{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 13863,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002164033758926639,
      "grad_norm": 1.1804725541377281,
      "learning_rate": 5e-06,
      "loss": 0.7931,
      "step": 10
    },
    {
      "epoch": 0.004328067517853278,
      "grad_norm": 1.0142507971826196,
      "learning_rate": 5e-06,
      "loss": 0.7085,
      "step": 20
    },
    {
      "epoch": 0.006492101276779918,
      "grad_norm": 1.0822486668830535,
      "learning_rate": 5e-06,
      "loss": 0.6971,
      "step": 30
    },
    {
      "epoch": 0.008656135035706556,
      "grad_norm": 0.9775955015066439,
      "learning_rate": 5e-06,
      "loss": 0.6931,
      "step": 40
    },
    {
      "epoch": 0.010820168794633196,
      "grad_norm": 1.0533969471863989,
      "learning_rate": 5e-06,
      "loss": 0.6748,
      "step": 50
    },
    {
      "epoch": 0.012984202553559835,
      "grad_norm": 0.9409532123493094,
      "learning_rate": 5e-06,
      "loss": 0.6582,
      "step": 60
    },
    {
      "epoch": 0.015148236312486475,
      "grad_norm": 0.8701409472453093,
      "learning_rate": 5e-06,
      "loss": 0.6612,
      "step": 70
    },
    {
      "epoch": 0.017312270071413113,
      "grad_norm": 0.8629522430209661,
      "learning_rate": 5e-06,
      "loss": 0.667,
      "step": 80
    },
    {
      "epoch": 0.019476303830339752,
      "grad_norm": 0.9005724399072088,
      "learning_rate": 5e-06,
      "loss": 0.6664,
      "step": 90
    },
    {
      "epoch": 0.02164033758926639,
      "grad_norm": 0.9031490400120155,
      "learning_rate": 5e-06,
      "loss": 0.6494,
      "step": 100
    },
    {
      "epoch": 0.02380437134819303,
      "grad_norm": 0.8491980446537303,
      "learning_rate": 5e-06,
      "loss": 0.6611,
      "step": 110
    },
    {
      "epoch": 0.02596840510711967,
      "grad_norm": 0.8544450914337819,
      "learning_rate": 5e-06,
      "loss": 0.6475,
      "step": 120
    },
    {
      "epoch": 0.02813243886604631,
      "grad_norm": 0.8940642349549588,
      "learning_rate": 5e-06,
      "loss": 0.6676,
      "step": 130
    },
    {
      "epoch": 0.03029647262497295,
      "grad_norm": 0.8624378682337847,
      "learning_rate": 5e-06,
      "loss": 0.6362,
      "step": 140
    },
    {
      "epoch": 0.032460506383899586,
      "grad_norm": 0.7937661473250804,
      "learning_rate": 5e-06,
      "loss": 0.6502,
      "step": 150
    },
    {
      "epoch": 0.034624540142826225,
      "grad_norm": 0.8511299785366766,
      "learning_rate": 5e-06,
      "loss": 0.6397,
      "step": 160
    },
    {
      "epoch": 0.036788573901752865,
      "grad_norm": 0.8247509378529784,
      "learning_rate": 5e-06,
      "loss": 0.6381,
      "step": 170
    },
    {
      "epoch": 0.038952607660679504,
      "grad_norm": 0.8606650790301921,
      "learning_rate": 5e-06,
      "loss": 0.6507,
      "step": 180
    },
    {
      "epoch": 0.041116641419606144,
      "grad_norm": 0.9000039308059945,
      "learning_rate": 5e-06,
      "loss": 0.6517,
      "step": 190
    },
    {
      "epoch": 0.04328067517853278,
      "grad_norm": 0.874046131263459,
      "learning_rate": 5e-06,
      "loss": 0.6372,
      "step": 200
    },
    {
      "epoch": 0.04544470893745942,
      "grad_norm": 0.826945338144314,
      "learning_rate": 5e-06,
      "loss": 0.6331,
      "step": 210
    },
    {
      "epoch": 0.04760874269638606,
      "grad_norm": 0.8647577576035517,
      "learning_rate": 5e-06,
      "loss": 0.6388,
      "step": 220
    },
    {
      "epoch": 0.0497727764553127,
      "grad_norm": 0.8357940501078124,
      "learning_rate": 5e-06,
      "loss": 0.6468,
      "step": 230
    },
    {
      "epoch": 0.05193681021423934,
      "grad_norm": 0.8205082815962632,
      "learning_rate": 5e-06,
      "loss": 0.6206,
      "step": 240
    },
    {
      "epoch": 0.05410084397316598,
      "grad_norm": 0.8029856499548153,
      "learning_rate": 5e-06,
      "loss": 0.6252,
      "step": 250
    },
    {
      "epoch": 0.05626487773209262,
      "grad_norm": 0.8266962137493432,
      "learning_rate": 5e-06,
      "loss": 0.6267,
      "step": 260
    },
    {
      "epoch": 0.05842891149101926,
      "grad_norm": 0.7760155385414025,
      "learning_rate": 5e-06,
      "loss": 0.6233,
      "step": 270
    },
    {
      "epoch": 0.0605929452499459,
      "grad_norm": 0.857439553904383,
      "learning_rate": 5e-06,
      "loss": 0.6247,
      "step": 280
    },
    {
      "epoch": 0.06275697900887253,
      "grad_norm": 0.8571645215071354,
