
      "learning_rate": 1.2671795233726212e-05,
      "loss": 1.1802,
      "step": 16890
    },
    {
      "epoch": 0.38565104285518687,
      "grad_norm": 3.359375,
      "learning_rate": 1.2667090126332135e-05,
      "loss": 1.1785,
      "step": 16900
    },
    {
      "epoch": 0.38587923873853314,
      "grad_norm": 3.0625,
      "learning_rate": 1.2662385018938059e-05,
      "loss": 1.2529,
      "step": 16910
    },
    {
      "epoch": 0.3861074346218794,
      "grad_norm": 3.21875,
      "learning_rate": 1.2657679911543983e-05,
      "loss": 1.165,
      "step": 16920
    },
    {
      "epoch": 0.3863356305052257,
      "grad_norm": 3.234375,
      "learning_rate": 1.2652974804149907e-05,
      "loss": 1.2021,
      "step": 16930
    },
    {
      "epoch": 0.38656382638857195,
      "grad_norm": 3.296875,
      "learning_rate": 1.264826969675583e-05,
      "loss": 1.1855,
      "step": 16940
    },
    {
      "epoch": 0.3867920222719182,
      "grad_norm": 3.515625,
      "learning_rate": 1.2643564589361754e-05,
      "loss": 1.2521,
      "step": 16950
    },
    {
      "epoch": 0.3870202181552645,
      "grad_norm": 2.953125,
      "learning_rate": 1.2638859481967678e-05,
      "loss": 1.1772,
      "step": 16960
    },
    {
      "epoch": 0.38724841403861077,
      "grad_norm": 3.15625,
      "learning_rate": 1.2634154374573601e-05,
      "loss": 1.2384,
      "step": 16970
    },
    {
      "epoch": 0.387476609921957,
      "grad_norm": 2.96875,
      "learning_rate": 1.2629449267179525e-05,
      "loss": 1.1623,
      "step": 16980
    },
    {
      "epoch": 0.38770480580530325,
      "grad_norm": 3.09375,
      "learning_rate": 1.2624744159785449e-05,
      "loss": 1.1692,
      "step": 16990
    },
    {
      "epoch": 0.3879330016886495,
      "grad_norm": 3.140625,
      "learning_rate": 1.2620039052391373e-05,
      "loss": 1.1925,
      "step": 17000
    },
    {
      "epoch": 0.3881611975719958,
      "grad_norm": 2.984375,
      "learning_rate": 1.2615333944997296e-05,
      "loss": 1.2224,
      "step": 17010
    },
    {
      "epoch": 0.38838939345534207,
      "grad_norm": 3.1875,
      "learning_rate": 1.261062883760322e-05,
      "loss": 1.1709,
      "step": 17020
    },
    {
      "epoch": 0.38861758933868834,
      "grad_norm": 2.9375,
      "learning_rate": 1.2605923730209142e-05,
      "loss": 1.2105,
      "step": 17030
    },
    {
      "epoch": 0.3888457852220346,
      "grad_norm": 3.078125,
      "learning_rate": 1.2601218622815066e-05,
      "loss": 1.1966,
      "step": 17040
    },
    {
      "epoch": 0.3890739811053809,
      "grad_norm": 3.203125,
      "learning_rate": 1.259651351542099e-05,
      "loss": 1.235,
      "step": 17050
    },
    {
      "epoch": 0.3893021769887271,
      "grad_norm": 3.421875,
      "learning_rate": 1.2591808408026913e-05,
      "loss": 1.208,
      "step": 17060
    },
    {
      "epoch": 0.38953037287207337,
      "grad_norm": 2.953125,
      "learning_rate": 1.2587103300632837e-05,
      "loss": 1.1666,
      "step": 17070
    },
    {
      "epoch": 0.38975856875541964,
      "grad_norm": 2.9375,
      "learning_rate": 1.258239819323876e-05,
      "loss": 1.2018,
      "step": 17080
    },
    {
      "epoch": 0.3899867646387659,
      "grad_norm": 3.359375,
      "learning_rate": 1.2577693085844684e-05,
      "loss": 1.2128,
      "step": 17090
    },
    {
      "epoch": 0.3902149605221122,
      "grad_norm": 3.0,
      "learning_rate": 1.2572987978450608e-05,
      "loss": 1.1985,
      "step": 17100
    },
    {
      "epoch": 0.39044315640545846,
      "grad_norm": 3.015625,
      "learning_rate": 1.2568282871056532e-05,
      "loss": 1.1114,
      "step": 17110
    },
    {
      "epoch": 0.39067135228880473,
      "grad_norm": 3.4375,
      "learning_rate": 1.2563577763662455e-05,
      "loss": 1.2021,
      "step": 17120
    },
    {
      "epoch": 0.390899548172151,
      "grad_norm": 3.0,
      "learning_rate": 1.2558872656268379e-05,
      "loss": 1.1807,
      "step": 17130
    },
    {
      "epoch": 0.3911277440554972,
      "grad_norm": 3.109375,
      "learning_rate": 1.2554167548874303e-05,
      "loss": 1.1691,
      "step": 17140
    },
    {
      "epoch": 0.3913559399388435,
      "grad_norm": 3.078125,
      "learning_rate": 1.2549462441480226e-05,
      "loss": 1.1731,
      "step": 17150
    },
    {
      "epoch": 0.39158413582218976,
      "grad_norm": 2.984375,
      "learning_rate": 1.254475733408615e-05,
      "loss": 1.2421,
      "step": 17160
    },
    {
      "epoch": 0.39181233170553603,
      "grad_norm": 2.96875,
      "learning_rate": 1.2540052226692074e-05,
      "loss": 1.171,
      "step": 17170
    },
    {
      "epoch": 0.3920405275888823,
      "grad_norm": 2.828125,
      "learning_rate": 1.2535347119297998e-05,
      "loss": 1.1595,
      "step": 17180
    },
    {
      "epoch": 0.3922687234722286,
      "grad_norm": 3.46875,
      "learning_rate": 1.2530642011903921e-05,
      "loss": 1.202,
      "step": 17190
    },
    {
      "epoch": 0.39249691935557485,
      "grad_norm": 3.296875,
      "learning_rate": 1.2525936904509845e-05,
      "loss": 1.1483,
      "step": 17200
    },
    {
      "epoch": 0.39272511523892106,
      "grad_norm": 3.15625,
      "learning_rate": 1.2521231797115769e-05,
      "loss": 1.1783,
      "step": 17210
    },
    {
      "epoch": 0.39295331112226733,
      "grad_norm": 3.015625,
      "learning_rate": 1.2516526689721692e-05,
      "loss": 1.263,
      "step": 17220
    },
    {
      "epoch": 0.3931815070056136,
      "grad_norm": 3.1875,
      "learning_rate": 1.2511821582327616e-05,
      "loss": 1.1593,
      "step": 17230
    },
    {
      "epoch": 0.3934097028889599,
      "grad_norm": 3.046875,
      "learning_rate": 1.250711647493354e-05,
      "loss": 1.2072,
      "step": 17240
    },
    {
      "epoch": 0.39363789877230615,
      "grad_norm": 3.125,
      "learning_rate": 1.2502411367539464e-05,
      "loss": 1.2129,
      "step": 17250
    },
    {
      "epoch": 0.3938660946556524,
      "grad_norm": 3.109375,
      "learning_rate": 1.2497706260145387e-05,
      "loss": 1.1871,
      "step": 17260
    },
    {
      "epoch": 0.3940942905389987,
      "grad_norm": 3.1875,
      "learning_rate": 1.2493001152751311e-05,
      "loss": 1.1784,
      "step": 17270
    },
    {
      "epoch": 0.39432248642234496,
      "grad_norm": 3.25,
      "learning_rate": 1.2488296045357235e-05,
      "loss": 1.1978,
      "step": 17280
    },
    {
      "epoch": 0.3945506823056912,
      "grad_norm": 3.125,
      "learning_rate": 1.2483590937963158e-05,
      "loss": 1.2005,
      "step": 17290
    },
    {
      "epoch": 0.39477887818903745,
      "grad_norm": 3.125,
      "learning_rate": 1.2478885830569082e-05,
      "loss": 1.196,
      "step": 17300
    },
    {
      "epoch": 0.3950070740723837,
      "grad_norm": 2.921875,
      "learning_rate": 1.2474180723175006e-05,
      "loss": 1.1712,
      "step": 17310
    },
    {
      "epoch": 0.39523526995573,
      "grad_norm": 3.546875,
      "learning_rate": 1.2469475615780931e-05,
      "loss": 1.2016,
      "step": 17320
    },
    {
      "epoch": 0.39546346583907627,
      "grad_norm": 3.234375,
      "learning_rate": 1.2464770508386855e-05,
      "loss": 1.1732,
      "step": 17330
    },
    {
      "epoch": 0.39569166172242254,
      "grad_norm": 3.1875,
      "learning_rate": 1.2460065400992779e-05,
      "loss": 1.171,
      "step": 17340
    },
    {
      "epoch": 0.3959198576057688,
      "grad_norm": 3.453125,
      "learning_rate": 1.2455360293598702e-05,
      "loss": 1.1783,
      "step": 17350
    },
    {
      "epoch": 0.3961480534891151,
      "grad_norm": 3.125,
      "learning_rate": 1.2450655186204626e-05,
      "loss": 1.1894,
      "step": 17360
    },
    {
      "epoch": 0.3963762493724613,
      "grad_norm": 3.0625,
      "learning_rate": 1.244595007881055e-05,
      "loss": 1.1984,
      "step": 17370
    },
    {
      "epoch": 0.39660444525580757,
      "grad_norm": 3.28125,
      "learning_rate": 1.2441244971416474e-05,
      "loss": 1.2006,
      "step": 17380
    },
    {
      "epoch": 0.39683264113915384,
      "grad_norm": 3.03125,
      "learning_rate": 1.2436539864022397e-05,
      "loss": 1.2496,
      "step": 17390
    },
    {
      "epoch": 0.3970608370225001,
      "grad_norm": 3.03125,
      "learning_rate": 1.2431834756628321e-05,
      "loss": 1.2214,
      "step": 17400
    },
    {
      "epoch": 0.3972890329058464,
      "grad_norm": 3.109375,
      "learning_rate": 1.2427129649234245e-05,
      "loss": 1.2505,
      "step": 17410
    },
    {
      "epoch": 0.39751722878919266,
      "grad_norm": 2.984375,
      "learning_rate": 1.2422424541840168e-05,
      "loss": 1.1794,
      "step": 17420
    },
    {
      "epoch": 0.3977454246725389,
      "grad_norm": 3.125,
      "learning_rate": 1.2417719434446092e-05,
      "loss": 1.2245,
      "step": 17430
    },
    {
      "epoch": 0.3979736205558852,
      "grad_norm": 2.96875,
      "learning_rate": 1.2413014327052016e-05,
      "loss": 1.2643,
      "step": 17440
    },
    {
      "epoch": 0.3982018164392314,
      "grad_norm": 3.515625,
      "learning_rate": 1.240830921965794e-05,
      "loss": 1.1768,
      "step": 17450
    },
    {
      "epoch": 0.3984300123225777,
      "grad_norm": 2.84375,
      "learning_rate": 1.2403604112263863e-05,
      "loss": 1.2245,
      "step": 17460
    },
    {
      "epoch": 0.39865820820592396,
      "grad_norm": 3.375,
      "learning_rate": 1.2398899004869787e-05,
      "loss": 1.2175,
      "step": 17470
    },
    {
      "epoch": 0.39888640408927023,
      "grad_norm": 2.828125,
      "learning_rate": 1.239419389747571e-05,
      "loss": 1.1696,
      "step": 17480
    },
    {
      "epoch": 0.3991145999726165,
      "grad_norm": 2.984375,
      "learning_rate": 1.2389488790081634e-05,
      "loss": 1.2177,
      "step": 17490
    },
    {
      "epoch": 0.3993427958559628,
      "grad_norm": 3.234375,
      "learning_rate": 1.2384783682687558e-05,
      "loss": 1.186,
      "step": 17500
    },
    {
      "epoch": 0.39957099173930904,
      "grad_norm": 3.140625,
      "learning_rate": 1.2380078575293482e-05,
      "loss": 1.2118,
      "step": 17510
    },
    {
      "epoch": 0.39979918762265526,
      "grad_norm": 3.046875,
      "learning_rate": 1.2375373467899406e-05,
      "loss": 1.1997,
      "step": 17520
    },
    {
      "epoch": 0.40002738350600153,
      "grad_norm": 2.984375,
      "learning_rate": 1.237066836050533e-05,
      "loss": 1.2258,
      "step": 17530
    },
    {
      "epoch": 0.4002555793893478,
      "grad_norm": 3.15625,
      "learning_rate": 1.2365963253111253e-05,
      "loss": 1.1998,
      "step": 17540
    },
    {
      "epoch": 0.4004837752726941,
      "grad_norm": 3.109375,
      "learning_rate": 1.2361258145717177e-05,
      "loss": 1.1496,
      "step": 17550
    },
    {
      "epoch": 0.40071197115604035,
      "grad_norm": 3.140625,
      "learning_rate": 1.23565530383231e-05,
      "loss": 1.1689,
      "step": 17560
    },
    {
      "epoch": 0.4009401670393866,
      "grad_norm": 2.90625,
      "learning_rate": 1.2351847930929024e-05,
      "loss": 1.2477,
      "step": 17570
    },
    {
      "epoch": 0.4011683629227329,
      "grad_norm": 3.03125,
      "learning_rate": 1.2347142823534948e-05,
      "loss": 1.2212,
      "step": 17580
    },
    {
      "epoch": 0.40139655880607916,
      "grad_norm": 3.0625,
      "learning_rate": 1.2342437716140872e-05,
      "loss": 1.1939,
      "step": 17590
    },
    {
      "epoch": 0.4016247546894254,
      "grad_norm": 2.96875,
      "learning_rate": 1.2337732608746795e-05,
      "loss": 1.167,
      "step": 17600
    },
    {
      "epoch": 0.40185295057277165,
      "grad_norm": 2.890625,
      "learning_rate": 1.2333027501352719e-05,
      "loss": 1.2247,
      "step": 17610
    },
    {
      "epoch": 0.4020811464561179,
      "grad_norm": 3.546875,
      "learning_rate": 1.2328322393958643e-05,
      "loss": 1.1727,
      "step": 17620
    },
    {
      "epoch": 0.4023093423394642,
      "grad_norm": 3.25,
      "learning_rate": 1.2323617286564566e-05,
      "loss": 1.1988,
      "step": 17630
    },
    {
      "epoch": 0.40253753822281046,
      "grad_norm": 2.890625,
      "learning_rate": 1.231891217917049e-05,
      "loss": 1.1494,
      "step": 17640
    },
    {
      "epoch": 0.40276573410615674,
      "grad_norm": 3.0625,
      "learning_rate": 1.2314207071776414e-05,
      "loss": 1.228,
      "step": 17650
    },
    {
      "epoch": 0.402993929989503,
      "grad_norm": 3.09375,
      "learning_rate": 1.2309501964382338e-05,
      "loss": 1.1423,
      "step": 17660
    },
    {
      "epoch": 0.4032221258728493,
      "grad_norm": 3.453125,
      "learning_rate": 1.2304796856988261e-05,
      "loss": 1.2354,
      "step": 17670
    },
    {
      "epoch": 0.4034503217561955,
      "grad_norm": 3.125,
      "learning_rate": 1.2300091749594185e-05,
      "loss": 1.2779,
      "step": 17680
    },
    {
      "epoch": 0.40367851763954177,
      "grad_norm": 3.0,
      "learning_rate": 1.2295386642200109e-05,
      "loss": 1.2069,
      "step": 17690
    },
    {
      "epoch": 0.40390671352288804,
      "grad_norm": 3.109375,
      "learning_rate": 1.2290681534806032e-05,
      "loss": 1.1496,
      "step": 17700
    },
    {
      "epoch": 0.4041349094062343,
      "grad_norm": 3.21875,
      "learning_rate": 1.2285976427411956e-05,
      "loss": 1.1515,
      "step": 17710
    },
    {
      "epoch": 0.4043631052895806,
      "grad_norm": 3.109375,
      "learning_rate": 1.228127132001788e-05,
      "loss": 1.1678,
      "step": 17720
    },
    {
      "epoch": 0.40459130117292685,
      "grad_norm": 3.046875,
      "learning_rate": 1.2276566212623804e-05,
      "loss": 1.1911,
      "step": 17730
    },
    {
      "epoch": 0.4048194970562731,
      "grad_norm": 3.140625,
      "learning_rate": 1.2271861105229727e-05,
      "loss": 1.2232,
      "step": 17740
    },
    {
      "epoch": 0.40504769293961934,
      "grad_norm": 3.296875,
      "learning_rate": 1.2267155997835651e-05,
      "loss": 1.211,
      "step": 17750
    },
    {
      "epoch": 0.4052758888229656,
      "grad_norm": 3.078125,
      "learning_rate": 1.2262450890441575e-05,
      "loss": 1.1933,
      "step": 17760
    },
    {
      "epoch": 0.4055040847063119,
      "grad_norm": 2.984375,
      "learning_rate": 1.2257745783047498e-05,
      "loss": 1.1882,
      "step": 17770
    },
    {
      "epoch": 0.40573228058965816,
      "grad_norm": 2.921875,