      "learning_rate": 5e-06,
      "loss": 0.6291,
      "step": 290
    },
    {
      "epoch": 0.06492101276779917,
      "grad_norm": 0.8324700525072513,
      "learning_rate": 5e-06,
      "loss": 0.6313,
      "step": 300
    },
    {
      "epoch": 0.06708504652672581,
      "grad_norm": 0.7832243300229523,
      "learning_rate": 5e-06,
      "loss": 0.6329,
      "step": 310
    },
    {
      "epoch": 0.06924908028565245,
      "grad_norm": 0.7493686602077012,
      "learning_rate": 5e-06,
      "loss": 0.6198,
      "step": 320
    },
    {
      "epoch": 0.07141311404457909,
      "grad_norm": 0.8110396523299721,
      "learning_rate": 5e-06,
      "loss": 0.6268,
      "step": 330
    },
    {
      "epoch": 0.07357714780350573,
      "grad_norm": 0.7729143403225553,
      "learning_rate": 5e-06,
      "loss": 0.6131,
      "step": 340
    },
    {
      "epoch": 0.07574118156243237,
      "grad_norm": 0.844299608680804,
      "learning_rate": 5e-06,
      "loss": 0.628,
      "step": 350
    },
    {
      "epoch": 0.07790521532135901,
      "grad_norm": 0.8311658262532373,
      "learning_rate": 5e-06,
      "loss": 0.6284,
      "step": 360
    },
    {
      "epoch": 0.08006924908028565,
      "grad_norm": 0.7958716863199661,
      "learning_rate": 5e-06,
      "loss": 0.6166,
      "step": 370
    },
    {
      "epoch": 0.08223328283921229,
      "grad_norm": 0.8412185496400149,
      "learning_rate": 5e-06,
      "loss": 0.6458,
      "step": 380
    },
    {
      "epoch": 0.08439731659813893,
      "grad_norm": 0.7950152179064658,
      "learning_rate": 5e-06,
      "loss": 0.625,
      "step": 390
    },
    {
      "epoch": 0.08656135035706557,
      "grad_norm": 0.7728715222121862,
      "learning_rate": 5e-06,
      "loss": 0.6246,
      "step": 400
    },
    {
      "epoch": 0.0887253841159922,
      "grad_norm": 0.806073545694218,
      "learning_rate": 5e-06,
      "loss": 0.6315,
      "step": 410
    },
    {
      "epoch": 0.09088941787491885,
      "grad_norm": 0.8489148782720433,
      "learning_rate": 5e-06,
      "loss": 0.6145,
      "step": 420
    },
    {
      "epoch": 0.09305345163384549,
      "grad_norm": 0.7918376736911522,
      "learning_rate": 5e-06,
      "loss": 0.6192,
      "step": 430
    },
    {
      "epoch": 0.09521748539277212,
      "grad_norm": 0.8400111831435656,
      "learning_rate": 5e-06,
      "loss": 0.621,
      "step": 440
    },
    {
      "epoch": 0.09738151915169876,
      "grad_norm": 0.7614028740715489,
      "learning_rate": 5e-06,
      "loss": 0.605,
      "step": 450
    },
    {
      "epoch": 0.0995455529106254,
      "grad_norm": 0.8577505860556417,
      "learning_rate": 5e-06,
      "loss": 0.607,
      "step": 460
    },
    {
      "epoch": 0.10170958666955204,
      "grad_norm": 0.8729565243092611,
      "learning_rate": 5e-06,
      "loss": 0.6229,
      "step": 470
    },
    {
      "epoch": 0.10387362042847868,
      "grad_norm": 0.822547387381443,
      "learning_rate": 5e-06,
      "loss": 0.6213,
      "step": 480
    },
    {
      "epoch": 0.10603765418740532,
      "grad_norm": 0.7797820836510193,
      "learning_rate": 5e-06,
      "loss": 0.6055,
      "step": 490
    },
    {
      "epoch": 0.10820168794633196,
      "grad_norm": 0.8039389365692253,
      "learning_rate": 5e-06,
      "loss": 0.6158,
      "step": 500
    },
    {
      "epoch": 0.1103657217052586,
      "grad_norm": 0.8014536678260006,
      "learning_rate": 5e-06,
      "loss": 0.6078,
      "step": 510
    },
    {
      "epoch": 0.11252975546418524,
      "grad_norm": 0.8655111256073293,
      "learning_rate": 5e-06,
      "loss": 0.6279,
      "step": 520
    },
    {
      "epoch": 0.11469378922311188,
      "grad_norm": 0.872957811479719,
      "learning_rate": 5e-06,
      "loss": 0.6172,
      "step": 530
    },
    {
      "epoch": 0.11685782298203852,
      "grad_norm": 0.8819039368989797,
      "learning_rate": 5e-06,
      "loss": 0.6121,
      "step": 540
    },
    {
      "epoch": 0.11902185674096516,
      "grad_norm": 0.8640564878023476,
      "learning_rate": 5e-06,
      "loss": 0.5962,
      "step": 550
    },
    {
      "epoch": 0.1211858904998918,
      "grad_norm": 0.8207671083993288,
      "learning_rate": 5e-06,
      "loss": 0.6061,
      "step": 560
    },
    {
      "epoch": 0.12334992425881844,
      "grad_norm": 0.8639259231144906,
      "learning_rate": 5e-06,
      "loss": 0.6102,
      "step": 570
    },
    {
      "epoch": 0.12551395801774506,
      "grad_norm": 0.8252682175137697,
      "learning_rate": 5e-06,
      "loss": 0.5919,
      "step": 580
    },
    {
      "epoch": 0.12767799177667172,
      "grad_norm": 0.8178974092958998,
      "learning_rate": 5e-06,
      "loss": 0.6166,
      "step": 590
    },
    {
      "epoch": 0.12984202553559834,
      "grad_norm": 0.8895228285922248,
      "learning_rate": 5e-06,
      "loss": 0.6098,
      "step": 600
    },
    {
      "epoch": 0.132006059294525,
      "grad_norm": 0.7969967604154639,
      "learning_rate": 5e-06,
      "loss": 0.6077,
      "step": 610
    },
    {
      "epoch": 0.13417009305345162,
      "grad_norm": 0.8683433820973785,
      "learning_rate": 5e-06,
      "loss": 0.6099,
      "step": 620
    },
    {
      "epoch": 0.13633412681237828,
      "grad_norm": 0.8063824032551594,
      "learning_rate": 5e-06,
      "loss": 0.6127,
      "step": 630
    },
    {
      "epoch": 0.1384981605713049,
      "grad_norm": 0.8136662571064663,
      "learning_rate": 5e-06,
      "loss": 0.5968,
      "step": 640
    },
    {
      "epoch": 0.14066219433023155,
      "grad_norm": 0.822893712320517,
      "learning_rate": 5e-06,
      "loss": 0.6038,
      "step": 650
    },
    {
      "epoch": 0.14282622808915818,
      "grad_norm": 0.8485980406618856,
      "learning_rate": 5e-06,
      "loss": 0.6127,
      "step": 660
    },
    {
      "epoch": 0.14499026184808483,
      "grad_norm": 0.8396390754266708,
      "learning_rate": 5e-06,
      "loss": 0.6105,
      "step": 670
    },
    {
      "epoch": 0.14715429560701146,
      "grad_norm": 0.863650354540203,
      "learning_rate": 5e-06,
      "loss": 0.5947,
      "step": 680
    },
    {
      "epoch": 0.1493183293659381,
      "grad_norm": 0.8767137343250653,
      "learning_rate": 5e-06,
      "loss": 0.5944,
      "step": 690
    },
    {
      "epoch": 0.15148236312486474,
      "grad_norm": 0.7785697177833184,
      "learning_rate": 5e-06,
      "loss": 0.5957,
      "step": 700
    },
    {
      "epoch": 0.1536463968837914,
      "grad_norm": 0.7903385744648517,
      "learning_rate": 5e-06,
      "loss": 0.5887,
      "step": 710
    },
    {
      "epoch": 0.15581043064271802,
      "grad_norm": 0.8304334914378555,
      "learning_rate": 5e-06,
      "loss": 0.5972,
      "step": 720
    },
    {
      "epoch": 0.15797446440164467,
      "grad_norm": 0.8032146774897264,
      "learning_rate": 5e-06,
      "loss": 0.6055,
      "step": 730
    },
    {
      "epoch": 0.1601384981605713,
      "grad_norm": 0.818462114006868,
      "learning_rate": 5e-06,
      "loss": 0.6086,
      "step": 740
    },
    {
      "epoch": 0.16230253191949795,
      "grad_norm": 0.7999227700105283,
      "learning_rate": 5e-06,
      "loss": 0.6,
      "step": 750
    },
    {
      "epoch": 0.16446656567842458,
      "grad_norm": 0.779725395098173,
      "learning_rate": 5e-06,
      "loss": 0.581,
      "step": 760
    },
    {
      "epoch": 0.16663059943735123,
      "grad_norm": 0.8501680445359217,
      "learning_rate": 5e-06,
      "loss": 0.6034,
      "step": 770
    },
    {
      "epoch": 0.16879463319627785,
      "grad_norm": 0.7955262980950765,
      "learning_rate": 5e-06,
      "loss": 0.591,
      "step": 780
    },
    {
      "epoch": 0.1709586669552045,
      "grad_norm": 0.7969375284741476,
      "learning_rate": 5e-06,
      "loss": 0.5917,
      "step": 790
    },
    {
      "epoch": 0.17312270071413113,
      "grad_norm": 0.7651500787357158,
      "learning_rate": 5e-06,
      "loss": 0.5925,
      "step": 800
    },
    {
      "epoch": 0.1752867344730578,
      "grad_norm": 0.7869498336245857,
      "learning_rate": 5e-06,
      "loss": 0.5919,
      "step": 810
    },
    {
      "epoch": 0.1774507682319844,
      "grad_norm": 0.8171995660498751,
      "learning_rate": 5e-06,
      "loss": 0.5938,
      "step": 820
    },
    {
      "epoch": 0.17961480199091107,
      "grad_norm": 0.7836587740083709,
      "learning_rate": 5e-06,
      "loss": 0.5786,
      "step": 830
    },
    {
      "epoch": 0.1817788357498377,
      "grad_norm": 0.8784878650037878,
      "learning_rate": 5e-06,
      "loss": 0.6044,
      "step": 840
    },
    {
      "epoch": 0.18394286950876435,
      "grad_norm": 0.800243869063219,
      "learning_rate": 5e-06,
      "loss": 0.5889,
      "step": 850
    },
    {
      "epoch": 0.18610690326769097,
      "grad_norm": 0.8018420320157924,
      "learning_rate": 5e-06,
      "loss": 0.6087,
      "step": 860
    },
    {
      "epoch": 0.18827093702661762,
      "grad_norm": 0.8180751714275638,
      "learning_rate": 5e-06,
      "loss": 0.6037,
      "step": 870