      "learning_rate": 1.2253040675653422e-05,
      "loss": 1.1635,
      "step": 17780
    },
    {
      "epoch": 0.40596047647300443,
      "grad_norm": 3.203125,
      "learning_rate": 1.2248335568259346e-05,
      "loss": 1.214,
      "step": 17790
    },
    {
      "epoch": 0.4061886723563507,
      "grad_norm": 3.03125,
      "learning_rate": 1.224363046086527e-05,
      "loss": 1.2201,
      "step": 17800
    },
    {
      "epoch": 0.40641686823969697,
      "grad_norm": 3.09375,
      "learning_rate": 1.2238925353471193e-05,
      "loss": 1.1606,
      "step": 17810
    },
    {
      "epoch": 0.40664506412304324,
      "grad_norm": 3.03125,
      "learning_rate": 1.2234220246077117e-05,
      "loss": 1.2044,
      "step": 17820
    },
    {
      "epoch": 0.40687326000638946,
      "grad_norm": 3.53125,
      "learning_rate": 1.222951513868304e-05,
      "loss": 1.2152,
      "step": 17830
    },
    {
      "epoch": 0.40710145588973573,
      "grad_norm": 3.015625,
      "learning_rate": 1.2224810031288964e-05,
      "loss": 1.1938,
      "step": 17840
    },
    {
      "epoch": 0.407329651773082,
      "grad_norm": 3.265625,
      "learning_rate": 1.2220104923894888e-05,
      "loss": 1.1626,
      "step": 17850
    },
    {
      "epoch": 0.4075578476564283,
      "grad_norm": 3.328125,
      "learning_rate": 1.2215399816500812e-05,
      "loss": 1.189,
      "step": 17860
    },
    {
      "epoch": 0.40778604353977455,
      "grad_norm": 3.34375,
      "learning_rate": 1.2210694709106735e-05,
      "loss": 1.223,
      "step": 17870
    },
    {
      "epoch": 0.4080142394231208,
      "grad_norm": 3.28125,
      "learning_rate": 1.220598960171266e-05,
      "loss": 1.2124,
      "step": 17880
    },
    {
      "epoch": 0.4082424353064671,
      "grad_norm": 3.03125,
      "learning_rate": 1.2201284494318583e-05,
      "loss": 1.1578,
      "step": 17890
    },
    {
      "epoch": 0.40847063118981336,
      "grad_norm": 3.03125,
      "learning_rate": 1.2196579386924508e-05,
      "loss": 1.2152,
      "step": 17900
    },
    {
      "epoch": 0.4086988270731596,
      "grad_norm": 3.296875,
      "learning_rate": 1.2191874279530432e-05,
      "loss": 1.2107,
      "step": 17910
    },
    {
      "epoch": 0.40892702295650585,
      "grad_norm": 3.046875,
      "learning_rate": 1.2187169172136356e-05,
      "loss": 1.2041,
      "step": 17920
    },
    {
      "epoch": 0.4091552188398521,
      "grad_norm": 3.15625,
      "learning_rate": 1.218246406474228e-05,
      "loss": 1.2158,
      "step": 17930
    },
    {
      "epoch": 0.4093834147231984,
      "grad_norm": 3.03125,
      "learning_rate": 1.2177758957348203e-05,
      "loss": 1.2221,
      "step": 17940
    },
    {
      "epoch": 0.40961161060654466,
      "grad_norm": 3.0,
      "learning_rate": 1.2173053849954127e-05,
      "loss": 1.1839,
      "step": 17950
    },
    {
      "epoch": 0.40983980648989093,
      "grad_norm": 3.109375,
      "learning_rate": 1.216834874256005e-05,
      "loss": 1.1514,
      "step": 17960
    },
    {
      "epoch": 0.4100680023732372,
      "grad_norm": 2.71875,
      "learning_rate": 1.2163643635165974e-05,
      "loss": 1.225,
      "step": 17970
    },
    {
      "epoch": 0.4102961982565835,
      "grad_norm": 3.28125,
      "learning_rate": 1.2158938527771898e-05,
      "loss": 1.2322,
      "step": 17980
    },
    {
      "epoch": 0.4105243941399297,
      "grad_norm": 3.125,
      "learning_rate": 1.2154233420377822e-05,
      "loss": 1.1888,
      "step": 17990
    },
    {
      "epoch": 0.41075259002327597,
      "grad_norm": 2.984375,
      "learning_rate": 1.2149528312983745e-05,
      "loss": 1.1896,
      "step": 18000
    },
    {
      "epoch": 0.41098078590662224,
      "grad_norm": 3.015625,
      "learning_rate": 1.2144823205589669e-05,
      "loss": 1.1396,
      "step": 18010
    },
    {
      "epoch": 0.4112089817899685,
      "grad_norm": 3.3125,
      "learning_rate": 1.2140118098195593e-05,
      "loss": 1.2043,
      "step": 18020
    },
    {
      "epoch": 0.4114371776733148,
      "grad_norm": 3.03125,
      "learning_rate": 1.2135412990801517e-05,
      "loss": 1.1788,
      "step": 18030
    },
    {
      "epoch": 0.41166537355666105,
      "grad_norm": 3.234375,
      "learning_rate": 1.213070788340744e-05,
      "loss": 1.1903,
      "step": 18040
    },
    {
      "epoch": 0.4118935694400073,
      "grad_norm": 3.09375,
      "learning_rate": 1.2126002776013364e-05,
      "loss": 1.2159,
      "step": 18050
    },
    {
      "epoch": 0.41212176532335354,
      "grad_norm": 3.171875,
      "learning_rate": 1.2121297668619288e-05,
      "loss": 1.2249,
      "step": 18060
    },
    {
      "epoch": 0.4123499612066998,
      "grad_norm": 2.90625,
      "learning_rate": 1.2116592561225211e-05,
      "loss": 1.1912,
      "step": 18070
    },
    {
      "epoch": 0.4125781570900461,
      "grad_norm": 3.21875,
      "learning_rate": 1.2111887453831135e-05,
      "loss": 1.1476,
      "step": 18080
    },
    {
      "epoch": 0.41280635297339235,
      "grad_norm": 2.96875,
      "learning_rate": 1.2107182346437059e-05,
      "loss": 1.127,
      "step": 18090
    },
    {
      "epoch": 0.4130345488567386,
      "grad_norm": 2.96875,
      "learning_rate": 1.2102477239042983e-05,
      "loss": 1.2087,
      "step": 18100
    },
    {
      "epoch": 0.4132627447400849,
      "grad_norm": 3.515625,
      "learning_rate": 1.2097772131648906e-05,
      "loss": 1.2124,
      "step": 18110
    },
    {
      "epoch": 0.41349094062343117,
      "grad_norm": 2.78125,
      "learning_rate": 1.209306702425483e-05,
      "loss": 1.1929,
      "step": 18120
    },
    {
      "epoch": 0.41371913650677744,
      "grad_norm": 3.390625,
      "learning_rate": 1.2088361916860754e-05,
      "loss": 1.1206,
      "step": 18130
    },
    {
      "epoch": 0.41394733239012366,
      "grad_norm": 3.234375,
      "learning_rate": 1.2083656809466677e-05,
      "loss": 1.1987,
      "step": 18140
    },
    {
      "epoch": 0.41417552827346993,
      "grad_norm": 3.125,
      "learning_rate": 1.2078951702072601e-05,
      "loss": 1.2017,
      "step": 18150
    },
    {
      "epoch": 0.4144037241568162,
      "grad_norm": 3.046875,
      "learning_rate": 1.2074246594678525e-05,
      "loss": 1.2101,
      "step": 18160
    },
    {
      "epoch": 0.41463192004016247,
      "grad_norm": 2.984375,
      "learning_rate": 1.2069541487284449e-05,
      "loss": 1.1651,
      "step": 18170
    },
    {
      "epoch": 0.41486011592350874,
      "grad_norm": 3.0,
      "learning_rate": 1.2064836379890372e-05,
      "loss": 1.2008,
      "step": 18180
    },
    {
      "epoch": 0.415088311806855,
      "grad_norm": 2.96875,
      "learning_rate": 1.2060131272496296e-05,
      "loss": 1.2015,
      "step": 18190
    },
    {
      "epoch": 0.4153165076902013,
      "grad_norm": 3.015625,
      "learning_rate": 1.205542616510222e-05,
      "loss": 1.2019,
      "step": 18200
    },
    {
      "epoch": 0.41554470357354756,
      "grad_norm": 3.0,
      "learning_rate": 1.2050721057708143e-05,
      "loss": 1.2194,
      "step": 18210
    },
    {
      "epoch": 0.4157728994568938,
      "grad_norm": 2.9375,
      "learning_rate": 1.2046015950314067e-05,
      "loss": 1.1554,
      "step": 18220
    },
    {
      "epoch": 0.41600109534024005,
      "grad_norm": 3.0625,
      "learning_rate": 1.204131084291999e-05,
      "loss": 1.2021,
      "step": 18230
    },
    {
      "epoch": 0.4162292912235863,
      "grad_norm": 3.125,
      "learning_rate": 1.2036605735525915e-05,
      "loss": 1.2025,
      "step": 18240
    },
    {
      "epoch": 0.4164574871069326,
      "grad_norm": 3.28125,
      "learning_rate": 1.2031900628131838e-05,
      "loss": 1.2331,
      "step": 18250
    },
    {
      "epoch": 0.41668568299027886,
      "grad_norm": 3.0,
      "learning_rate": 1.2027195520737762e-05,
      "loss": 1.1547,
      "step": 18260
    },
    {
      "epoch": 0.41691387887362513,
      "grad_norm": 3.21875,
      "learning_rate": 1.2022490413343686e-05,
      "loss": 1.1885,
      "step": 18270
    },
    {
      "epoch": 0.4171420747569714,
      "grad_norm": 3.1875,
      "learning_rate": 1.201778530594961e-05,
      "loss": 1.1407,
      "step": 18280
    },
    {
      "epoch": 0.4173702706403177,
      "grad_norm": 2.875,
      "learning_rate": 1.2013080198555533e-05,
      "loss": 1.1895,
      "step": 18290
    },
    {
      "epoch": 0.4175984665236639,
      "grad_norm": 3.296875,
      "learning_rate": 1.2008375091161457e-05,
      "loss": 1.2464,
      "step": 18300
    },
    {
      "epoch": 0.41782666240701016,
      "grad_norm": 3.109375,
      "learning_rate": 1.200366998376738e-05,
      "loss": 1.1903,
      "step": 18310
    },
    {
      "epoch": 0.41805485829035643,
      "grad_norm": 3.1875,
      "learning_rate": 1.1998964876373304e-05,
      "loss": 1.2123,
      "step": 18320
    },
    {
      "epoch": 0.4182830541737027,
      "grad_norm": 2.890625,
      "learning_rate": 1.1994259768979228e-05,
      "loss": 1.1598,
      "step": 18330
    },
    {
      "epoch": 0.418511250057049,
      "grad_norm": 2.9375,
      "learning_rate": 1.1989554661585152e-05,
      "loss": 1.1744,
      "step": 18340
    },
    {
      "epoch": 0.41873944594039525,
      "grad_norm": 2.890625,
      "learning_rate": 1.1984849554191075e-05,
      "loss": 1.1854,
      "step": 18350
    },
    {
      "epoch": 0.4189676418237415,
      "grad_norm": 3.0,
      "learning_rate": 1.1980144446796999e-05,
      "loss": 1.2072,
      "step": 18360
    },
    {
      "epoch": 0.41919583770708774,
      "grad_norm": 3.328125,
      "learning_rate": 1.1975439339402923e-05,
      "loss": 1.2394,
      "step": 18370
    },
    {
      "epoch": 0.419424033590434,
      "grad_norm": 3.296875,
      "learning_rate": 1.1970734232008847e-05,
      "loss": 1.1943,
      "step": 18380
    },
    {
      "epoch": 0.4196522294737803,
      "grad_norm": 2.8125,
      "learning_rate": 1.196602912461477e-05,
      "loss": 1.2144,
      "step": 18390
    },
    {
      "epoch": 0.41988042535712655,
      "grad_norm": 3.234375,
      "learning_rate": 1.1961324017220694e-05,
      "loss": 1.1736,
      "step": 18400
    },
    {
      "epoch": 0.4201086212404728,
      "grad_norm": 3.078125,
      "learning_rate": 1.1956618909826618e-05,
      "loss": 1.1917,
      "step": 18410
    },
    {
      "epoch": 0.4203368171238191,
      "grad_norm": 3.203125,
      "learning_rate": 1.1951913802432541e-05,
      "loss": 1.1776,
      "step": 18420
    },
    {
      "epoch": 0.42056501300716537,
      "grad_norm": 2.890625,
      "learning_rate": 1.1947208695038465e-05,
      "loss": 1.1483,
      "step": 18430
    },
    {
      "epoch": 0.42079320889051164,
      "grad_norm": 3.09375,
      "learning_rate": 1.1942503587644389e-05,
      "loss": 1.146,
      "step": 18440
    },
    {
      "epoch": 0.42102140477385785,
      "grad_norm": 3.03125,
      "learning_rate": 1.1937798480250312e-05,
      "loss": 1.1688,
      "step": 18450
    },
    {
      "epoch": 0.4212496006572041,
      "grad_norm": 3.015625,
      "learning_rate": 1.1933093372856236e-05,
      "loss": 1.1654,
      "step": 18460
    },
    {
      "epoch": 0.4214777965405504,
      "grad_norm": 3.40625,
      "learning_rate": 1.1928388265462162e-05,
      "loss": 1.2247,
      "step": 18470
    },
    {
      "epoch": 0.42170599242389667,
      "grad_norm": 2.96875,
      "learning_rate": 1.1923683158068085e-05,
      "loss": 1.2314,
      "step": 18480
    },
    {
      "epoch": 0.42193418830724294,
      "grad_norm": 3.0,
      "learning_rate": 1.1918978050674009e-05,
      "loss": 1.2299,
      "step": 18490
    },
    {
      "epoch": 0.4221623841905892,
      "grad_norm": 3.203125,
      "learning_rate": 1.1914272943279933e-05,
      "loss": 1.2182,
      "step": 18500
    },
    {
      "epoch": 0.4223905800739355,
      "grad_norm": 2.96875,
      "learning_rate": 1.1909567835885856e-05,
      "loss": 1.1833,
      "step": 18510
    },
    {
      "epoch": 0.42261877595728176,
      "grad_norm": 3.265625,
      "learning_rate": 1.190486272849178e-05,
      "loss": 1.1457,
      "step": 18520
    },
    {
      "epoch": 0.42284697184062797,
      "grad_norm": 2.75,
      "learning_rate": 1.1900157621097704e-05,
      "loss": 1.2203,
      "step": 18530
    },
    {
      "epoch": 0.42307516772397424,
      "grad_norm": 3.5625,
      "learning_rate": 1.1895452513703628e-05,
      "loss": 1.2072,
      "step": 18540
    },
    {
      "epoch": 0.4233033636073205,
      "grad_norm": 2.96875,
      "learning_rate": 1.1890747406309551e-05,
      "loss": 1.1445,
      "step": 18550
    },
    {
      "epoch": 0.4235315594906668,
      "grad_norm": 3.046875,
      "learning_rate": 1.1886042298915475e-05,
      "loss": 1.1207,
      "step": 18560
    },
    {
      "epoch": 0.42375975537401306,
      "grad_norm": 3.34375,
      "learning_rate": 1.1881337191521399e-05,
      "loss": 1.183,
      "step": 18570
    },
    {
      "epoch": 0.42398795125735933,
      "grad_norm": 3.03125,
      "learning_rate": 1.1876632084127322e-05,
      "loss": 1.1124,
      "step": 18580
    },
    {
      "epoch": 0.4242161471407056,
      "grad_norm": 3.046875,
      "learning_rate": 1.1871926976733246e-05,
      "loss": 1.2069,
      "step": 18590
    },
    {
      "epoch": 0.4244443430240519,
      "grad_norm": 3.171875,
      "learning_rate": 1.186722186933917e-05,
      "loss": 1.219,
      "step": 18600
    },
    {
      "epoch": 0.4246725389073981,
      "grad_norm": 3.234375,
      "learning_rate": 1.1862516761945094e-05,
      "loss": 1.2535,
      "step": 18610
    },
    {
      "epoch": 0.42490073479074436,
      "grad_norm": 3.46875,
      "learning_rate": 1.1857811654551017e-05,
      "loss": 1.1989,
      "step": 18620
    },
    {
      "epoch": 0.42512893067409063,
      "grad_norm": 3.21875,
      "learning_rate": 1.1853106547156941e-05,
      "loss": 1.1824,
      "step": 18630
    },
    {
      "epoch": 0.4253571265574369,
      "grad_norm": 3.078125,
      "learning_rate": 1.1848401439762865e-05,
      "loss": 1.2194,
      "step": 18640
    },
    {
      "epoch": 0.4255853224407832,
      "grad_norm": 3.1875,
      "learning_rate": 1.1843696332368788e-05,
      "loss": 1.1714,
      "step": 18650
    },
    {
      "epoch": 0.42581351832412945,
      "grad_norm": 3.140625,
      "learning_rate": 1.1838991224974712e-05,
      "loss": 1.2262,
      "step": 18660
    },
    {
      "epoch": 0.4260417142074757,
      "grad_norm": 3.34375,
      "learning_rate": 1.1834286117580636e-05,
      "loss": 1.2267,
      "step": 18670
    },
    {
      "epoch": 0.42626991009082194,
      "grad_norm": 3.125,
      "learning_rate": 1.182958101018656e-05,
      "loss": 1.1738,
      "step": 18680
    },
    {
      "epoch": 0.4264981059741682,
      "grad_norm": 2.90625,
      "learning_rate": 1.1824875902792483e-05,
      "loss": 1.2036,
      "step": 18690
    },
    {
      "epoch": 0.4267263018575145,
      "grad_norm": 2.984375,
      "learning_rate": 1.1820170795398407e-05,
      "loss": 1.1922,
      "step": 18700
    },
    {
      "epoch": 0.42695449774086075,
      "grad_norm": 3.015625,
      "learning_rate": 1.181546568800433e-05,
      "loss": 1.2037,
      "step": 18710
    },
    {
      "epoch": 0.427182693624207,
      "grad_norm": 2.859375,
      "learning_rate": 1.1810760580610254e-05,
      "loss": 1.225,
      "step": 18720
    },
    {
      "epoch": 0.4274108895075533,
      "grad_norm": 3.40625,
      "learning_rate": 1.1806055473216178e-05,
      "loss": 1.2116,
      "step": 18730
    },
    {
      "epoch": 0.42763908539089956,
      "grad_norm": 3.046875,
      "learning_rate": 1.1801350365822102e-05,
      "loss": 1.2013,
      "step": 18740
    },
    {
      "epoch": 0.42786728127424584,
      "grad_norm": 3.046875,
      "learning_rate": 1.1796645258428026e-05,
      "loss": 1.1903,
      "step": 18750
    },
    {
      "epoch": 0.42809547715759205,
      "grad_norm": 2.953125,
      "learning_rate": 1.179194015103395e-05,
      "loss": 1.1825,
      "step": 18760
    },
    {
      "epoch": 0.4283236730409383,
      "grad_norm": 2.921875,
      "learning_rate": 1.1787235043639873e-05,
      "loss": 1.2005,
      "step": 18770
    },
    {
      "epoch": 0.4285518689242846,
      "grad_norm": 3.109375,
      "learning_rate": 1.1782529936245797e-05,
      "loss": 1.2062,
      "step": 18780
    },
    {
      "epoch": 0.42878006480763087,
      "grad_norm": 2.984375,
      "learning_rate": 1.177782482885172e-05,
      "loss": 1.2366,
      "step": 18790
    },
    {
      "epoch": 0.42900826069097714,
      "grad_norm": 2.96875,
      "learning_rate": 1.1773119721457644e-05,
      "loss": 1.186,
      "step": 18800
    },
    {
      "epoch": 0.4292364565743234,
      "grad_norm": 3.703125,
      "learning_rate": 1.1768414614063568e-05,
      "loss": 1.1374,
      "step": 18810
    },
    {
      "epoch": 0.4294646524576697,
      "grad_norm": 3.046875,
      "learning_rate": 1.1763709506669492e-05,
      "loss": 1.2339,
      "step": 18820
    },
    {
      "epoch": 0.42969284834101595,
      "grad_norm": 3.390625,
      "learning_rate": 1.1759004399275415e-05,
      "loss": 1.1762,
      "step": 18830
    },
    {
      "epoch": 0.42992104422436217,
      "grad_norm": 3.421875,
      "learning_rate": 1.1754299291881339e-05,
      "loss": 1.2382,
      "step": 18840
    },
    {
      "epoch": 0.43014924010770844,
      "grad_norm": 3.046875,
      "learning_rate": 1.1749594184487263e-05,
      "loss": 1.196,
      "step": 18850
    },
    {
      "epoch": 0.4303774359910547,
      "grad_norm": 2.890625,
      "learning_rate": 1.1744889077093186e-05,
      "loss": 1.2174,
      "step": 18860
    },
    {
      "epoch": 0.430605631874401,
      "grad_norm": 3.015625,
      "learning_rate": 1.174018396969911e-05,
      "loss": 1.1659,
      "step": 18870
    },
    {
      "epoch": 0.43083382775774726,
      "grad_norm": 2.984375,
      "learning_rate": 1.1735478862305032e-05,
      "loss": 1.2361,
      "step": 18880
    },
    {
      "epoch": 0.43106202364109353,
      "grad_norm": 2.890625,
      "learning_rate": 1.1730773754910956e-05,
      "loss": 1.2098,
      "step": 18890
    },
    {
      "epoch": 0.4312902195244398,
      "grad_norm": 3.96875,
      "learning_rate": 1.172606864751688e-05,
      "loss": 1.2084,
      "step": 18900
    },
    {
      "epoch": 0.43151841540778607,
      "grad_norm": 3.0,
      "learning_rate": 1.1721363540122803e-05,
      "loss": 1.2031,
      "step": 18910
    },
    {
      "epoch": 0.4317466112911323,
      "grad_norm": 2.9375,
      "learning_rate": 1.1716658432728727e-05,
      "loss": 1.1708,
      "step": 18920
    },
    {
      "epoch": 0.43197480717447856,
      "grad_norm": 3.171875,
      "learning_rate": 1.171195332533465e-05,
      "loss": 1.2127,
      "step": 18930
    },
    {
      "epoch": 0.43220300305782483,
      "grad_norm": 3.25,
      "learning_rate": 1.1707248217940574e-05,
      "loss": 1.1885,
      "step": 18940
    },
    {
      "epoch": 0.4324311989411711,
      "grad_norm": 2.859375,
      "learning_rate": 1.1702543110546498e-05,
      "loss": 1.2142,
      "step": 18950
    },
    {
      "epoch": 0.4326593948245174,
      "grad_norm": 2.984375,
      "learning_rate": 1.1697838003152422e-05,
      "loss": 1.17,
      "step": 18960
    },
    {
      "epoch": 0.43288759070786365,
      "grad_norm": 3.234375,
      "learning_rate": 1.1693132895758346e-05,
      "loss": 1.1718,
      "step": 18970
    },
    {
      "epoch": 0.4331157865912099,
      "grad_norm": 3.0,
      "learning_rate": 1.168842778836427e-05,
      "loss": 1.2227,
      "step": 18980
    },
    {
      "epoch": 0.43334398247455613,
      "grad_norm": 2.84375,
      "learning_rate": 1.1683722680970193e-05,
      "loss": 1.1726,
      "step": 18990
    },
    {
      "epoch": 0.4335721783579024,
      "grad_norm": 2.984375,
      "learning_rate": 1.1679017573576117e-05,
      "loss": 1.2254,
      "step": 19000
    },
    {
      "epoch": 0.4338003742412487,
      "grad_norm": 3.171875,
      "learning_rate": 1.167431246618204e-05,
      "loss": 1.2047,
      "step": 19010
    },
    {
      "epoch": 0.43402857012459495,
      "grad_norm": 3.21875,
      "learning_rate": 1.1669607358787964e-05,
      "loss": 1.2149,
      "step": 19020
    },
    {
      "epoch": 0.4342567660079412,
      "grad_norm": 3.015625,
      "learning_rate": 1.1664902251393888e-05,
      "loss": 1.1431,
      "step": 19030
    },
    {
      "epoch": 0.4344849618912875,
      "grad_norm": 3.171875,
      "learning_rate": 1.1660197143999812e-05,
      "loss": 1.1877,
      "step": 19040
    },
    {
      "epoch": 0.43471315777463376,
      "grad_norm": 3.125,
      "learning_rate": 1.1655492036605735e-05,
      "loss": 1.1537,
      "step": 19050
    },
    {
      "epoch": 0.43494135365798003,
      "grad_norm": 3.703125,
      "learning_rate": 1.1650786929211659e-05,
      "loss": 1.2254,
      "step": 19060
    },
    {
      "epoch": 0.43516954954132625,
      "grad_norm": 3.09375,
      "learning_rate": 1.1646081821817583e-05,
      "loss": 1.1734,
      "step": 19070
    },
    {
      "epoch": 0.4353977454246725,
      "grad_norm": 2.953125,
      "learning_rate": 1.1641376714423506e-05,
      "loss": 1.203,
      "step": 19080
    },
    {
      "epoch": 0.4356259413080188,
      "grad_norm": 3.140625,
      "learning_rate": 1.163667160702943e-05,
      "loss": 1.1626,
      "step": 19090
    },
    {
      "epoch": 0.43585413719136507,
      "grad_norm": 3.046875,
      "learning_rate": 1.1631966499635354e-05,
      "loss": 1.1942,
      "step": 19100
    },
    {
      "epoch": 0.43608233307471134,
      "grad_norm": 3.09375,
      "learning_rate": 1.1627261392241278e-05,
      "loss": 1.2225,
      "step": 19110
    },
    {
      "epoch": 0.4363105289580576,
      "grad_norm": 3.265625,
      "learning_rate": 1.1622556284847201e-05,
      "loss": 1.192,
      "step": 19120
    },
    {
      "epoch": 0.4365387248414039,
      "grad_norm": 2.796875,
      "learning_rate": 1.1617851177453125e-05,
      "loss": 1.1856,
      "step": 19130
    },
    {
      "epoch": 0.43676692072475015,
      "grad_norm": 3.1875,
      "learning_rate": 1.1613146070059049e-05,
      "loss": 1.1835,
      "step": 19140
    },
    {
      "epoch": 0.43699511660809637,
      "grad_norm": 3.1875,
      "learning_rate": 1.1608440962664972e-05,
      "loss": 1.1669,
      "step": 19150
    },
    {
      "epoch": 0.43722331249144264,
      "grad_norm": 3.203125,
      "learning_rate": 1.1603735855270896e-05,
      "loss": 1.2184,
      "step": 19160
    },
    {
      "epoch": 0.4374515083747889,
      "grad_norm": 3.03125,
      "learning_rate": 1.159903074787682e-05,
      "loss": 1.1657,
      "step": 19170
    },
    {
      "epoch": 0.4376797042581352,
      "grad_norm": 2.84375,
      "learning_rate": 1.1594325640482743e-05,
      "loss": 1.1671,
      "step": 19180
    },
    {
      "epoch": 0.43790790014148145,
      "grad_norm": 2.984375,
      "learning_rate": 1.1589620533088667e-05,
      "loss": 1.2467,
      "step": 19190
    },
    {
      "epoch": 0.4381360960248277,
      "grad_norm": 2.734375,
      "learning_rate": 1.1584915425694591e-05,
      "loss": 1.1184,
      "step": 19200
    },
    {
      "epoch": 0.438364291908174,
      "grad_norm": 2.90625,
      "learning_rate": 1.1580210318300515e-05,
      "loss": 1.2081,
      "step": 19210
    },
    {
      "epoch": 0.4385924877915202,
      "grad_norm": 3.703125,
      "learning_rate": 1.1575505210906438e-05,
      "loss": 1.1613,
      "step": 19220
    },
    {
      "epoch": 0.4388206836748665,
      "grad_norm": 3.375,
      "learning_rate": 1.1570800103512364e-05,
      "loss": 1.186,
      "step": 19230
    },
    {
      "epoch": 0.43904887955821276,
      "grad_norm": 3.625,
      "learning_rate": 1.1566094996118287e-05,
      "loss": 1.1795,
      "step": 19240
    },
    {
      "epoch": 0.43927707544155903,
      "grad_norm": 3.15625,
      "learning_rate": 1.1561389888724211e-05,
      "loss": 1.2347,
      "step": 19250
    },
    {
      "epoch": 0.4395052713249053,
      "grad_norm": 3.046875,
      "learning_rate": 1.1556684781330135e-05,
      "loss": 1.2002,
      "step": 19260
    },
    {
      "epoch": 0.43973346720825157,
      "grad_norm": 3.40625,
      "learning_rate": 1.1551979673936059e-05,
      "loss": 1.2053,
      "step": 19270
    },
    {
      "epoch": 0.43996166309159784,
      "grad_norm": 3.28125,
      "learning_rate": 1.1547274566541982e-05,
      "loss": 1.1783,
      "step": 19280
    },
    {
      "epoch": 0.4401898589749441,
      "grad_norm": 3.078125,
      "learning_rate": 1.1542569459147906e-05,
      "loss": 1.2048,
      "step": 19290
    },
    {
      "epoch": 0.44041805485829033,
      "grad_norm": 2.984375,
      "learning_rate": 1.153786435175383e-05,
      "loss": 1.233,
      "step": 19300
    },
    {
      "epoch": 0.4406462507416366,
      "grad_norm": 2.96875,
      "learning_rate": 1.1533159244359753e-05,
      "loss": 1.2006,
      "step": 19310
    },
    {
      "epoch": 0.4408744466249829,
      "grad_norm": 2.9375,
      "learning_rate": 1.1528454136965677e-05,
      "loss": 1.1704,
      "step": 19320
    },
    {
      "epoch": 0.44110264250832915,
      "grad_norm": 3.0625,
      "learning_rate": 1.1523749029571601e-05,
      "loss": 1.1712,
      "step": 19330
    },
    {
      "epoch": 0.4413308383916754,
      "grad_norm": 3.328125,
      "learning_rate": 1.1519043922177525e-05,
      "loss": 1.208,
      "step": 19340
    },
    {
      "epoch": 0.4415590342750217,
      "grad_norm": 2.96875,
      "learning_rate": 1.1514338814783448e-05,
      "loss": 1.2004,
      "step": 19350
    },
    {
      "epoch": 0.44178723015836796,
      "grad_norm": 3.140625,
      "learning_rate": 1.1509633707389372e-05,
      "loss": 1.1869,
      "step": 19360
    },
    {
      "epoch": 0.44201542604171423,
      "grad_norm": 2.828125,
      "learning_rate": 1.1504928599995296e-05,
      "loss": 1.2113,
      "step": 19370
    },
    {
      "epoch": 0.44224362192506045,
      "grad_norm": 3.015625,
      "learning_rate": 1.150022349260122e-05,
      "loss": 1.1501,
      "step": 19380
    },
    {
      "epoch": 0.4424718178084067,
      "grad_norm": 3.125,
      "learning_rate": 1.1495518385207143e-05,
      "loss": 1.1711,
      "step": 19390
    },
    {
      "epoch": 0.442700013691753,
      "grad_norm": 3.1875,
      "learning_rate": 1.1490813277813067e-05,
      "loss": 1.2138,
      "step": 19400
    },
    {
      "epoch": 0.44292820957509926,
      "grad_norm": 3.15625,
      "learning_rate": 1.148610817041899e-05,
      "loss": 1.2578,
      "step": 19410
    },
    {
      "epoch": 0.44315640545844553,
      "grad_norm": 3.015625,
      "learning_rate": 1.1481403063024914e-05,
      "loss": 1.1842,
      "step": 19420
    },
    {
      "epoch": 0.4433846013417918,
      "grad_norm": 3.140625,
      "learning_rate": 1.1476697955630838e-05,
      "loss": 1.2109,
      "step": 19430
    },
    {
      "epoch": 0.4436127972251381,
      "grad_norm": 3.109375,
      "learning_rate": 1.1471992848236762e-05,
      "loss": 1.1795,
      "step": 19440
    },
    {
      "epoch": 0.44384099310848435,
      "grad_norm": 2.828125,
      "learning_rate": 1.1467287740842685e-05,
      "loss": 1.2428,
      "step": 19450
    },
    {
      "epoch": 0.44406918899183057,
      "grad_norm": 3.109375,
      "learning_rate": 1.1462582633448609e-05,
      "loss": 1.1561,
      "step": 19460
    },
    {
      "epoch": 0.44429738487517684,
      "grad_norm": 3.328125,
      "learning_rate": 1.1457877526054533e-05,
      "loss": 1.2757,
      "step": 19470
    },
    {
      "epoch": 0.4445255807585231,
      "grad_norm": 3.203125,
      "learning_rate": 1.1453172418660457e-05,
      "loss": 1.2242,
      "step": 19480
    },
    {
      "epoch": 0.4447537766418694,
      "grad_norm": 3.03125,
      "learning_rate": 1.144846731126638e-05,
      "loss": 1.2187,
      "step": 19490
    },
    {
      "epoch": 0.44498197252521565,
      "grad_norm": 3.03125,
      "learning_rate": 1.1443762203872304e-05,
      "loss": 1.209,
      "step": 19500
    },
    {
      "epoch": 0.4452101684085619,
      "grad_norm": 3.078125,
      "learning_rate": 1.1439057096478228e-05,
      "loss": 1.1854,
      "step": 19510
    },
    {
      "epoch": 0.4454383642919082,
      "grad_norm": 3.1875,
      "learning_rate": 1.1434351989084151e-05,
      "loss": 1.2314,
      "step": 19520
    },
    {
      "epoch": 0.4456665601752544,
      "grad_norm": 3.796875,
      "learning_rate": 1.1429646881690075e-05,
      "loss": 1.2632,
      "step": 19530
    },
    {
      "epoch": 0.4458947560586007,
      "grad_norm": 3.15625,
      "learning_rate": 1.1424941774295999e-05,
      "loss": 1.2216,
      "step": 19540
    },
    {
      "epoch": 0.44612295194194695,
      "grad_norm": 3.0625,
      "learning_rate": 1.1420236666901923e-05,
      "loss": 1.2109,
      "step": 19550
    },
    {
      "epoch": 0.4463511478252932,