    },
    {
      "epoch": 0.19043497078554425,
      "grad_norm": 0.785600856617542,
      "learning_rate": 5e-06,
      "loss": 0.5837,
      "step": 880
    },
    {
      "epoch": 0.1925990045444709,
      "grad_norm": 0.8412200127939977,
      "learning_rate": 5e-06,
      "loss": 0.5896,
      "step": 890
    },
    {
      "epoch": 0.19476303830339753,
      "grad_norm": 0.8025538669907342,
      "learning_rate": 5e-06,
      "loss": 0.5991,
      "step": 900
    },
    {
      "epoch": 0.19692707206232418,
      "grad_norm": 0.7720798100780025,
      "learning_rate": 5e-06,
      "loss": 0.5891,
      "step": 910
    },
    {
      "epoch": 0.1990911058212508,
      "grad_norm": 0.8661764118040983,
      "learning_rate": 5e-06,
      "loss": 0.5931,
      "step": 920
    },
    {
      "epoch": 0.20125513958017746,
      "grad_norm": 0.8120743515574185,
      "learning_rate": 5e-06,
      "loss": 0.5957,
      "step": 930
    },
    {
      "epoch": 0.2034191733391041,
      "grad_norm": 0.8345160783897795,
      "learning_rate": 5e-06,
      "loss": 0.5984,
      "step": 940
    },
    {
      "epoch": 0.20558320709803074,
      "grad_norm": 0.811502391003602,
      "learning_rate": 5e-06,
      "loss": 0.6081,
      "step": 950
    },
    {
      "epoch": 0.20774724085695737,
      "grad_norm": 0.7825461582026372,
      "learning_rate": 5e-06,
      "loss": 0.5948,
      "step": 960
    },
    {
      "epoch": 0.20991127461588402,
      "grad_norm": 0.8055902800373363,
      "learning_rate": 5e-06,
      "loss": 0.5982,
      "step": 970
    },
    {
      "epoch": 0.21207530837481064,
      "grad_norm": 0.8388084051375468,
      "learning_rate": 5e-06,
      "loss": 0.5893,
      "step": 980
    },
    {
      "epoch": 0.2142393421337373,
      "grad_norm": 0.7679133259037194,
      "learning_rate": 5e-06,
      "loss": 0.5907,
      "step": 990
    },
    {
      "epoch": 0.21640337589266392,
      "grad_norm": 0.8624733193059415,
      "learning_rate": 5e-06,
      "loss": 0.5814,
      "step": 1000
    },
    {
      "epoch": 0.21856740965159058,
      "grad_norm": 0.7984486425858306,
      "learning_rate": 5e-06,
      "loss": 0.5913,
      "step": 1010
    },
    {
      "epoch": 0.2207314434105172,
      "grad_norm": 0.8510028087137639,
      "learning_rate": 5e-06,
      "loss": 0.5949,
      "step": 1020
    },
    {
      "epoch": 0.22289547716944386,
      "grad_norm": 0.7996439613799075,
      "learning_rate": 5e-06,
      "loss": 0.601,
      "step": 1030
    },
    {
      "epoch": 0.22505951092837048,
      "grad_norm": 0.8088661553876958,
      "learning_rate": 5e-06,
      "loss": 0.6074,
      "step": 1040
    },
    {
      "epoch": 0.22722354468729714,
      "grad_norm": 0.8163892550046303,
      "learning_rate": 5e-06,
      "loss": 0.5863,
      "step": 1050
    },
    {
      "epoch": 0.22938757844622376,
      "grad_norm": 0.9418880337035802,
      "learning_rate": 5e-06,
      "loss": 0.6033,
      "step": 1060
    },
    {
      "epoch": 0.2315516122051504,
      "grad_norm": 0.8308760617654204,
      "learning_rate": 5e-06,
      "loss": 0.603,
      "step": 1070
    },
    {
      "epoch": 0.23371564596407704,
      "grad_norm": 0.8145198410135479,
      "learning_rate": 5e-06,
      "loss": 0.5947,
      "step": 1080
    },
    {
      "epoch": 0.23587967972300367,
      "grad_norm": 0.8623357006048706,
      "learning_rate": 5e-06,
      "loss": 0.6056,
      "step": 1090
    },
    {
      "epoch": 0.23804371348193032,
      "grad_norm": 0.7965551896548404,
      "learning_rate": 5e-06,
      "loss": 0.5923,
      "step": 1100
    },
    {
      "epoch": 0.24020774724085694,
      "grad_norm": 0.8052327006114539,
      "learning_rate": 5e-06,
      "loss": 0.6007,
      "step": 1110
    },
    {
      "epoch": 0.2423717809997836,
      "grad_norm": 0.8213993699298745,
      "learning_rate": 5e-06,
      "loss": 0.5797,
      "step": 1120
    },
    {
      "epoch": 0.24453581475871022,
      "grad_norm": 0.8149866187723401,
      "learning_rate": 5e-06,
      "loss": 0.5941,
      "step": 1130
    },
    {
      "epoch": 0.24669984851763688,
      "grad_norm": 0.7837216156150254,
      "learning_rate": 5e-06,
      "loss": 0.59,
      "step": 1140
    },
    {
      "epoch": 0.2488638822765635,
      "grad_norm": 0.8126197328739624,
      "learning_rate": 5e-06,
      "loss": 0.5952,
      "step": 1150
    },
    {
      "epoch": 0.25102791603549013,
      "grad_norm": 0.8221669781147581,
      "learning_rate": 5e-06,
      "loss": 0.6005,
      "step": 1160
    },
    {