      "grad_norm": 3.171875,
      "learning_rate": 1.1415531559507846e-05,
      "loss": 1.2006,
      "step": 19560
    },
    {
      "epoch": 0.4465793437086395,
      "grad_norm": 2.875,
      "learning_rate": 1.141082645211377e-05,
      "loss": 1.1739,
      "step": 19570
    },
    {
      "epoch": 0.44680753959198577,
      "grad_norm": 3.203125,
      "learning_rate": 1.1406121344719694e-05,
      "loss": 1.2249,
      "step": 19580
    },
    {
      "epoch": 0.44703573547533204,
      "grad_norm": 2.90625,
      "learning_rate": 1.1401416237325617e-05,
      "loss": 1.1932,
      "step": 19590
    },
    {
      "epoch": 0.4472639313586783,
      "grad_norm": 3.09375,
      "learning_rate": 1.1396711129931541e-05,
      "loss": 1.1845,
      "step": 19600
    },
    {
      "epoch": 0.44749212724202453,
      "grad_norm": 3.46875,
      "learning_rate": 1.1392006022537465e-05,
      "loss": 1.1927,
      "step": 19610
    },
    {
      "epoch": 0.4477203231253708,
      "grad_norm": 3.234375,
      "learning_rate": 1.1387300915143389e-05,
      "loss": 1.2246,
      "step": 19620
    },
    {
      "epoch": 0.44794851900871707,
      "grad_norm": 3.40625,
      "learning_rate": 1.1382595807749312e-05,
      "loss": 1.1443,
      "step": 19630
    },
    {
      "epoch": 0.44817671489206334,
      "grad_norm": 3.078125,
      "learning_rate": 1.1377890700355236e-05,
      "loss": 1.1713,
      "step": 19640
    },
    {
      "epoch": 0.4484049107754096,
      "grad_norm": 3.09375,
      "learning_rate": 1.137318559296116e-05,
      "loss": 1.2166,
      "step": 19650
    },
    {
      "epoch": 0.4486331066587559,
      "grad_norm": 2.984375,
      "learning_rate": 1.1368480485567083e-05,
      "loss": 1.1685,
      "step": 19660
    },
    {
      "epoch": 0.44886130254210216,
      "grad_norm": 2.859375,
      "learning_rate": 1.1363775378173007e-05,
      "loss": 1.1804,
      "step": 19670
    },
    {
      "epoch": 0.44908949842544843,
      "grad_norm": 3.15625,
      "learning_rate": 1.135907027077893e-05,
      "loss": 1.1702,
      "step": 19680
    },
    {
      "epoch": 0.44931769430879465,
      "grad_norm": 3.0625,
      "learning_rate": 1.1354365163384855e-05,
      "loss": 1.18,
      "step": 19690
    },
    {
      "epoch": 0.4495458901921409,
      "grad_norm": 3.328125,
      "learning_rate": 1.1349660055990778e-05,
      "loss": 1.2117,
      "step": 19700
    },
    {
      "epoch": 0.4497740860754872,
      "grad_norm": 3.25,
      "learning_rate": 1.1344954948596702e-05,
      "loss": 1.164,
      "step": 19710
    },
    {
      "epoch": 0.45000228195883346,
      "grad_norm": 3.28125,
      "learning_rate": 1.1340249841202626e-05,
      "loss": 1.2209,
      "step": 19720
    },
    {
      "epoch": 0.45023047784217973,
      "grad_norm": 3.484375,
      "learning_rate": 1.133554473380855e-05,
      "loss": 1.1876,
      "step": 19730
    },
    {
      "epoch": 0.450458673725526,
      "grad_norm": 3.328125,
      "learning_rate": 1.1330839626414473e-05,
      "loss": 1.1777,
      "step": 19740
    },
    {
      "epoch": 0.4506868696088723,
      "grad_norm": 3.03125,
      "learning_rate": 1.1326134519020397e-05,
      "loss": 1.1754,
      "step": 19750
    },
    {
      "epoch": 0.45091506549221855,
      "grad_norm": 3.09375,
      "learning_rate": 1.132142941162632e-05,
      "loss": 1.1656,
      "step": 19760
    },
    {
      "epoch": 0.45114326137556476,
      "grad_norm": 3.046875,
      "learning_rate": 1.1316724304232244e-05,
      "loss": 1.1942,
      "step": 19770
    },
    {
      "epoch": 0.45137145725891104,
      "grad_norm": 3.09375,
      "learning_rate": 1.1312019196838168e-05,
      "loss": 1.1456,
      "step": 19780
    },
    {
      "epoch": 0.4515996531422573,
      "grad_norm": 2.828125,
      "learning_rate": 1.1307314089444092e-05,
      "loss": 1.2239,
      "step": 19790
    },
    {
      "epoch": 0.4518278490256036,
      "grad_norm": 2.9375,
      "learning_rate": 1.1302608982050017e-05,
      "loss": 1.2019,
      "step": 19800
    },
    {
      "epoch": 0.45205604490894985,
      "grad_norm": 3.234375,
      "learning_rate": 1.129790387465594e-05,
      "loss": 1.1859,
      "step": 19810
    },
    {
      "epoch": 0.4522842407922961,
      "grad_norm": 3.046875,
      "learning_rate": 1.1293198767261864e-05,
      "loss": 1.157,
      "step": 19820
    },
    {
      "epoch": 0.4525124366756424,
      "grad_norm": 3.109375,
      "learning_rate": 1.1288493659867788e-05,
      "loss": 1.2025,
      "step": 19830
    },
    {
      "epoch": 0.4527406325589886,
      "grad_norm": 2.90625,
      "learning_rate": 1.1283788552473712e-05,
      "loss": 1.1781,
      "step": 19840
    },
    {
      "epoch": 0.4529688284423349,
      "grad_norm": 2.921875,
      "learning_rate": 1.1279083445079636e-05,
      "loss": 1.2091,
      "step": 19850
    },
    {
      "epoch": 0.45319702432568115,
      "grad_norm": 3.390625,
      "learning_rate": 1.127437833768556e-05,
      "loss": 1.201,
      "step": 19860
    },
    {
      "epoch": 0.4534252202090274,
      "grad_norm": 3.171875,
      "learning_rate": 1.1269673230291483e-05,
      "loss": 1.2125,
      "step": 19870
    },
    {
      "epoch": 0.4536534160923737,
      "grad_norm": 3.234375,
      "learning_rate": 1.1264968122897407e-05,
      "loss": 1.2361,
      "step": 19880
    },
    {
      "epoch": 0.45388161197571997,
      "grad_norm": 3.140625,
      "learning_rate": 1.126026301550333e-05,
      "loss": 1.1994,
      "step": 19890
    },
    {
      "epoch": 0.45410980785906624,
      "grad_norm": 2.78125,
      "learning_rate": 1.1255557908109254e-05,
      "loss": 1.2358,
      "step": 19900
    },
    {
      "epoch": 0.4543380037424125,
      "grad_norm": 3.375,
      "learning_rate": 1.1250852800715178e-05,
      "loss": 1.2054,
      "step": 19910
    },
    {
      "epoch": 0.4545661996257587,
      "grad_norm": 3.0,
      "learning_rate": 1.1246147693321102e-05,
      "loss": 1.1829,
      "step": 19920
    },
    {
      "epoch": 0.454794395509105,
      "grad_norm": 3.0625,
      "learning_rate": 1.1241442585927025e-05,
      "loss": 1.179,
      "step": 19930
    },
    {
      "epoch": 0.45502259139245127,
      "grad_norm": 3.109375,
      "learning_rate": 1.1236737478532949e-05,
      "loss": 1.179,
      "step": 19940
    },
    {
      "epoch": 0.45525078727579754,
      "grad_norm": 3.03125,
      "learning_rate": 1.1232032371138873e-05,
      "loss": 1.1881,
      "step": 19950
    },
    {
      "epoch": 0.4554789831591438,
      "grad_norm": 3.140625,
      "learning_rate": 1.1227327263744796e-05,
      "loss": 1.1692,
      "step": 19960
    },
    {
      "epoch": 0.4557071790424901,
      "grad_norm": 3.09375,
      "learning_rate": 1.122262215635072e-05,
      "loss": 1.1412,
      "step": 19970
    },
    {
      "epoch": 0.45593537492583636,
      "grad_norm": 3.15625,
      "learning_rate": 1.1217917048956644e-05,
      "loss": 1.1652,
      "step": 19980
    },
    {
      "epoch": 0.45616357080918263,
      "grad_norm": 3.171875,
      "learning_rate": 1.1213211941562568e-05,
      "loss": 1.127,
      "step": 19990
    },
    {
      "epoch": 0.45639176669252884,
      "grad_norm": 3.296875,
      "learning_rate": 1.1208506834168491e-05,
      "loss": 1.2107,
      "step": 20000
    },
    {
      "epoch": 0.4566199625758751,
      "grad_norm": 3.03125,
      "learning_rate": 1.1203801726774415e-05,
      "loss": 1.1498,
      "step": 20010
    },
    {
      "epoch": 0.4568481584592214,
      "grad_norm": 2.9375,
      "learning_rate": 1.1199096619380339e-05,
      "loss": 1.2078,
      "step": 20020
    },
    {
      "epoch": 0.45707635434256766,
      "grad_norm": 2.9375,
      "learning_rate": 1.1194391511986262e-05,
      "loss": 1.175,
      "step": 20030
    },
    {
      "epoch": 0.45730455022591393,
      "grad_norm": 3.390625,
      "learning_rate": 1.1189686404592186e-05,
      "loss": 1.2801,
      "step": 20040
    },
    {
      "epoch": 0.4575327461092602,
      "grad_norm": 3.234375,
      "learning_rate": 1.118498129719811e-05,
      "loss": 1.2289,
      "step": 20050
    },
    {
      "epoch": 0.4577609419926065,
      "grad_norm": 3.203125,
      "learning_rate": 1.1180276189804034e-05,
      "loss": 1.179,
      "step": 20060
    },
    {
      "epoch": 0.45798913787595275,
      "grad_norm": 3.109375,
      "learning_rate": 1.1175571082409957e-05,
      "loss": 1.2351,
      "step": 20070
    },
    {
      "epoch": 0.45821733375929896,
      "grad_norm": 3.046875,
      "learning_rate": 1.1170865975015881e-05,
      "loss": 1.1801,
      "step": 20080
    },
    {
      "epoch": 0.45844552964264523,
      "grad_norm": 3.234375,
      "learning_rate": 1.1166160867621805e-05,
      "loss": 1.1623,
      "step": 20090
    },
    {
      "epoch": 0.4586737255259915,
      "grad_norm": 2.953125,
      "learning_rate": 1.1161455760227728e-05,
      "loss": 1.1873,
      "step": 20100
    },
    {
      "epoch": 0.4589019214093378,
      "grad_norm": 3.25,
      "learning_rate": 1.1156750652833652e-05,
      "loss": 1.1697,
      "step": 20110
    },
    {
      "epoch": 0.45913011729268405,
      "grad_norm": 3.015625,
      "learning_rate": 1.1152045545439576e-05,
      "loss": 1.2319,
      "step": 20120
    },
    {
      "epoch": 0.4593583131760303,
      "grad_norm": 3.0625,
      "learning_rate": 1.11473404380455e-05,
      "loss": 1.1543,
      "step": 20130
    },
    {
      "epoch": 0.4595865090593766,
      "grad_norm": 3.3125,
      "learning_rate": 1.1142635330651423e-05,
      "loss": 1.2516,
      "step": 20140
    },
    {
      "epoch": 0.4598147049427228,
      "grad_norm": 2.875,
      "learning_rate": 1.1137930223257347e-05,
      "loss": 1.1923,
      "step": 20150
    },
    {
      "epoch": 0.4600429008260691,
      "grad_norm": 3.15625,
      "learning_rate": 1.113322511586327e-05,
      "loss": 1.154,
      "step": 20160
    },
    {
      "epoch": 0.46027109670941535,
      "grad_norm": 3.296875,
      "learning_rate": 1.1128520008469194e-05,
      "loss": 1.1847,
      "step": 20170
    },
    {
      "epoch": 0.4604992925927616,
      "grad_norm": 2.921875,
      "learning_rate": 1.1123814901075118e-05,
      "loss": 1.2058,
      "step": 20180
    },
    {
      "epoch": 0.4607274884761079,
      "grad_norm": 3.390625,
      "learning_rate": 1.1119109793681042e-05,
      "loss": 1.1677,
      "step": 20190
    },
    {
      "epoch": 0.46095568435945417,
      "grad_norm": 3.140625,
      "learning_rate": 1.1114404686286966e-05,
      "loss": 1.211,
      "step": 20200
    },
    {
      "epoch": 0.46118388024280044,
      "grad_norm": 2.859375,
      "learning_rate": 1.110969957889289e-05,
      "loss": 1.1993,
      "step": 20210
    },
    {
      "epoch": 0.4614120761261467,
      "grad_norm": 3.375,
      "learning_rate": 1.1104994471498813e-05,
      "loss": 1.1172,
      "step": 20220
    },
    {
      "epoch": 0.4616402720094929,
      "grad_norm": 3.28125,
      "learning_rate": 1.1100289364104737e-05,
      "loss": 1.2414,
      "step": 20230
    },
    {
      "epoch": 0.4618684678928392,
      "grad_norm": 3.265625,
      "learning_rate": 1.109558425671066e-05,
      "loss": 1.251,
      "step": 20240
    },
    {
      "epoch": 0.46209666377618547,
      "grad_norm": 3.078125,
      "learning_rate": 1.1090879149316584e-05,
      "loss": 1.1799,
      "step": 20250
    },
    {
      "epoch": 0.46232485965953174,
      "grad_norm": 2.859375,
      "learning_rate": 1.1086174041922508e-05,
      "loss": 1.1637,
      "step": 20260
    },
    {
      "epoch": 0.462553055542878,
      "grad_norm": 3.65625,
      "learning_rate": 1.1081468934528432e-05,
      "loss": 1.1311,
      "step": 20270
    },
    {
      "epoch": 0.4627812514262243,
      "grad_norm": 3.0,
      "learning_rate": 1.1076763827134355e-05,
      "loss": 1.2163,
      "step": 20280
    },
    {
      "epoch": 0.46300944730957055,
      "grad_norm": 2.84375,
      "learning_rate": 1.1072058719740279e-05,
      "loss": 1.2106,
      "step": 20290
    },
    {
      "epoch": 0.4632376431929168,
      "grad_norm": 2.9375,
      "learning_rate": 1.1067353612346203e-05,
      "loss": 1.1607,
      "step": 20300
    },
    {
      "epoch": 0.46346583907626304,
      "grad_norm": 3.484375,
      "learning_rate": 1.1062648504952126e-05,
      "loss": 1.2217,
      "step": 20310
    },
    {
      "epoch": 0.4636940349596093,
      "grad_norm": 2.984375,
      "learning_rate": 1.105794339755805e-05,
      "loss": 1.1702,
      "step": 20320
    },
    {
      "epoch": 0.4639222308429556,
      "grad_norm": 2.953125,
      "learning_rate": 1.1053238290163974e-05,
      "loss": 1.1422,
      "step": 20330
    },
    {
      "epoch": 0.46415042672630186,
      "grad_norm": 2.96875,
      "learning_rate": 1.1048533182769898e-05,
      "loss": 1.1669,
      "step": 20340
    },
    {
      "epoch": 0.46437862260964813,
      "grad_norm": 3.171875,
      "learning_rate": 1.1043828075375821e-05,
      "loss": 1.2771,
      "step": 20350
    },
    {
      "epoch": 0.4646068184929944,
      "grad_norm": 3.0,
      "learning_rate": 1.1039122967981745e-05,
      "loss": 1.1827,
      "step": 20360
    },
    {
      "epoch": 0.46483501437634067,
      "grad_norm": 3.21875,
      "learning_rate": 1.1034417860587669e-05,
      "loss": 1.2069,
      "step": 20370
    },
    {
      "epoch": 0.4650632102596869,
      "grad_norm": 3.15625,
      "learning_rate": 1.1029712753193594e-05,
      "loss": 1.2365,
      "step": 20380
    },
    {
      "epoch": 0.46529140614303316,
      "grad_norm": 3.296875,
      "learning_rate": 1.1025007645799518e-05,
      "loss": 1.1643,
      "step": 20390
    },
    {
      "epoch": 0.46551960202637943,
      "grad_norm": 3.40625,
      "learning_rate": 1.1020302538405441e-05,
      "loss": 1.2117,
      "step": 20400
    },
    {
      "epoch": 0.4657477979097257,
      "grad_norm": 3.171875,
      "learning_rate": 1.1015597431011365e-05,
      "loss": 1.2296,
      "step": 20410
    },
    {
      "epoch": 0.465975993793072,
      "grad_norm": 2.859375,
      "learning_rate": 1.1010892323617289e-05,
      "loss": 1.1658,
      "step": 20420
    },
    {
      "epoch": 0.46620418967641825,
      "grad_norm": 3.09375,
      "learning_rate": 1.1006187216223213e-05,
      "loss": 1.1877,
      "step": 20430
    },
    {
      "epoch": 0.4664323855597645,
      "grad_norm": 2.953125,
      "learning_rate": 1.1001482108829136e-05,
      "loss": 1.1795,
      "step": 20440
    },
    {
      "epoch": 0.4666605814431108,
      "grad_norm": 3.203125,
      "learning_rate": 1.099677700143506e-05,
      "loss": 1.2146,
      "step": 20450
    },
    {
      "epoch": 0.466888777326457,
      "grad_norm": 3.3125,
      "learning_rate": 1.0992071894040984e-05,
      "loss": 1.226,
      "step": 20460
    },
    {
      "epoch": 0.4671169732098033,
      "grad_norm": 3.25,
      "learning_rate": 1.0987366786646907e-05,
      "loss": 1.1867,
      "step": 20470
    },
    {
      "epoch": 0.46734516909314955,
      "grad_norm": 3.296875,
      "learning_rate": 1.0982661679252831e-05,
      "loss": 1.1652,
      "step": 20480
    },
    {
      "epoch": 0.4675733649764958,
      "grad_norm": 3.0,
      "learning_rate": 1.0977956571858755e-05,
      "loss": 1.2671,
      "step": 20490
    },
    {
      "epoch": 0.4678015608598421,
      "grad_norm": 3.171875,
      "learning_rate": 1.0973251464464679e-05,
      "loss": 1.206,
      "step": 20500
    },
    {
      "epoch": 0.46802975674318836,
      "grad_norm": 3.09375,
      "learning_rate": 1.0968546357070602e-05,
      "loss": 1.1853,
      "step": 20510
    },
    {
      "epoch": 0.46825795262653463,
      "grad_norm": 3.125,
      "learning_rate": 1.0963841249676526e-05,
      "loss": 1.2295,
      "step": 20520
    },
    {
      "epoch": 0.4684861485098809,
      "grad_norm": 3.171875,
      "learning_rate": 1.095913614228245e-05,
      "loss": 1.1413,
      "step": 20530
    },
    {
      "epoch": 0.4687143443932271,
      "grad_norm": 3.140625,
      "learning_rate": 1.0954431034888373e-05,
      "loss": 1.2104,
      "step": 20540
    },
    {
      "epoch": 0.4689425402765734,
      "grad_norm": 3.046875,
      "learning_rate": 1.0949725927494297e-05,
      "loss": 1.2009,
      "step": 20550
    },
    {
      "epoch": 0.46917073615991967,
      "grad_norm": 3.109375,
      "learning_rate": 1.0945020820100221e-05,
      "loss": 1.1869,
      "step": 20560
    },
    {
      "epoch": 0.46939893204326594,
      "grad_norm": 3.34375,
      "learning_rate": 1.0940315712706145e-05,
      "loss": 1.1954,
      "step": 20570
    },
    {
      "epoch": 0.4696271279266122,
      "grad_norm": 3.140625,
      "learning_rate": 1.0935610605312068e-05,
      "loss": 1.1961,
      "step": 20580
    },
    {
      "epoch": 0.4698553238099585,
      "grad_norm": 3.359375,
      "learning_rate": 1.0930905497917992e-05,
      "loss": 1.2336,
      "step": 20590
    },
    {
      "epoch": 0.47008351969330475,
      "grad_norm": 2.984375,
      "learning_rate": 1.0926200390523916e-05,
      "loss": 1.1844,
      "step": 20600
    },
    {
      "epoch": 0.470311715576651,
      "grad_norm": 2.90625,
      "learning_rate": 1.092149528312984e-05,
      "loss": 1.1732,
      "step": 20610
    },
    {
      "epoch": 0.47053991145999724,
      "grad_norm": 3.28125,
      "learning_rate": 1.0916790175735763e-05,
      "loss": 1.1617,
      "step": 20620
    },
    {
      "epoch": 0.4707681073433435,
      "grad_norm": 3.375,
      "learning_rate": 1.0912085068341687e-05,
      "loss": 1.2161,
      "step": 20630
    },
    {
      "epoch": 0.4709963032266898,
      "grad_norm": 3.0,
      "learning_rate": 1.090737996094761e-05,
      "loss": 1.1784,
      "step": 20640
    },
    {
      "epoch": 0.47122449911003605,
      "grad_norm": 3.234375,
      "learning_rate": 1.0902674853553534e-05,
      "loss": 1.2284,
      "step": 20650
    },
    {
      "epoch": 0.4714526949933823,
      "grad_norm": 3.21875,
      "learning_rate": 1.0897969746159458e-05,
      "loss": 1.1903,
      "step": 20660
    },
    {
      "epoch": 0.4716808908767286,
      "grad_norm": 3.6875,
      "learning_rate": 1.0893264638765382e-05,
      "loss": 1.1779,
      "step": 20670
    },
    {
      "epoch": 0.47190908676007487,
      "grad_norm": 3.328125,
      "learning_rate": 1.0888559531371305e-05,
      "loss": 1.1971,
      "step": 20680
    },
    {
      "epoch": 0.4721372826434211,
      "grad_norm": 3.015625,
      "learning_rate": 1.088385442397723e-05,
      "loss": 1.1631,
      "step": 20690
    },
    {
      "epoch": 0.47236547852676736,
      "grad_norm": 3.015625,
      "learning_rate": 1.0879149316583153e-05,
      "loss": 1.1239,
      "step": 20700
    },
    {
      "epoch": 0.47259367441011363,
      "grad_norm": 3.21875,
      "learning_rate": 1.0874444209189077e-05,
      "loss": 1.1815,
      "step": 20710
    },
    {
      "epoch": 0.4728218702934599,
      "grad_norm": 2.90625,
      "learning_rate": 1.0869739101795e-05,
      "loss": 1.2071,
      "step": 20720
    },
    {
      "epoch": 0.47305006617680617,
      "grad_norm": 3.34375,
      "learning_rate": 1.0865033994400922e-05,
      "loss": 1.1838,
      "step": 20730
    },
    {
      "epoch": 0.47327826206015244,
      "grad_norm": 3.28125,
      "learning_rate": 1.0860328887006846e-05,
      "loss": 1.1167,
      "step": 20740
    },
    {
      "epoch": 0.4735064579434987,
      "grad_norm": 3.03125,
      "learning_rate": 1.085562377961277e-05,
      "loss": 1.1515,
      "step": 20750
    },
    {
      "epoch": 0.473734653826845,
      "grad_norm": 3.078125,
      "learning_rate": 1.0850918672218693e-05,
      "loss": 1.1787,
      "step": 20760
    },
    {
      "epoch": 0.4739628497101912,
      "grad_norm": 3.0,
      "learning_rate": 1.0846213564824617e-05,
      "loss": 1.173,
      "step": 20770
    },
    {
      "epoch": 0.4741910455935375,
      "grad_norm": 3.140625,
      "learning_rate": 1.0841508457430541e-05,
      "loss": 1.206,
      "step": 20780
    },
    {
      "epoch": 0.47441924147688375,
      "grad_norm": 3.40625,
      "learning_rate": 1.0836803350036465e-05,
      "loss": 1.1953,
      "step": 20790
    },
    {
      "epoch": 0.47464743736023,
      "grad_norm": 2.859375,
      "learning_rate": 1.0832098242642388e-05,
      "loss": 1.1485,
      "step": 20800
    },
    {
      "epoch": 0.4748756332435763,
      "grad_norm": 3.1875,
      "learning_rate": 1.0827393135248312e-05,
      "loss": 1.1831,
      "step": 20810
    },
    {
      "epoch": 0.47510382912692256,
      "grad_norm": 3.640625,
      "learning_rate": 1.0822688027854236e-05,
      "loss": 1.1935,
      "step": 20820
    },
    {
      "epoch": 0.47533202501026883,
      "grad_norm": 3.203125,
      "learning_rate": 1.081798292046016e-05,
      "loss": 1.1049,
      "step": 20830
    },
    {
      "epoch": 0.4755602208936151,
      "grad_norm": 3.140625,
      "learning_rate": 1.0813277813066083e-05,
      "loss": 1.1454,
      "step": 20840
    },
    {
      "epoch": 0.4757884167769613,
      "grad_norm": 3.0625,
      "learning_rate": 1.0808572705672007e-05,
      "loss": 1.1724,
      "step": 20850
    },
    {
      "epoch": 0.4760166126603076,
      "grad_norm": 2.9375,
      "learning_rate": 1.080386759827793e-05,
      "loss": 1.2098,
      "step": 20860
    },
    {
      "epoch": 0.47624480854365386,
      "grad_norm": 2.9375,
      "learning_rate": 1.0799162490883854e-05,
      "loss": 1.1882,
      "step": 20870
    },
    {
      "epoch": 0.47647300442700014,
      "grad_norm": 2.921875,
      "learning_rate": 1.0794457383489778e-05,
      "loss": 1.1985,
      "step": 20880
    },
    {
      "epoch": 0.4767012003103464,
      "grad_norm": 3.375,
      "learning_rate": 1.0789752276095702e-05,
      "loss": 1.2099,
      "step": 20890
    },
    {
      "epoch": 0.4769293961936927,
      "grad_norm": 3.109375,
      "learning_rate": 1.0785047168701625e-05,
      "loss": 1.2056,
      "step": 20900
    },
    {
      "epoch": 0.47715759207703895,
      "grad_norm": 2.96875,
      "learning_rate": 1.0780342061307549e-05,
      "loss": 1.2291,
      "step": 20910
    },
    {
      "epoch": 0.4773857879603852,
      "grad_norm": 3.09375,
      "learning_rate": 1.0775636953913473e-05,
      "loss": 1.1504,
      "step": 20920
    },
    {
      "epoch": 0.47761398384373144,
      "grad_norm": 3.390625,
      "learning_rate": 1.0770931846519397e-05,
      "loss": 1.247,
      "step": 20930
    },
    {
      "epoch": 0.4778421797270777,
      "grad_norm": 3.125,
      "learning_rate": 1.076622673912532e-05,
      "loss": 1.1403,
      "step": 20940
    },
    {
      "epoch": 0.478070375610424,
      "grad_norm": 3.359375,
      "learning_rate": 1.0761521631731244e-05,
      "loss": 1.2189,
      "step": 20950
    },
    {
      "epoch": 0.47829857149377025,
      "grad_norm": 3.015625,
      "learning_rate": 1.0756816524337168e-05,
      "loss": 1.1905,
      "step": 20960
    },
    {
      "epoch": 0.4785267673771165,
      "grad_norm": 2.953125,
      "learning_rate": 1.0752111416943091e-05,
      "loss": 1.148,
      "step": 20970
    },
    {
      "epoch": 0.4787549632604628,
      "grad_norm": 3.109375,
      "learning_rate": 1.0747406309549015e-05,
      "loss": 1.2455,
      "step": 20980
    },
    {
      "epoch": 0.47898315914380907,
      "grad_norm": 2.96875,
      "learning_rate": 1.0742701202154939e-05,
      "loss": 1.2109,
      "step": 20990
    },
    {
      "epoch": 0.4792113550271553,
      "grad_norm": 3.140625,
      "learning_rate": 1.0737996094760863e-05,
      "loss": 1.2105,
      "step": 21000
    },
    {
      "epoch": 0.47943955091050156,
      "grad_norm": 2.984375,
      "learning_rate": 1.0733290987366786e-05,
      "loss": 1.2342,
      "step": 21010
    },
    {
      "epoch": 0.4796677467938478,
      "grad_norm": 3.359375,
      "learning_rate": 1.072858587997271e-05,
      "loss": 1.2238,
      "step": 21020
    },
    {
      "epoch": 0.4798959426771941,
      "grad_norm": 3.375,
      "learning_rate": 1.0723880772578634e-05,
      "loss": 1.1651,
      "step": 21030
    },
    {
      "epoch": 0.48012413856054037,
      "grad_norm": 2.96875,
      "learning_rate": 1.0719175665184557e-05,
      "loss": 1.176,
      "step": 21040
    },
    {
      "epoch": 0.48035233444388664,
      "grad_norm": 3.1875,
      "learning_rate": 1.0714470557790481e-05,
      "loss": 1.1214,
      "step": 21050
    },
    {
      "epoch": 0.4805805303272329,
      "grad_norm": 3.046875,
      "learning_rate": 1.0709765450396405e-05,
      "loss": 1.1954,
      "step": 21060
    },
    {
      "epoch": 0.4808087262105792,
      "grad_norm": 2.96875,
      "learning_rate": 1.0705060343002329e-05,
      "loss": 1.1867,
      "step": 21070
    },
    {
      "epoch": 0.4810369220939254,
      "grad_norm": 3.328125,
      "learning_rate": 1.0700355235608252e-05,
      "loss": 1.1699,
      "step": 21080
    },
    {
      "epoch": 0.4812651179772717,
      "grad_norm": 2.890625,
      "learning_rate": 1.0695650128214176e-05,
      "loss": 1.1709,
      "step": 21090
    },
    {
      "epoch": 0.48149331386061794,
      "grad_norm": 3.28125,
      "learning_rate": 1.06909450208201e-05,
      "loss": 1.2597,
      "step": 21100
    },
    {
      "epoch": 0.4817215097439642,
      "grad_norm": 3.59375,
      "learning_rate": 1.0686239913426023e-05,
      "loss": 1.2044,
      "step": 21110
    },
    {
      "epoch": 0.4819497056273105,
      "grad_norm": 3.453125,
      "learning_rate": 1.0681534806031947e-05,
      "loss": 1.2003,
      "step": 21120
    },
    {
      "epoch": 0.48217790151065676,
      "grad_norm": 2.953125,
      "learning_rate": 1.0676829698637873e-05,
      "loss": 1.1694,
      "step": 21130
    },
    {
      "epoch": 0.48240609739400303,
      "grad_norm": 3.09375,
      "learning_rate": 1.0672124591243796e-05,
      "loss": 1.1957,
      "step": 21140
    },
    {
      "epoch": 0.4826342932773493,
      "grad_norm": 3.28125,
      "learning_rate": 1.066741948384972e-05,
      "loss": 1.182,
      "step": 21150
    },
    {
      "epoch": 0.4828624891606955,
      "grad_norm": 3.21875,
      "learning_rate": 1.0662714376455644e-05,
      "loss": 1.1409,
      "step": 21160
    },
    {
      "epoch": 0.4830906850440418,
      "grad_norm": 3.125,
      "learning_rate": 1.0658009269061567e-05,
      "loss": 1.1581,
      "step": 21170
    },
    {
      "epoch": 0.48331888092738806,
      "grad_norm": 3.015625,
      "learning_rate": 1.0653304161667491e-05,
      "loss": 1.1866,
      "step": 21180
    },
    {
      "epoch": 0.48354707681073433,
      "grad_norm": 3.25,
      "learning_rate": 1.0648599054273415e-05,
      "loss": 1.1784,
      "step": 21190
    },
    {
      "epoch": 0.4837752726940806,
      "grad_norm": 2.953125,
      "learning_rate": 1.0643893946879338e-05,
      "loss": 1.2022,
      "step": 21200
    },
    {
      "epoch": 0.4840034685774269,
      "grad_norm": 3.25,
      "learning_rate": 1.0639188839485262e-05,
      "loss": 1.2429,
      "step": 21210
    },
    {
      "epoch": 0.48423166446077315,
      "grad_norm": 3.390625,
      "learning_rate": 1.0634483732091186e-05,
      "loss": 1.2021,
      "step": 21220
    },
    {
      "epoch": 0.4844598603441194,
      "grad_norm": 3.15625,
      "learning_rate": 1.062977862469711e-05,
      "loss": 1.1826,
      "step": 21230
    },
    {
      "epoch": 0.48468805622746564,
      "grad_norm": 3.21875,
      "learning_rate": 1.0625073517303033e-05,
      "loss": 1.2665,
      "step": 21240
    },
    {
      "epoch": 0.4849162521108119,
      "grad_norm": 3.25,
      "learning_rate": 1.0620368409908957e-05,
      "loss": 1.1124,
      "step": 21250
    },
    {
      "epoch": 0.4851444479941582,
      "grad_norm": 3.25,
      "learning_rate": 1.061566330251488e-05,
      "loss": 1.1961,
      "step": 21260
    },
    {
      "epoch": 0.48537264387750445,
      "grad_norm": 3.5,
      "learning_rate": 1.0610958195120804e-05,
      "loss": 1.1626,
      "step": 21270
    },
    {
      "epoch": 0.4856008397608507,
      "grad_norm": 3.015625,
      "learning_rate": 1.0606253087726728e-05,
      "loss": 1.2416,
      "step": 21280
    },
    {
      "epoch": 0.485829035644197,
      "grad_norm": 2.828125,
      "learning_rate": 1.0601547980332652e-05,
      "loss": 1.1319,
      "step": 21290
    },
    {
      "epoch": 0.48605723152754327,
      "grad_norm": 3.28125,
      "learning_rate": 1.0596842872938576e-05,
      "loss": 1.2151,
      "step": 21300
    },
    {
      "epoch": 0.4862854274108895,
      "grad_norm": 3.046875,
      "learning_rate": 1.05921377655445e-05,
      "loss": 1.2236,
      "step": 21310
    },
    {
      "epoch": 0.48651362329423575,
      "grad_norm": 3.4375,
      "learning_rate": 1.0587432658150423e-05,
      "loss": 1.2043,
      "step": 21320
    },
    {
      "epoch": 0.486741819177582,
      "grad_norm": 3.1875,
      "learning_rate": 1.0582727550756347e-05,
      "loss": 1.179,
      "step": 21330
    },
    {
      "epoch": 0.4869700150609283,