      "epoch": 0.2531919497944168,
      "grad_norm": 0.7963300719869713,
      "learning_rate": 5e-06,
      "loss": 0.5877,
      "step": 1170
    },
    {
      "epoch": 0.25535598355334344,
      "grad_norm": 0.788058435618901,
      "learning_rate": 5e-06,
      "loss": 0.591,
      "step": 1180
    },
    {
      "epoch": 0.2575200173122701,
      "grad_norm": 0.7519308880120247,
      "learning_rate": 5e-06,
      "loss": 0.5928,
      "step": 1190
    },
    {
      "epoch": 0.2596840510711967,
      "grad_norm": 0.8266442618730233,
      "learning_rate": 5e-06,
      "loss": 0.5839,
      "step": 1200
    },
    {
      "epoch": 0.26184808483012334,
      "grad_norm": 0.8292557026765464,
      "learning_rate": 5e-06,
      "loss": 0.5865,
      "step": 1210
    },
    {
      "epoch": 0.26401211858905,
      "grad_norm": 0.8853558777310622,
      "learning_rate": 5e-06,
      "loss": 0.5983,
      "step": 1220
    },
    {
      "epoch": 0.26617615234797665,
      "grad_norm": 0.9109356382425066,
      "learning_rate": 5e-06,
      "loss": 0.5977,
      "step": 1230
    },
    {
      "epoch": 0.26834018610690324,
      "grad_norm": 0.89504318082267,
      "learning_rate": 5e-06,
      "loss": 0.592,
      "step": 1240
    },
    {
      "epoch": 0.2705042198658299,
      "grad_norm": 0.8160219829275853,
      "learning_rate": 5e-06,
      "loss": 0.593,
      "step": 1250
    },
    {
      "epoch": 0.27266825362475655,
      "grad_norm": 0.8076508336570862,
      "learning_rate": 5e-06,
      "loss": 0.5856,
      "step": 1260
    },
    {
      "epoch": 0.2748322873836832,
      "grad_norm": 0.8022139372756553,
      "learning_rate": 5e-06,
      "loss": 0.5747,
      "step": 1270
    },
    {
      "epoch": 0.2769963211426098,
      "grad_norm": 0.8145806416794947,
      "learning_rate": 5e-06,
      "loss": 0.5887,
      "step": 1280
    },
    {
      "epoch": 0.27916035490153646,
      "grad_norm": 0.7591150136319158,
      "learning_rate": 5e-06,
      "loss": 0.5732,
      "step": 1290
    },
    {
      "epoch": 0.2813243886604631,
      "grad_norm": 0.8140968921103322,
      "learning_rate": 5e-06,
      "loss": 0.5912,
      "step": 1300
    },
    {
      "epoch": 0.28348842241938976,
      "grad_norm": 0.7944048590141184,
      "learning_rate": 5e-06,
      "loss": 0.5895,
      "step": 1310
    },
    {
      "epoch": 0.28565245617831636,
      "grad_norm": 0.8789105404070925,
      "learning_rate": 5e-06,
      "loss": 0.5747,
      "step": 1320
    },
    {
      "epoch": 0.287816489937243,
      "grad_norm": 0.7796055521428117,
      "learning_rate": 5e-06,
      "loss": 0.5881,
      "step": 1330
    },
    {
      "epoch": 0.28998052369616967,
      "grad_norm": 0.817568265263925,
      "learning_rate": 5e-06,
      "loss": 0.5818,
      "step": 1340
    },
    {
      "epoch": 0.2921445574550963,
      "grad_norm": 0.7781293420250672,
      "learning_rate": 5e-06,
      "loss": 0.5912,
      "step": 1350
    },
    {
      "epoch": 0.2943085912140229,
      "grad_norm": 0.8504639244998208,
      "learning_rate": 5e-06,
      "loss": 0.5914,
      "step": 1360
    },
    {
      "epoch": 0.29647262497294957,
      "grad_norm": 0.848929905244684,
      "learning_rate": 5e-06,
      "loss": 0.5705,
      "step": 1370
    },
    {
      "epoch": 0.2986366587318762,
      "grad_norm": 0.8401835594137422,
      "learning_rate": 5e-06,
      "loss": 0.5859,
      "step": 1380
    },
    {
      "epoch": 0.3008006924908029,
      "grad_norm": 0.8430038720121172,
      "learning_rate": 5e-06,
      "loss": 0.5909,
      "step": 1390
    },
    {
      "epoch": 0.3029647262497295,
      "grad_norm": 0.7925617711932909,
      "learning_rate": 5e-06,
      "loss": 0.5822,
      "step": 1400
    },
    {
      "epoch": 0.30512876000865613,
      "grad_norm": 0.8081455173268478,
      "learning_rate": 5e-06,
      "loss": 0.5745,
      "step": 1410
    },
    {
      "epoch": 0.3072927937675828,
      "grad_norm": 0.7903132825985777,
      "learning_rate": 5e-06,
      "loss": 0.5759,
      "step": 1420
    },
    {
      "epoch": 0.30945682752650944,
      "grad_norm": 0.8243849826447428,
      "learning_rate": 5e-06,
      "loss": 0.5968,
      "step": 1430
    },
    {
      "epoch": 0.31162086128543603,
      "grad_norm": 0.7603966355823882,
      "learning_rate": 5e-06,
      "loss": 0.5873,
      "step": 1440
    },
    {
      "epoch": 0.3137848950443627,
      "grad_norm": 0.801455870233685,
      "learning_rate": 5e-06,
      "loss": 0.5786,
      "step": 1450
    },
    {
      "epoch": 0.31594892880328934,