      "grad_norm": 2.953125,
      "learning_rate": 1.057802244336227e-05,
      "loss": 1.1472,
      "step": 21340
    },
    {
      "epoch": 0.48719821094427457,
      "grad_norm": 3.0625,
      "learning_rate": 1.0573317335968194e-05,
      "loss": 1.1818,
      "step": 21350
    },
    {
      "epoch": 0.48742640682762084,
      "grad_norm": 3.171875,
      "learning_rate": 1.0568612228574118e-05,
      "loss": 1.2285,
      "step": 21360
    },
    {
      "epoch": 0.4876546027109671,
      "grad_norm": 3.09375,
      "learning_rate": 1.0563907121180042e-05,
      "loss": 1.1853,
      "step": 21370
    },
    {
      "epoch": 0.4878827985943134,
      "grad_norm": 3.171875,
      "learning_rate": 1.0559202013785965e-05,
      "loss": 1.1717,
      "step": 21380
    },
    {
      "epoch": 0.4881109944776596,
      "grad_norm": 3.203125,
      "learning_rate": 1.0554496906391889e-05,
      "loss": 1.1674,
      "step": 21390
    },
    {
      "epoch": 0.48833919036100587,
      "grad_norm": 2.859375,
      "learning_rate": 1.0549791798997813e-05,
      "loss": 1.1727,
      "step": 21400
    },
    {
      "epoch": 0.48856738624435214,
      "grad_norm": 3.21875,
      "learning_rate": 1.0545086691603736e-05,
      "loss": 1.1635,
      "step": 21410
    },
    {
      "epoch": 0.4887955821276984,
      "grad_norm": 3.3125,
      "learning_rate": 1.054038158420966e-05,
      "loss": 1.1686,
      "step": 21420
    },
    {
      "epoch": 0.4890237780110447,
      "grad_norm": 3.125,
      "learning_rate": 1.0535676476815584e-05,
      "loss": 1.1537,
      "step": 21430
    },
    {
      "epoch": 0.48925197389439096,
      "grad_norm": 3.21875,
      "learning_rate": 1.0530971369421508e-05,
      "loss": 1.1704,
      "step": 21440
    },
    {
      "epoch": 0.48948016977773723,
      "grad_norm": 2.8125,
      "learning_rate": 1.0526266262027431e-05,
      "loss": 1.1625,
      "step": 21450
    },
    {
      "epoch": 0.4897083656610835,
      "grad_norm": 3.0625,
      "learning_rate": 1.0521561154633355e-05,
      "loss": 1.2308,
      "step": 21460
    },
    {
      "epoch": 0.4899365615444297,
      "grad_norm": 3.09375,
      "learning_rate": 1.0516856047239279e-05,
      "loss": 1.2013,
      "step": 21470
    },
    {
      "epoch": 0.490164757427776,
      "grad_norm": 3.265625,
      "learning_rate": 1.0512150939845202e-05,
      "loss": 1.2345,
      "step": 21480
    },
    {
      "epoch": 0.49039295331112226,
      "grad_norm": 3.109375,
      "learning_rate": 1.0507445832451126e-05,
      "loss": 1.1829,
      "step": 21490
    },
    {
      "epoch": 0.49062114919446853,
      "grad_norm": 3.078125,
      "learning_rate": 1.050274072505705e-05,
      "loss": 1.2147,
      "step": 21500
    },
    {
      "epoch": 0.4908493450778148,
      "grad_norm": 3.171875,
      "learning_rate": 1.0498035617662974e-05,
      "loss": 1.1961,
      "step": 21510
    },
    {
      "epoch": 0.4910775409611611,
      "grad_norm": 3.0625,
      "learning_rate": 1.0493330510268897e-05,
      "loss": 1.1689,
      "step": 21520
    },
    {
      "epoch": 0.49130573684450735,
      "grad_norm": 3.1875,
      "learning_rate": 1.0488625402874821e-05,
      "loss": 1.1826,
      "step": 21530
    },
    {
      "epoch": 0.49153393272785356,
      "grad_norm": 3.203125,
      "learning_rate": 1.0483920295480745e-05,
      "loss": 1.2234,
      "step": 21540
    },
    {
      "epoch": 0.49176212861119983,
      "grad_norm": 3.25,
      "learning_rate": 1.0479215188086668e-05,
      "loss": 1.1556,
      "step": 21550
    },
    {
      "epoch": 0.4919903244945461,
      "grad_norm": 3.09375,
      "learning_rate": 1.0474510080692592e-05,
      "loss": 1.2398,
      "step": 21560
    },
    {
      "epoch": 0.4922185203778924,
      "grad_norm": 3.203125,
      "learning_rate": 1.0469804973298516e-05,
      "loss": 1.1993,
      "step": 21570
    },
    {
      "epoch": 0.49244671626123865,
      "grad_norm": 3.15625,
      "learning_rate": 1.046509986590444e-05,
      "loss": 1.1878,
      "step": 21580
    },
    {
      "epoch": 0.4926749121445849,
      "grad_norm": 3.3125,
      "learning_rate": 1.0460394758510363e-05,
      "loss": 1.1687,
      "step": 21590
    },
    {
      "epoch": 0.4929031080279312,
      "grad_norm": 3.125,
      "learning_rate": 1.0455689651116287e-05,
      "loss": 1.2103,
      "step": 21600
    },
    {
      "epoch": 0.49313130391127746,
      "grad_norm": 3.265625,
      "learning_rate": 1.045098454372221e-05,
      "loss": 1.1904,
      "step": 21610
    },
    {
      "epoch": 0.4933594997946237,
      "grad_norm": 3.015625,
      "learning_rate": 1.0446279436328134e-05,
      "loss": 1.1383,
      "step": 21620
    },
    {
      "epoch": 0.49358769567796995,
      "grad_norm": 2.84375,
      "learning_rate": 1.0441574328934058e-05,
      "loss": 1.2506,
      "step": 21630
    },
    {
      "epoch": 0.4938158915613162,
      "grad_norm": 3.75,
      "learning_rate": 1.0436869221539982e-05,
      "loss": 1.1705,
      "step": 21640
    },
    {
      "epoch": 0.4940440874446625,
      "grad_norm": 3.109375,
      "learning_rate": 1.0432164114145906e-05,
      "loss": 1.1842,
      "step": 21650
    },
    {
      "epoch": 0.49427228332800877,
      "grad_norm": 3.203125,
      "learning_rate": 1.042745900675183e-05,
      "loss": 1.1475,
      "step": 21660
    },
    {
      "epoch": 0.49450047921135504,
      "grad_norm": 2.9375,
      "learning_rate": 1.0422753899357753e-05,
      "loss": 1.1852,
      "step": 21670
    },
    {
      "epoch": 0.4947286750947013,
      "grad_norm": 3.046875,
      "learning_rate": 1.0418048791963677e-05,
      "loss": 1.2278,
      "step": 21680
    },
    {
      "epoch": 0.4949568709780476,
      "grad_norm": 3.234375,
      "learning_rate": 1.04133436845696e-05,
      "loss": 1.1759,
      "step": 21690
    },
    {
      "epoch": 0.4951850668613938,
      "grad_norm": 3.140625,
      "learning_rate": 1.0408638577175524e-05,
      "loss": 1.225,
      "step": 21700
    },
    {
      "epoch": 0.49541326274474007,
      "grad_norm": 3.078125,
      "learning_rate": 1.040393346978145e-05,
      "loss": 1.2385,
      "step": 21710
    },
    {
      "epoch": 0.49564145862808634,
      "grad_norm": 2.921875,
      "learning_rate": 1.0399228362387373e-05,
      "loss": 1.2278,
      "step": 21720
    },
    {
      "epoch": 0.4958696545114326,
      "grad_norm": 2.96875,
      "learning_rate": 1.0394523254993297e-05,
      "loss": 1.1787,
      "step": 21730
    },
    {
      "epoch": 0.4960978503947789,
      "grad_norm": 3.0,
      "learning_rate": 1.038981814759922e-05,
      "loss": 1.1357,
      "step": 21740
    },
    {
      "epoch": 0.49632604627812515,
      "grad_norm": 3.1875,
      "learning_rate": 1.0385113040205144e-05,
      "loss": 1.239,
      "step": 21750
    },
    {
      "epoch": 0.4965542421614714,
      "grad_norm": 3.34375,
      "learning_rate": 1.0380407932811068e-05,
      "loss": 1.2074,
      "step": 21760
    },
    {
      "epoch": 0.4967824380448177,
      "grad_norm": 2.984375,
      "learning_rate": 1.0375702825416992e-05,
      "loss": 1.169,
      "step": 21770
    },
    {
      "epoch": 0.4970106339281639,
      "grad_norm": 3.515625,
      "learning_rate": 1.0370997718022916e-05,
      "loss": 1.1339,
      "step": 21780
    },
    {
      "epoch": 0.4972388298115102,
      "grad_norm": 3.078125,
      "learning_rate": 1.036629261062884e-05,
      "loss": 1.1691,
      "step": 21790
    },
    {
      "epoch": 0.49746702569485646,
      "grad_norm": 3.15625,
      "learning_rate": 1.0361587503234763e-05,
      "loss": 1.1783,
      "step": 21800
    },
    {
      "epoch": 0.49769522157820273,
      "grad_norm": 3.171875,
      "learning_rate": 1.0356882395840687e-05,
      "loss": 1.1801,
      "step": 21810
    },
    {
      "epoch": 0.497923417461549,
      "grad_norm": 3.25,
      "learning_rate": 1.035217728844661e-05,
      "loss": 1.1825,
      "step": 21820
    },
    {
      "epoch": 0.49815161334489527,
      "grad_norm": 3.28125,
      "learning_rate": 1.0347472181052534e-05,
      "loss": 1.1948,
      "step": 21830
    },
    {
      "epoch": 0.49837980922824154,
      "grad_norm": 3.125,
      "learning_rate": 1.0342767073658458e-05,
      "loss": 1.178,
      "step": 21840
    },
    {
      "epoch": 0.49860800511158776,
      "grad_norm": 3.21875,
      "learning_rate": 1.0338061966264381e-05,
      "loss": 1.1966,
      "step": 21850
    },
    {
      "epoch": 0.49883620099493403,
      "grad_norm": 3.09375,
      "learning_rate": 1.0333356858870305e-05,
      "loss": 1.1886,
      "step": 21860
    },
    {
      "epoch": 0.4990643968782803,
      "grad_norm": 3.453125,
      "learning_rate": 1.0328651751476229e-05,
      "loss": 1.162,
      "step": 21870
    },
    {
      "epoch": 0.4992925927616266,
      "grad_norm": 2.875,
      "learning_rate": 1.0323946644082153e-05,
      "loss": 1.1814,
      "step": 21880
    },
    {
      "epoch": 0.49952078864497285,
      "grad_norm": 3.078125,
      "learning_rate": 1.0319241536688076e-05,
      "loss": 1.268,
      "step": 21890
    },
    {
      "epoch": 0.4997489845283191,
      "grad_norm": 3.0625,
      "learning_rate": 1.0314536429294e-05,
      "loss": 1.1998,
      "step": 21900
    },
    {
      "epoch": 0.4999771804116654,
      "grad_norm": 2.984375,
      "learning_rate": 1.0309831321899924e-05,
      "loss": 1.151,
      "step": 21910
    },
    {
      "epoch": 0.5002053762950116,
      "grad_norm": 3.109375,
      "learning_rate": 1.0305126214505847e-05,
      "loss": 1.1854,
      "step": 21920
    },
    {
      "epoch": 0.5004335721783579,
      "grad_norm": 3.21875,
      "learning_rate": 1.0300421107111771e-05,
      "loss": 1.1575,
      "step": 21930
    },
    {
      "epoch": 0.5006617680617041,
      "grad_norm": 2.78125,
      "learning_rate": 1.0295715999717695e-05,
      "loss": 1.1929,
      "step": 21940
    },
    {
      "epoch": 0.5008899639450505,
      "grad_norm": 3.046875,
      "learning_rate": 1.0291010892323619e-05,
      "loss": 1.1987,
      "step": 21950
    },
    {
      "epoch": 0.5011181598283967,
      "grad_norm": 3.15625,
      "learning_rate": 1.0286305784929542e-05,
      "loss": 1.199,
      "step": 21960
    },
    {
      "epoch": 0.5013463557117429,
      "grad_norm": 2.796875,
      "learning_rate": 1.0281600677535466e-05,
      "loss": 1.192,
      "step": 21970
    },
    {
      "epoch": 0.5015745515950892,
      "grad_norm": 3.109375,
      "learning_rate": 1.027689557014139e-05,
      "loss": 1.2163,
      "step": 21980
    },
    {
      "epoch": 0.5018027474784355,
      "grad_norm": 2.984375,
      "learning_rate": 1.0272190462747313e-05,
      "loss": 1.1839,
      "step": 21990
    },
    {
      "epoch": 0.5020309433617818,
      "grad_norm": 2.953125,
      "learning_rate": 1.0267485355353237e-05,
      "loss": 1.2306,
      "step": 22000
    },
    {
      "epoch": 0.502259139245128,
      "grad_norm": 3.203125,
      "learning_rate": 1.0262780247959161e-05,
      "loss": 1.2356,
      "step": 22010
    },
    {
      "epoch": 0.5024873351284743,
      "grad_norm": 3.0625,
      "learning_rate": 1.0258075140565085e-05,
      "loss": 1.1783,
      "step": 22020
    },
    {
      "epoch": 0.5027155310118205,
      "grad_norm": 3.109375,
      "learning_rate": 1.0253370033171008e-05,
      "loss": 1.181,
      "step": 22030
    },
    {
      "epoch": 0.5029437268951669,
      "grad_norm": 3.15625,
      "learning_rate": 1.0248664925776932e-05,
      "loss": 1.1735,
      "step": 22040
    },
    {
      "epoch": 0.5031719227785131,
      "grad_norm": 3.046875,
      "learning_rate": 1.0243959818382856e-05,
      "loss": 1.1803,
      "step": 22050
    },
    {
      "epoch": 0.5034001186618593,
      "grad_norm": 3.21875,
      "learning_rate": 1.023925471098878e-05,
      "loss": 1.1959,
      "step": 22060
    },
    {
      "epoch": 0.5036283145452056,
      "grad_norm": 2.890625,
      "learning_rate": 1.0234549603594703e-05,
      "loss": 1.1598,
      "step": 22070
    },
    {
      "epoch": 0.5038565104285518,
      "grad_norm": 3.140625,
      "learning_rate": 1.0229844496200627e-05,
      "loss": 1.1932,
      "step": 22080
    },
    {
      "epoch": 0.5040847063118982,
      "grad_norm": 3.34375,
      "learning_rate": 1.022513938880655e-05,
      "loss": 1.1645,
      "step": 22090
    },
    {
      "epoch": 0.5043129021952444,
      "grad_norm": 3.09375,
      "learning_rate": 1.0220434281412474e-05,
      "loss": 1.1856,
      "step": 22100
    },
    {
      "epoch": 0.5045410980785907,
      "grad_norm": 3.09375,
      "learning_rate": 1.0215729174018398e-05,
      "loss": 1.1796,
      "step": 22110
    },
    {
      "epoch": 0.5047692939619369,
      "grad_norm": 3.171875,
      "learning_rate": 1.0211024066624322e-05,
      "loss": 1.2156,
      "step": 22120
    },
    {
      "epoch": 0.5049974898452831,
      "grad_norm": 3.265625,
      "learning_rate": 1.0206318959230245e-05,
      "loss": 1.2779,
      "step": 22130
    },
    {
      "epoch": 0.5052256857286295,
      "grad_norm": 3.421875,
      "learning_rate": 1.0201613851836169e-05,
      "loss": 1.2281,
      "step": 22140
    },
    {
      "epoch": 0.5054538816119757,
      "grad_norm": 2.953125,
      "learning_rate": 1.0196908744442093e-05,
      "loss": 1.1926,
      "step": 22150
    },
    {
      "epoch": 0.505682077495322,
      "grad_norm": 2.90625,
      "learning_rate": 1.0192203637048017e-05,
      "loss": 1.2588,
      "step": 22160
    },
    {
      "epoch": 0.5059102733786682,
      "grad_norm": 3.234375,
      "learning_rate": 1.018749852965394e-05,
      "loss": 1.1385,
      "step": 22170
    },
    {
      "epoch": 0.5061384692620146,
      "grad_norm": 3.03125,
      "learning_rate": 1.0182793422259864e-05,
      "loss": 1.2105,
      "step": 22180
    },
    {
      "epoch": 0.5063666651453608,
      "grad_norm": 3.296875,
      "learning_rate": 1.0178088314865788e-05,
      "loss": 1.1773,
      "step": 22190
    },
    {
      "epoch": 0.506594861028707,
      "grad_norm": 3.28125,
      "learning_rate": 1.0173383207471711e-05,
      "loss": 1.2494,
      "step": 22200
    },
    {
      "epoch": 0.5068230569120533,
      "grad_norm": 3.375,
      "learning_rate": 1.0168678100077635e-05,
      "loss": 1.1463,
      "step": 22210
    },
    {
      "epoch": 0.5070512527953995,
      "grad_norm": 3.25,
      "learning_rate": 1.0163972992683559e-05,
      "loss": 1.2166,
      "step": 22220
    },
    {
      "epoch": 0.5072794486787459,