      "grad_norm": 0.7686291740407882,
      "learning_rate": 5e-06,
      "loss": 0.5789,
      "step": 1460
    },
    {
      "epoch": 0.318112962562216,
      "grad_norm": 0.8554910506261316,
      "learning_rate": 5e-06,
      "loss": 0.5795,
      "step": 1470
    },
    {
      "epoch": 0.3202769963211426,
      "grad_norm": 0.820274584453091,
      "learning_rate": 5e-06,
      "loss": 0.5756,
      "step": 1480
    },
    {
      "epoch": 0.32244103008006925,
      "grad_norm": 0.835330743611247,
      "learning_rate": 5e-06,
      "loss": 0.5848,
      "step": 1490
    },
    {
      "epoch": 0.3246050638389959,
      "grad_norm": 0.8067787217591124,
      "learning_rate": 5e-06,
      "loss": 0.5651,
      "step": 1500
    },
    {
      "epoch": 0.32676909759792255,
      "grad_norm": 0.8055298656551628,
      "learning_rate": 5e-06,
      "loss": 0.5806,
      "step": 1510
    },
    {
      "epoch": 0.32893313135684915,
      "grad_norm": 0.7643337696325749,
      "learning_rate": 5e-06,
      "loss": 0.5836,
      "step": 1520
    },
    {
      "epoch": 0.3310971651157758,
      "grad_norm": 0.802754989395744,
      "learning_rate": 5e-06,
      "loss": 0.577,
      "step": 1530
    },
    {
      "epoch": 0.33326119887470246,
      "grad_norm": 0.8227460460988874,
      "learning_rate": 5e-06,
      "loss": 0.5781,
      "step": 1540
    },
    {
      "epoch": 0.3354252326336291,
      "grad_norm": 0.8157555450231291,
      "learning_rate": 5e-06,
      "loss": 0.5915,
      "step": 1550
    },
    {
      "epoch": 0.3375892663925557,
      "grad_norm": 0.8178511979744351,
      "learning_rate": 5e-06,
      "loss": 0.5844,
      "step": 1560
    },
    {
      "epoch": 0.33975330015148236,
      "grad_norm": 0.8640976250921749,
      "learning_rate": 5e-06,
      "loss": 0.5867,
      "step": 1570
    },
    {
      "epoch": 0.341917333910409,
      "grad_norm": 0.8225622251467227,
      "learning_rate": 5e-06,
      "loss": 0.5626,
      "step": 1580
    },
    {
      "epoch": 0.34408136766933567,
      "grad_norm": 0.809101202716206,
      "learning_rate": 5e-06,
      "loss": 0.5841,
      "step": 1590
    },
    {
      "epoch": 0.34624540142826227,
      "grad_norm": 0.8007366862509202,
      "learning_rate": 5e-06,
      "loss": 0.6002,
      "step": 1600
    },
    {
      "epoch": 0.3484094351871889,
      "grad_norm": 0.80921682472726,
      "learning_rate": 5e-06,
      "loss": 0.5756,
      "step": 1610
    },
    {
      "epoch": 0.3505734689461156,
      "grad_norm": 0.7919027933880338,
      "learning_rate": 5e-06,
      "loss": 0.5811,
      "step": 1620
    },
    {
      "epoch": 0.35273750270504217,
      "grad_norm": 0.8578709720525358,
      "learning_rate": 5e-06,
      "loss": 0.5844,
      "step": 1630
    },
    {
      "epoch": 0.3549015364639688,
      "grad_norm": 0.8919305010438147,
      "learning_rate": 5e-06,
      "loss": 0.5867,
      "step": 1640
    },
    {
      "epoch": 0.3570655702228955,
      "grad_norm": 0.8360856901044694,
      "learning_rate": 5e-06,
      "loss": 0.5959,
      "step": 1650
    },
    {
      "epoch": 0.35922960398182213,
      "grad_norm": 0.8675129695665484,
      "learning_rate": 5e-06,
      "loss": 0.5761,
      "step": 1660
    },
    {
      "epoch": 0.36139363774074873,
      "grad_norm": 0.7689087827995047,
      "learning_rate": 5e-06,
      "loss": 0.5861,
      "step": 1670
    },
    {
      "epoch": 0.3635576714996754,
      "grad_norm": 0.7720960735276896,
      "learning_rate": 5e-06,
      "loss": 0.5743,
      "step": 1680
    },
    {
      "epoch": 0.36572170525860204,
      "grad_norm": 0.7721866885573503,
      "learning_rate": 5e-06,
      "loss": 0.59,
      "step": 1690
    },
    {
      "epoch": 0.3678857390175287,
      "grad_norm": 0.7882290467361662,
      "learning_rate": 5e-06,
      "loss": 0.5755,
      "step": 1700
    },
    {
      "epoch": 0.3700497727764553,
      "grad_norm": 0.7556952201847303,
      "learning_rate": 5e-06,
      "loss": 0.5665,
      "step": 1710
    },
    {
      "epoch": 0.37221380653538194,
      "grad_norm": 0.791031843589153,
      "learning_rate": 5e-06,
      "loss": 0.5771,
      "step": 1720
    },
    {
      "epoch": 0.3743778402943086,
      "grad_norm": 0.7809696368341134,
      "learning_rate": 5e-06,
      "loss": 0.5653,
      "step": 1730
    },
    {
      "epoch": 0.37654187405323525,
      "grad_norm": 0.8629957045419663,
      "learning_rate": 5e-06,
      "loss": 0.5744,
      "step": 1740
    },
    {
      "epoch": 0.37870590781216185,