      "grad_norm": 3.09375,
      "learning_rate": 1.0159267885289483e-05,
      "loss": 1.1242,
      "step": 22230
    },
    {
      "epoch": 0.5075076445620921,
      "grad_norm": 3.171875,
      "learning_rate": 1.0154562777895406e-05,
      "loss": 1.1862,
      "step": 22240
    },
    {
      "epoch": 0.5077358404454384,
      "grad_norm": 3.21875,
      "learning_rate": 1.014985767050133e-05,
      "loss": 1.1722,
      "step": 22250
    },
    {
      "epoch": 0.5079640363287846,
      "grad_norm": 3.296875,
      "learning_rate": 1.0145152563107254e-05,
      "loss": 1.1789,
      "step": 22260
    },
    {
      "epoch": 0.508192232212131,
      "grad_norm": 3.078125,
      "learning_rate": 1.0140447455713177e-05,
      "loss": 1.1103,
      "step": 22270
    },
    {
      "epoch": 0.5084204280954772,
      "grad_norm": 3.21875,
      "learning_rate": 1.0135742348319103e-05,
      "loss": 1.1696,
      "step": 22280
    },
    {
      "epoch": 0.5086486239788234,
      "grad_norm": 2.953125,
      "learning_rate": 1.0131037240925027e-05,
      "loss": 1.1942,
      "step": 22290
    },
    {
      "epoch": 0.5088768198621697,
      "grad_norm": 2.953125,
      "learning_rate": 1.012633213353095e-05,
      "loss": 1.1496,
      "step": 22300
    },
    {
      "epoch": 0.5091050157455159,
      "grad_norm": 3.03125,
      "learning_rate": 1.0121627026136874e-05,
      "loss": 1.1725,
      "step": 22310
    },
    {
      "epoch": 0.5093332116288622,
      "grad_norm": 3.0625,
      "learning_rate": 1.0116921918742798e-05,
      "loss": 1.1968,
      "step": 22320
    },
    {
      "epoch": 0.5095614075122085,
      "grad_norm": 3.28125,
      "learning_rate": 1.0112216811348721e-05,
      "loss": 1.1811,
      "step": 22330
    },
    {
      "epoch": 0.5097896033955548,
      "grad_norm": 2.828125,
      "learning_rate": 1.0107511703954645e-05,
      "loss": 1.1459,
      "step": 22340
    },
    {
      "epoch": 0.510017799278901,
      "grad_norm": 2.921875,
      "learning_rate": 1.0102806596560569e-05,
      "loss": 1.1944,
      "step": 22350
    },
    {
      "epoch": 0.5102459951622472,
      "grad_norm": 2.984375,
      "learning_rate": 1.0098101489166493e-05,
      "loss": 1.1583,
      "step": 22360
    },
    {
      "epoch": 0.5104741910455936,
      "grad_norm": 3.015625,
      "learning_rate": 1.0093396381772416e-05,
      "loss": 1.1857,
      "step": 22370
    },
    {
      "epoch": 0.5107023869289398,
      "grad_norm": 3.234375,
      "learning_rate": 1.008869127437834e-05,
      "loss": 1.2132,
      "step": 22380
    },
    {
      "epoch": 0.5109305828122861,
      "grad_norm": 3.265625,
      "learning_rate": 1.0083986166984264e-05,
      "loss": 1.2213,
      "step": 22390
    },
    {
      "epoch": 0.5111587786956323,
      "grad_norm": 2.84375,
      "learning_rate": 1.0079281059590187e-05,
      "loss": 1.2007,
      "step": 22400
    },
    {
      "epoch": 0.5113869745789786,
      "grad_norm": 3.015625,
      "learning_rate": 1.0074575952196111e-05,
      "loss": 1.1755,
      "step": 22410
    },
    {
      "epoch": 0.5116151704623249,
      "grad_norm": 3.171875,
      "learning_rate": 1.0069870844802035e-05,
      "loss": 1.1918,
      "step": 22420
    },
    {
      "epoch": 0.5118433663456711,
      "grad_norm": 3.34375,
      "learning_rate": 1.0065165737407959e-05,
      "loss": 1.2352,
      "step": 22430
    },
    {
      "epoch": 0.5120715622290174,
      "grad_norm": 3.0625,
      "learning_rate": 1.0060460630013882e-05,
      "loss": 1.1724,
      "step": 22440
    },
    {
      "epoch": 0.5122997581123636,
      "grad_norm": 3.1875,
      "learning_rate": 1.0055755522619806e-05,
      "loss": 1.1761,
      "step": 22450
    },
    {
      "epoch": 0.5125279539957099,
      "grad_norm": 2.828125,
      "learning_rate": 1.005105041522573e-05,
      "loss": 1.1558,
      "step": 22460
    },
    {
      "epoch": 0.5127561498790562,
      "grad_norm": 3.046875,
      "learning_rate": 1.0046345307831653e-05,
      "loss": 1.2227,
      "step": 22470
    },
    {
      "epoch": 0.5129843457624025,
      "grad_norm": 3.140625,
      "learning_rate": 1.0041640200437577e-05,
      "loss": 1.2184,
      "step": 22480
    },
    {
      "epoch": 0.5132125416457487,
      "grad_norm": 3.25,
      "learning_rate": 1.00369350930435e-05,
      "loss": 1.1805,
      "step": 22490
    },
    {
      "epoch": 0.513440737529095,
      "grad_norm": 3.125,
      "learning_rate": 1.0032229985649424e-05,
      "loss": 1.1374,
      "step": 22500
    },
    {
      "epoch": 0.5136689334124412,
      "grad_norm": 3.03125,
      "learning_rate": 1.0027524878255348e-05,
      "loss": 1.1819,
      "step": 22510
    },
    {
      "epoch": 0.5138971292957875,
      "grad_norm": 2.921875,
      "learning_rate": 1.0022819770861272e-05,
      "loss": 1.1385,
      "step": 22520
    },
    {
      "epoch": 0.5141253251791338,
      "grad_norm": 3.03125,
      "learning_rate": 1.0018114663467196e-05,
      "loss": 1.1976,
      "step": 22530
    },
    {
      "epoch": 0.51435352106248,
      "grad_norm": 2.90625,
      "learning_rate": 1.001340955607312e-05,
      "loss": 1.2008,
      "step": 22540
    },
    {
      "epoch": 0.5145817169458263,
      "grad_norm": 2.96875,
      "learning_rate": 1.0008704448679043e-05,
      "loss": 1.1654,
      "step": 22550
    },
    {
      "epoch": 0.5148099128291725,
      "grad_norm": 3.203125,
      "learning_rate": 1.0003999341284967e-05,
      "loss": 1.2386,
      "step": 22560
    },
    {
      "epoch": 0.5150381087125189,
      "grad_norm": 3.234375,
      "learning_rate": 9.999294233890889e-06,
      "loss": 1.17,
      "step": 22570
    },
    {
      "epoch": 0.5152663045958651,
      "grad_norm": 2.9375,
      "learning_rate": 9.994589126496812e-06,
      "loss": 1.1875,
      "step": 22580
    },
    {
      "epoch": 0.5154945004792113,
      "grad_norm": 3.015625,
      "learning_rate": 9.989884019102736e-06,
      "loss": 1.172,
      "step": 22590
    },
    {
      "epoch": 0.5157226963625576,
      "grad_norm": 3.203125,
      "learning_rate": 9.98517891170866e-06,
      "loss": 1.2115,
      "step": 22600
    },
    {
      "epoch": 0.5159508922459038,
      "grad_norm": 3.046875,
      "learning_rate": 9.980473804314584e-06,
      "loss": 1.18,
      "step": 22610
    },
    {
      "epoch": 0.5161790881292502,
      "grad_norm": 3.109375,
      "learning_rate": 9.975768696920507e-06,
      "loss": 1.2236,
      "step": 22620
    },
    {
      "epoch": 0.5164072840125964,
      "grad_norm": 3.21875,
      "learning_rate": 9.971063589526431e-06,
      "loss": 1.2208,
      "step": 22630
    },
    {
      "epoch": 0.5166354798959427,
      "grad_norm": 3.234375,
      "learning_rate": 9.966358482132355e-06,
      "loss": 1.1926,
      "step": 22640
    },
    {
      "epoch": 0.5168636757792889,
      "grad_norm": 3.0,
      "learning_rate": 9.961653374738278e-06,
      "loss": 1.1943,
      "step": 22650
    },
    {
      "epoch": 0.5170918716626353,
      "grad_norm": 2.859375,
      "learning_rate": 9.956948267344204e-06,
      "loss": 1.1958,
      "step": 22660
    },
    {
      "epoch": 0.5173200675459815,
      "grad_norm": 3.390625,
      "learning_rate": 9.952243159950128e-06,
      "loss": 1.2341,
      "step": 22670
    },
    {
      "epoch": 0.5175482634293277,
      "grad_norm": 2.9375,
      "learning_rate": 9.947538052556051e-06,
      "loss": 1.1283,
      "step": 22680
    },
    {
      "epoch": 0.517776459312674,
      "grad_norm": 3.25,
      "learning_rate": 9.942832945161975e-06,
      "loss": 1.2147,
      "step": 22690
    },
    {
      "epoch": 0.5180046551960202,
      "grad_norm": 3.109375,
      "learning_rate": 9.938127837767899e-06,
      "loss": 1.2064,
      "step": 22700
    },
    {
      "epoch": 0.5182328510793666,
      "grad_norm": 3.21875,
      "learning_rate": 9.933422730373822e-06,
      "loss": 1.1966,
      "step": 22710
    },
    {
      "epoch": 0.5184610469627128,
      "grad_norm": 3.046875,
      "learning_rate": 9.928717622979746e-06,
      "loss": 1.1433,
      "step": 22720
    },
    {
      "epoch": 0.5186892428460591,
      "grad_norm": 3.578125,
      "learning_rate": 9.92401251558567e-06,
      "loss": 1.1749,
      "step": 22730
    },
    {
      "epoch": 0.5189174387294053,
      "grad_norm": 2.984375,
      "learning_rate": 9.919307408191594e-06,
      "loss": 1.1505,
      "step": 22740
    },
    {
      "epoch": 0.5191456346127515,
      "grad_norm": 3.265625,
      "learning_rate": 9.914602300797517e-06,
      "loss": 1.1551,
      "step": 22750
    },
    {
      "epoch": 0.5193738304960979,
      "grad_norm": 3.46875,
      "learning_rate": 9.909897193403441e-06,
      "loss": 1.2364,
      "step": 22760
    },
    {
      "epoch": 0.5196020263794441,
      "grad_norm": 3.171875,
      "learning_rate": 9.905192086009365e-06,
      "loss": 1.15,
      "step": 22770
    },
    {
      "epoch": 0.5198302222627904,
      "grad_norm": 3.109375,
      "learning_rate": 9.900486978615288e-06,
      "loss": 1.162,
      "step": 22780
    },
    {
      "epoch": 0.5200584181461366,
      "grad_norm": 2.984375,
      "learning_rate": 9.895781871221212e-06,
      "loss": 1.17,
      "step": 22790
    },
    {
      "epoch": 0.520286614029483,
      "grad_norm": 3.1875,
      "learning_rate": 9.891076763827136e-06,
      "loss": 1.1752,
      "step": 22800
    },
    {
      "epoch": 0.5205148099128292,
      "grad_norm": 3.3125,
      "learning_rate": 9.88637165643306e-06,
      "loss": 1.1778,
      "step": 22810
    },
    {
      "epoch": 0.5207430057961754,
      "grad_norm": 3.0625,
      "learning_rate": 9.881666549038983e-06,
      "loss": 1.1443,
      "step": 22820
    },
    {
      "epoch": 0.5209712016795217,
      "grad_norm": 3.109375,
      "learning_rate": 9.876961441644907e-06,
      "loss": 1.202,
      "step": 22830
    },
    {
      "epoch": 0.5211993975628679,
      "grad_norm": 3.25,
      "learning_rate": 9.87225633425083e-06,
      "loss": 1.1955,
      "step": 22840
    },
    {
      "epoch": 0.5214275934462143,
      "grad_norm": 3.5625,
      "learning_rate": 9.867551226856754e-06,
      "loss": 1.1982,
      "step": 22850
    },
    {
      "epoch": 0.5216557893295605,
      "grad_norm": 3.078125,
      "learning_rate": 9.862846119462678e-06,
      "loss": 1.2022,
      "step": 22860
    },
    {
      "epoch": 0.5218839852129068,
      "grad_norm": 3.265625,
      "learning_rate": 9.858141012068602e-06,
      "loss": 1.1837,
      "step": 22870
    },
    {
      "epoch": 0.522112181096253,
      "grad_norm": 3.078125,
      "learning_rate": 9.853435904674526e-06,
      "loss": 1.1612,
      "step": 22880
    },
    {
      "epoch": 0.5223403769795993,
      "grad_norm": 3.09375,
      "learning_rate": 9.84873079728045e-06,
      "loss": 1.2066,
      "step": 22890
    },
    {
      "epoch": 0.5225685728629456,
      "grad_norm": 3.140625,
      "learning_rate": 9.844025689886373e-06,
      "loss": 1.2247,
      "step": 22900
    },
    {
      "epoch": 0.5227967687462918,
      "grad_norm": 3.21875,
      "learning_rate": 9.839320582492297e-06,
      "loss": 1.1915,
      "step": 22910
    },
    {
      "epoch": 0.5230249646296381,
      "grad_norm": 2.984375,
      "learning_rate": 9.83461547509822e-06,
      "loss": 1.1708,
      "step": 22920
    },
    {
      "epoch": 0.5232531605129843,
      "grad_norm": 2.875,
      "learning_rate": 9.829910367704144e-06,
      "loss": 1.1726,
      "step": 22930
    },
    {
      "epoch": 0.5234813563963306,
      "grad_norm": 3.1875,
      "learning_rate": 9.825205260310068e-06,
      "loss": 1.1807,
      "step": 22940
    },
    {
      "epoch": 0.5237095522796769,
      "grad_norm": 2.984375,
      "learning_rate": 9.820500152915992e-06,
      "loss": 1.2047,
      "step": 22950
    },
    {
      "epoch": 0.5239377481630232,
      "grad_norm": 3.390625,
      "learning_rate": 9.815795045521915e-06,
      "loss": 1.1829,
      "step": 22960
    },
    {
      "epoch": 0.5241659440463694,
      "grad_norm": 3.046875,
      "learning_rate": 9.811089938127839e-06,
      "loss": 1.2405,
      "step": 22970
    },
    {
      "epoch": 0.5243941399297156,
      "grad_norm": 3.1875,
      "learning_rate": 9.806384830733763e-06,
      "loss": 1.2471,
      "step": 22980
    },
    {
      "epoch": 0.524622335813062,
      "grad_norm": 3.078125,
      "learning_rate": 9.801679723339686e-06,
      "loss": 1.1545,
      "step": 22990
    },
    {
      "epoch": 0.5248505316964082,
      "grad_norm": 3.8125,
      "learning_rate": 9.79697461594561e-06,
      "loss": 1.1923,
      "step": 23000
    },
    {
      "epoch": 0.5250787275797545,
      "grad_norm": 3.15625,
      "learning_rate": 9.792269508551534e-06,
      "loss": 1.1855,
      "step": 23010
    },
    {
      "epoch": 0.5253069234631007,
      "grad_norm": 3.25,
      "learning_rate": 9.787564401157458e-06,
      "loss": 1.1555,
      "step": 23020
    },
    {
      "epoch": 0.525535119346447,
      "grad_norm": 3.140625,
      "learning_rate": 9.782859293763381e-06,
      "loss": 1.2446,
      "step": 23030
    },
    {
      "epoch": 0.5257633152297932,
      "grad_norm": 3.421875,
      "learning_rate": 9.778154186369305e-06,
      "loss": 1.1935,
      "step": 23040
    },
    {
      "epoch": 0.5259915111131395,
      "grad_norm": 3.421875,
      "learning_rate": 9.773449078975229e-06,
      "loss": 1.1888,
      "step": 23050
    },
    {
      "epoch": 0.5262197069964858,
      "grad_norm": 3.65625,
      "learning_rate": 9.768743971581152e-06,
      "loss": 1.1971,
      "step": 23060
    },
    {
      "epoch": 0.526447902879832,
      "grad_norm": 3.3125,
      "learning_rate": 9.764038864187076e-06,
      "loss": 1.1609,
      "step": 23070
    },
    {
      "epoch": 0.5266760987631783,
      "grad_norm": 3.25,
      "learning_rate": 9.759333756793e-06,
      "loss": 1.2449,
      "step": 23080
    },
    {
      "epoch": 0.5269042946465246,
      "grad_norm": 3.109375,
      "learning_rate": 9.754628649398924e-06,
      "loss": 1.1966,
      "step": 23090
    },
    {
      "epoch": 0.5271324905298709,
      "grad_norm": 2.859375,
      "learning_rate": 9.749923542004847e-06,
      "loss": 1.1745,
      "step": 23100
    },
    {
      "epoch": 0.5273606864132171,
      "grad_norm": 3.640625,
      "learning_rate": 9.745218434610771e-06,
      "loss": 1.2,
      "step": 23110
    },
    {
      "epoch": 0.5275888822965634,
      "grad_norm": 3.0,
      "learning_rate": 9.740513327216695e-06,
      "loss": 1.1743,