      "grad_norm": 0.735264733320356,
      "learning_rate": 5e-06,
      "loss": 0.568,
      "step": 1750
    },
    {
      "epoch": 0.3808699415710885,
      "grad_norm": 0.79820790046158,
      "learning_rate": 5e-06,
      "loss": 0.5756,
      "step": 1760
    },
    {
      "epoch": 0.38303397533001515,
      "grad_norm": 0.7990867854378455,
      "learning_rate": 5e-06,
      "loss": 0.5741,
      "step": 1770
    },
    {
      "epoch": 0.3851980090889418,
      "grad_norm": 0.8121981810970615,
      "learning_rate": 5e-06,
      "loss": 0.5784,
      "step": 1780
    },
    {
      "epoch": 0.3873620428478684,
      "grad_norm": 0.8133852852205034,
      "learning_rate": 5e-06,
      "loss": 0.5843,
      "step": 1790
    },
    {
      "epoch": 0.38952607660679506,
      "grad_norm": 0.8244168262600475,
      "learning_rate": 5e-06,
      "loss": 0.5711,
      "step": 1800
    },
    {
      "epoch": 0.3916901103657217,
      "grad_norm": 0.7650337026676514,
      "learning_rate": 5e-06,
      "loss": 0.567,
      "step": 1810
    },
    {
      "epoch": 0.39385414412464836,
      "grad_norm": 0.7884841014495525,
      "learning_rate": 5e-06,
      "loss": 0.587,
      "step": 1820
    },
    {
      "epoch": 0.39601817788357496,
      "grad_norm": 0.8218222575278361,
      "learning_rate": 5e-06,
      "loss": 0.5729,
      "step": 1830
    },
    {
      "epoch": 0.3981822116425016,
      "grad_norm": 0.7928094425107333,
      "learning_rate": 5e-06,
      "loss": 0.5677,
      "step": 1840
    },
    {
      "epoch": 0.40034624540142827,
      "grad_norm": 0.7768628011319236,
      "learning_rate": 5e-06,
      "loss": 0.5914,
      "step": 1850
    },
    {
      "epoch": 0.4025102791603549,
      "grad_norm": 0.7947010305390309,
      "learning_rate": 5e-06,
      "loss": 0.5748,
      "step": 1860
    },
    {
      "epoch": 0.4046743129192815,
      "grad_norm": 0.7697471130550405,
      "learning_rate": 5e-06,
      "loss": 0.5759,
      "step": 1870
    },
    {
      "epoch": 0.4068383466782082,
      "grad_norm": 0.796128902115479,
      "learning_rate": 5e-06,
      "loss": 0.5734,
      "step": 1880
    },
    {
      "epoch": 0.4090023804371348,
      "grad_norm": 0.7659679821804417,
      "learning_rate": 5e-06,
      "loss": 0.5631,
      "step": 1890
    },
    {
      "epoch": 0.4111664141960615,
      "grad_norm": 0.8580458852557561,
      "learning_rate": 5e-06,
      "loss": 0.5765,
      "step": 1900
    },
    {
      "epoch": 0.4133304479549881,
      "grad_norm": 0.7957186485957107,
      "learning_rate": 5e-06,
      "loss": 0.5658,
      "step": 1910
    },
    {
      "epoch": 0.41549448171391473,
      "grad_norm": 0.7853295564565586,
      "learning_rate": 5e-06,
      "loss": 0.5621,
      "step": 1920
    },
    {
      "epoch": 0.4176585154728414,
      "grad_norm": 0.720431628977774,
      "learning_rate": 5e-06,
      "loss": 0.565,
      "step": 1930
    },
    {
      "epoch": 0.41982254923176804,
      "grad_norm": 0.7826448016952756,
      "learning_rate": 5e-06,
      "loss": 0.5677,
      "step": 1940
    },
    {
      "epoch": 0.42198658299069464,
      "grad_norm": 0.7904824818785057,
      "learning_rate": 5e-06,
      "loss": 0.567,
      "step": 1950
    },
    {
      "epoch": 0.4241506167496213,
      "grad_norm": 0.7871987194350701,
      "learning_rate": 5e-06,
      "loss": 0.5812,
      "step": 1960
    },
    {
      "epoch": 0.42631465050854794,
      "grad_norm": 0.7967155088578817,
      "learning_rate": 5e-06,
      "loss": 0.559,
      "step": 1970
    },
    {
      "epoch": 0.4284786842674746,
      "grad_norm": 0.8505915926157366,
      "learning_rate": 5e-06,
      "loss": 0.5757,
      "step": 1980
    },
    {
      "epoch": 0.4306427180264012,
      "grad_norm": 0.7667510030206882,
      "learning_rate": 5e-06,
      "loss": 0.557,
      "step": 1990
    },
    {
      "epoch": 0.43280675178532785,
      "grad_norm": 0.8161411756414166,
      "learning_rate": 5e-06,
      "loss": 0.5886,
      "step": 2000
    },
    {
      "epoch": 0.4349707855442545,
      "grad_norm": 0.8557821729072761,
      "learning_rate": 5e-06,
      "loss": 0.5815,
      "step": 2010
    },
    {
      "epoch": 0.43713481930318115,
      "grad_norm": 0.7980272351612475,
      "learning_rate": 5e-06,
      "loss": 0.5782,
      "step": 2020
    },
    {
      "epoch": 0.43929885306210775,
      "grad_norm": 0.7855484364592483,
      "learning_rate": 5e-06,
      "loss": 0.565,
      "step": 2030
    },
    {
      "epoch": 0.4414628868210344,
      "grad_norm": 0.742503374275201,