      "step": 23120
    },
    {
      "epoch": 0.5278170781799096,
      "grad_norm": 3.046875,
      "learning_rate": 9.735808219822618e-06,
      "loss": 1.1899,
      "step": 23130
    },
    {
      "epoch": 0.5280452740632559,
      "grad_norm": 3.15625,
      "learning_rate": 9.731103112428542e-06,
      "loss": 1.2168,
      "step": 23140
    },
    {
      "epoch": 0.5282734699466022,
      "grad_norm": 3.234375,
      "learning_rate": 9.726398005034466e-06,
      "loss": 1.214,
      "step": 23150
    },
    {
      "epoch": 0.5285016658299484,
      "grad_norm": 2.890625,
      "learning_rate": 9.72169289764039e-06,
      "loss": 1.1966,
      "step": 23160
    },
    {
      "epoch": 0.5287298617132947,
      "grad_norm": 3.484375,
      "learning_rate": 9.716987790246313e-06,
      "loss": 1.194,
      "step": 23170
    },
    {
      "epoch": 0.5289580575966409,
      "grad_norm": 2.96875,
      "learning_rate": 9.712282682852237e-06,
      "loss": 1.1503,
      "step": 23180
    },
    {
      "epoch": 0.5291862534799873,
      "grad_norm": 3.03125,
      "learning_rate": 9.70757757545816e-06,
      "loss": 1.1589,
      "step": 23190
    },
    {
      "epoch": 0.5294144493633335,
      "grad_norm": 3.0,
      "learning_rate": 9.702872468064084e-06,
      "loss": 1.2013,
      "step": 23200
    },
    {
      "epoch": 0.5296426452466797,
      "grad_norm": 2.71875,
      "learning_rate": 9.698167360670008e-06,
      "loss": 1.1561,
      "step": 23210
    },
    {
      "epoch": 0.529870841130026,
      "grad_norm": 3.359375,
      "learning_rate": 9.693462253275932e-06,
      "loss": 1.1978,
      "step": 23220
    },
    {
      "epoch": 0.5300990370133722,
      "grad_norm": 3.515625,
      "learning_rate": 9.688757145881855e-06,
      "loss": 1.191,
      "step": 23230
    },
    {
      "epoch": 0.5303272328967186,
      "grad_norm": 3.3125,
      "learning_rate": 9.68405203848778e-06,
      "loss": 1.1157,
      "step": 23240
    },
    {
      "epoch": 0.5305554287800648,
      "grad_norm": 3.109375,
      "learning_rate": 9.679346931093703e-06,
      "loss": 1.2118,
      "step": 23250
    },
    {
      "epoch": 0.5307836246634111,
      "grad_norm": 3.5,
      "learning_rate": 9.674641823699627e-06,
      "loss": 1.2073,
      "step": 23260
    },
    {
      "epoch": 0.5310118205467573,
      "grad_norm": 3.453125,
      "learning_rate": 9.66993671630555e-06,
      "loss": 1.2592,
      "step": 23270
    },
    {
      "epoch": 0.5312400164301035,
      "grad_norm": 2.96875,
      "learning_rate": 9.665231608911474e-06,
      "loss": 1.2489,
      "step": 23280
    },
    {
      "epoch": 0.5314682123134499,
      "grad_norm": 3.015625,
      "learning_rate": 9.660526501517398e-06,
      "loss": 1.1553,
      "step": 23290
    },
    {
      "epoch": 0.5316964081967961,
      "grad_norm": 3.453125,
      "learning_rate": 9.655821394123321e-06,
      "loss": 1.1613,
      "step": 23300
    },
    {
      "epoch": 0.5319246040801424,
      "grad_norm": 2.96875,
      "learning_rate": 9.651116286729245e-06,
      "loss": 1.1812,
      "step": 23310
    },
    {
      "epoch": 0.5321527999634886,
      "grad_norm": 3.03125,
      "learning_rate": 9.646411179335169e-06,
      "loss": 1.1499,
      "step": 23320
    },
    {
      "epoch": 0.532380995846835,
      "grad_norm": 3.953125,
      "learning_rate": 9.641706071941093e-06,
      "loss": 1.2569,
      "step": 23330
    },
    {
      "epoch": 0.5326091917301812,
      "grad_norm": 3.546875,
      "learning_rate": 9.637000964547016e-06,
      "loss": 1.1641,
      "step": 23340
    },
    {
      "epoch": 0.5328373876135275,
      "grad_norm": 2.828125,
      "learning_rate": 9.63229585715294e-06,
      "loss": 1.1657,
      "step": 23350
    },
    {
      "epoch": 0.5330655834968737,
      "grad_norm": 3.140625,
      "learning_rate": 9.627590749758864e-06,
      "loss": 1.2368,
      "step": 23360
    },
    {
      "epoch": 0.5332937793802199,
      "grad_norm": 3.0,
      "learning_rate": 9.622885642364787e-06,
      "loss": 1.178,
      "step": 23370
    },
    {
      "epoch": 0.5335219752635663,
      "grad_norm": 3.140625,
      "learning_rate": 9.618180534970711e-06,
      "loss": 1.1706,
      "step": 23380
    },
    {
      "epoch": 0.5337501711469125,
      "grad_norm": 3.203125,
      "learning_rate": 9.613475427576635e-06,
      "loss": 1.216,
      "step": 23390
    },
    {
      "epoch": 0.5339783670302588,
      "grad_norm": 3.28125,
      "learning_rate": 9.608770320182559e-06,
      "loss": 1.1745,
      "step": 23400
    },
    {
      "epoch": 0.534206562913605,
      "grad_norm": 3.1875,
      "learning_rate": 9.604065212788482e-06,
      "loss": 1.1996,
      "step": 23410
    },
    {
      "epoch": 0.5344347587969513,
      "grad_norm": 3.0625,
      "learning_rate": 9.599360105394406e-06,
      "loss": 1.2022,
      "step": 23420
    },
    {
      "epoch": 0.5346629546802976,
      "grad_norm": 3.0,
      "learning_rate": 9.59465499800033e-06,
      "loss": 1.1701,
      "step": 23430
    },
    {
      "epoch": 0.5348911505636438,
      "grad_norm": 3.109375,
      "learning_rate": 9.589949890606253e-06,
      "loss": 1.262,
      "step": 23440
    },
    {
      "epoch": 0.5351193464469901,
      "grad_norm": 3.015625,
      "learning_rate": 9.585244783212177e-06,
      "loss": 1.1765,
      "step": 23450
    },
    {
      "epoch": 0.5353475423303363,
      "grad_norm": 3.046875,
      "learning_rate": 9.580539675818101e-06,
      "loss": 1.2254,
      "step": 23460
    },
    {
      "epoch": 0.5355757382136827,
      "grad_norm": 3.171875,
      "learning_rate": 9.575834568424025e-06,
      "loss": 1.1985,
      "step": 23470
    },
    {
      "epoch": 0.5358039340970289,
      "grad_norm": 3.609375,
      "learning_rate": 9.571129461029948e-06,
      "loss": 1.2008,
      "step": 23480
    },
    {
      "epoch": 0.5360321299803752,
      "grad_norm": 3.15625,
      "learning_rate": 9.566424353635872e-06,
      "loss": 1.1915,
      "step": 23490
    },
    {
      "epoch": 0.5362603258637214,
      "grad_norm": 3.109375,
      "learning_rate": 9.561719246241796e-06,
      "loss": 1.1864,
      "step": 23500
    },
    {
      "epoch": 0.5364885217470677,
      "grad_norm": 3.171875,
      "learning_rate": 9.55701413884772e-06,
      "loss": 1.1793,
      "step": 23510
    },
    {
      "epoch": 0.536716717630414,
      "grad_norm": 3.234375,
      "learning_rate": 9.552309031453643e-06,
      "loss": 1.1998,
      "step": 23520
    },
    {
      "epoch": 0.5369449135137602,
      "grad_norm": 3.078125,
      "learning_rate": 9.547603924059567e-06,
      "loss": 1.159,
      "step": 23530
    },
    {
      "epoch": 0.5371731093971065,
      "grad_norm": 3.203125,
      "learning_rate": 9.54289881666549e-06,
      "loss": 1.1725,
      "step": 23540
    },
    {
      "epoch": 0.5374013052804527,
      "grad_norm": 3.171875,
      "learning_rate": 9.538193709271414e-06,
      "loss": 1.1466,
      "step": 23550
    },
    {
      "epoch": 0.537629501163799,
      "grad_norm": 3.515625,
      "learning_rate": 9.533488601877338e-06,
      "loss": 1.1598,
      "step": 23560
    },
    {
      "epoch": 0.5378576970471453,
      "grad_norm": 3.21875,
      "learning_rate": 9.528783494483262e-06,
      "loss": 1.2259,
      "step": 23570
    },
    {
      "epoch": 0.5380858929304916,
      "grad_norm": 3.109375,
      "learning_rate": 9.524078387089185e-06,
      "loss": 1.1877,
      "step": 23580
    },
    {
      "epoch": 0.5383140888138378,
      "grad_norm": 3.15625,
      "learning_rate": 9.519373279695109e-06,
      "loss": 1.225,
      "step": 23590
    },
    {
      "epoch": 0.538542284697184,
      "grad_norm": 3.3125,
      "learning_rate": 9.514668172301033e-06,
      "loss": 1.1707,
      "step": 23600
    },
    {
      "epoch": 0.5387704805805303,
      "grad_norm": 3.53125,
      "learning_rate": 9.509963064906958e-06,
      "loss": 1.1553,
      "step": 23610
    },
    {
      "epoch": 0.5389986764638766,
      "grad_norm": 3.03125,
      "learning_rate": 9.505257957512882e-06,
      "loss": 1.181,
      "step": 23620
    },
    {
      "epoch": 0.5392268723472229,
      "grad_norm": 3.265625,
      "learning_rate": 9.500552850118806e-06,
      "loss": 1.1696,
      "step": 23630
    },
    {
      "epoch": 0.5394550682305691,
      "grad_norm": 3.28125,
      "learning_rate": 9.49584774272473e-06,
      "loss": 1.2249,
      "step": 23640
    },
    {
      "epoch": 0.5396832641139154,
      "grad_norm": 3.015625,
      "learning_rate": 9.491142635330653e-06,
      "loss": 1.1908,
      "step": 23650
    },
    {
      "epoch": 0.5399114599972616,
      "grad_norm": 3.25,
      "learning_rate": 9.486437527936577e-06,
      "loss": 1.2338,
      "step": 23660
    },
    {
      "epoch": 0.5401396558806079,
      "grad_norm": 3.015625,
      "learning_rate": 9.4817324205425e-06,
      "loss": 1.1759,
      "step": 23670
    },
    {
      "epoch": 0.5403678517639542,
      "grad_norm": 3.0625,
      "learning_rate": 9.477027313148424e-06,
      "loss": 1.1898,
      "step": 23680
    },
    {
      "epoch": 0.5405960476473004,
      "grad_norm": 3.0,
      "learning_rate": 9.472322205754348e-06,
      "loss": 1.232,
      "step": 23690
    },
    {
      "epoch": 0.5408242435306467,
      "grad_norm": 3.359375,
      "learning_rate": 9.467617098360272e-06,
      "loss": 1.1785,
      "step": 23700
    },
    {
      "epoch": 0.541052439413993,
      "grad_norm": 3.078125,
      "learning_rate": 9.462911990966195e-06,
      "loss": 1.2013,
      "step": 23710
    },
    {
      "epoch": 0.5412806352973393,
      "grad_norm": 3.625,
      "learning_rate": 9.458206883572119e-06,
      "loss": 1.2148,
      "step": 23720
    },
    {
      "epoch": 0.5415088311806855,
      "grad_norm": 3.171875,
      "learning_rate": 9.453501776178043e-06,
      "loss": 1.124,
      "step": 23730
    },
    {
      "epoch": 0.5417370270640318,
      "grad_norm": 3.25,
      "learning_rate": 9.448796668783967e-06,
      "loss": 1.1739,
      "step": 23740
    },
    {
      "epoch": 0.541965222947378,
      "grad_norm": 3.078125,
      "learning_rate": 9.44409156138989e-06,
      "loss": 1.1493,
      "step": 23750
    },
    {
      "epoch": 0.5421934188307242,
      "grad_norm": 3.171875,
      "learning_rate": 9.439386453995814e-06,
      "loss": 1.1697,
      "step": 23760
    },
    {
      "epoch": 0.5424216147140706,
      "grad_norm": 3.3125,
      "learning_rate": 9.434681346601738e-06,
      "loss": 1.2229,
      "step": 23770
    },
    {
      "epoch": 0.5426498105974168,
      "grad_norm": 3.296875,
      "learning_rate": 9.429976239207661e-06,
      "loss": 1.167,
      "step": 23780
    },
    {
      "epoch": 0.5428780064807631,
      "grad_norm": 3.15625,
      "learning_rate": 9.425271131813585e-06,
      "loss": 1.178,
      "step": 23790
    },
    {
      "epoch": 0.5431062023641093,
      "grad_norm": 3.109375,
      "learning_rate": 9.420566024419509e-06,
      "loss": 1.1554,
      "step": 23800
    },
    {
      "epoch": 0.5433343982474557,
      "grad_norm": 2.84375,
      "learning_rate": 9.415860917025433e-06,
      "loss": 1.1729,
      "step": 23810
    },
    {
      "epoch": 0.5435625941308019,
      "grad_norm": 3.1875,
      "learning_rate": 9.411155809631356e-06,
      "loss": 1.1609,
      "step": 23820
    },
    {
      "epoch": 0.5437907900141481,
      "grad_norm": 3.140625,
      "learning_rate": 9.40645070223728e-06,
      "loss": 1.1474,
      "step": 23830
    },
    {
      "epoch": 0.5440189858974944,
      "grad_norm": 2.90625,
      "learning_rate": 9.401745594843204e-06,
      "loss": 1.2039,
      "step": 23840
    },
    {
      "epoch": 0.5442471817808406,
      "grad_norm": 3.03125,
      "learning_rate": 9.397040487449127e-06,
      "loss": 1.1313,
      "step": 23850
    },
    {
      "epoch": 0.544475377664187,
      "grad_norm": 3.140625,
      "learning_rate": 9.392335380055051e-06,
      "loss": 1.1886,
      "step": 23860
    },
    {
      "epoch": 0.5447035735475332,
      "grad_norm": 3.171875,
      "learning_rate": 9.387630272660975e-06,
      "loss": 1.1862,
      "step": 23870
    },
    {
      "epoch": 0.5449317694308795,
      "grad_norm": 3.046875,
      "learning_rate": 9.382925165266898e-06,
      "loss": 1.1237,
      "step": 23880
    },
    {
      "epoch": 0.5451599653142257,
      "grad_norm": 3.15625,
      "learning_rate": 9.378220057872822e-06,
      "loss": 1.1727,
      "step": 23890
    },
    {
      "epoch": 0.5453881611975719,
      "grad_norm": 3.15625,
      "learning_rate": 9.373514950478746e-06,
      "loss": 1.141,
      "step": 23900
    },
    {
      "epoch": 0.5456163570809183,
      "grad_norm": 3.21875,
      "learning_rate": 9.36880984308467e-06,
      "loss": 1.2254,
      "step": 23910
    },
    {
      "epoch": 0.5458445529642645,
      "grad_norm": 3.3125,
      "learning_rate": 9.364104735690593e-06,
      "loss": 1.1653,
      "step": 23920
    },
    {
      "epoch": 0.5460727488476108,
      "grad_norm": 3.53125,
      "learning_rate": 9.359399628296517e-06,
      "loss": 1.1048,
      "step": 23930
    },
    {
      "epoch": 0.546300944730957,
      "grad_norm": 3.234375,
      "learning_rate": 9.35469452090244e-06,
      "loss": 1.2593,
      "step": 23940
    },
    {
      "epoch": 0.5465291406143034,
      "grad_norm": 3.609375,
      "learning_rate": 9.349989413508364e-06,
      "loss": 1.1977,
      "step": 23950
    },
    {
      "epoch": 0.5467573364976496,
      "grad_norm": 3.03125,
      "learning_rate": 9.345284306114287e-06,
      "loss": 1.1122,
      "step": 23960
    },
    {
      "epoch": 0.5469855323809959,
      "grad_norm": 3.03125,
      "learning_rate": 9.34057919872021e-06,
      "loss": 1.172,
      "step": 23970
    },
    {
      "epoch": 0.5472137282643421,
      "grad_norm": 3.296875,
      "learning_rate": 9.335874091326134e-06,
      "loss": 1.2086,
      "step": 23980
    },
    {
      "epoch": 0.5474419241476883,
      "grad_norm": 2.984375,
      "learning_rate": 9.33116898393206e-06,
      "loss": 1.1904,
      "step": 23990
    },
    {
      "epoch": 0.5476701200310347,
      "grad_norm": 3.375,
      "learning_rate": 9.326463876537983e-06,
      "loss": 1.1954,
      "step": 24000
    }
  ],
  "logging_steps": 10,
  "max_steps": 43822,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 2000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 6.209087320939823e+18,
  "train_batch_size": 10,
  "trial_name": null,
  "trial_params": null
}