      "learning_rate": 5e-06,
      "loss": 0.5726,
      "step": 2040
    },
    {
      "epoch": 0.44362692057996106,
      "grad_norm": 0.8265176138612144,
      "learning_rate": 5e-06,
      "loss": 0.565,
      "step": 2050
    },
    {
      "epoch": 0.4457909543388877,
      "grad_norm": 0.7910125817205834,
      "learning_rate": 5e-06,
      "loss": 0.5738,
      "step": 2060
    },
    {
      "epoch": 0.4479549880978143,
      "grad_norm": 0.7624732216492297,
      "learning_rate": 5e-06,
      "loss": 0.5539,
      "step": 2070
    },
    {
      "epoch": 0.45011902185674096,
      "grad_norm": 0.8024443330805618,
      "learning_rate": 5e-06,
      "loss": 0.5684,
      "step": 2080
    },
    {
      "epoch": 0.4522830556156676,
      "grad_norm": 0.8252242841372383,
      "learning_rate": 5e-06,
      "loss": 0.5638,
      "step": 2090
    },
    {
      "epoch": 0.45444708937459427,
      "grad_norm": 0.7918799931003818,
      "learning_rate": 5e-06,
      "loss": 0.5656,
      "step": 2100
    },
    {
      "epoch": 0.45661112313352087,
      "grad_norm": 0.8503297988500775,
      "learning_rate": 5e-06,
      "loss": 0.5878,
      "step": 2110
    },
    {
      "epoch": 0.4587751568924475,
      "grad_norm": 0.7881054994019235,
      "learning_rate": 5e-06,
      "loss": 0.5688,
      "step": 2120
    },
    {
      "epoch": 0.4609391906513742,
      "grad_norm": 0.8298283284928384,
      "learning_rate": 5e-06,
      "loss": 0.5662,
      "step": 2130
    },
    {
      "epoch": 0.4631032244103008,
      "grad_norm": 2.6984894013557374,
      "learning_rate": 5e-06,
      "loss": 0.5691,
      "step": 2140
    },
    {
      "epoch": 0.4652672581692274,
      "grad_norm": 0.8293615927885423,
      "learning_rate": 5e-06,
      "loss": 0.5624,
      "step": 2150
    },
    {
      "epoch": 0.4674312919281541,
      "grad_norm": 0.7908261626595905,
      "learning_rate": 5e-06,
      "loss": 0.5601,
      "step": 2160
    },
    {
      "epoch": 0.46959532568708073,
      "grad_norm": 0.8407395655282288,
      "learning_rate": 5e-06,
      "loss": 0.5789,
      "step": 2170
    },
    {
      "epoch": 0.47175935944600733,
      "grad_norm": 0.7855312267072224,
      "learning_rate": 5e-06,
      "loss": 0.5592,
      "step": 2180
    },
    {
      "epoch": 0.473923393204934,
      "grad_norm": 0.8177276022998019,
      "learning_rate": 5e-06,
      "loss": 0.5642,
      "step": 2190
    },
    {
      "epoch": 0.47608742696386064,
      "grad_norm": 0.7674903085772374,
      "learning_rate": 5e-06,
      "loss": 0.559,
      "step": 2200
    },
    {
      "epoch": 0.4782514607227873,
      "grad_norm": 0.8138473478647174,
      "learning_rate": 5e-06,
      "loss": 0.5744,
      "step": 2210
    },
    {
      "epoch": 0.4804154944817139,
      "grad_norm": 0.7985572678524506,
      "learning_rate": 5e-06,
      "loss": 0.5608,
      "step": 2220
    },
    {
      "epoch": 0.48257952824064054,
      "grad_norm": 0.7978238400519106,
      "learning_rate": 5e-06,
      "loss": 0.56,
      "step": 2230
    },
    {
      "epoch": 0.4847435619995672,
      "grad_norm": 0.786485961749514,
      "learning_rate": 5e-06,
      "loss": 0.5806,
      "step": 2240
    },
    {
      "epoch": 0.48690759575849385,
      "grad_norm": 0.7835696169083621,
      "learning_rate": 5e-06,
      "loss": 0.5813,
      "step": 2250
    },
    {
      "epoch": 0.48907162951742045,
      "grad_norm": 0.8038286577487808,
      "learning_rate": 5e-06,
      "loss": 0.5787,
      "step": 2260
    },
    {
      "epoch": 0.4912356632763471,
      "grad_norm": 0.7949579187720448,
      "learning_rate": 5e-06,
      "loss": 0.5603,
      "step": 2270
    },
    {
      "epoch": 0.49339969703527375,
      "grad_norm": 0.7996876857887395,
      "learning_rate": 5e-06,
      "loss": 0.564,
      "step": 2280
    },
    {
      "epoch": 0.4955637307942004,
      "grad_norm": 0.7523979963095976,
      "learning_rate": 5e-06,
      "loss": 0.572,
      "step": 2290
    },
    {
      "epoch": 0.497727764553127,
      "grad_norm": 0.7799060855112318,
      "learning_rate": 5e-06,
      "loss": 0.5512,
      "step": 2300
    },
    {
      "epoch": 0.49989179831205366,
      "grad_norm": 0.8354130726326037,
      "learning_rate": 5e-06,
      "loss": 0.5672,
      "step": 2310
    },
    {
      "epoch": 0.5020558320709803,
      "grad_norm": 0.8174289167975042,
      "learning_rate": 5e-06,
      "loss": 0.5745,
      "step": 2320
    },
    {
      "epoch": 0.504219865829907,
      "grad_norm": 0.7797704636452041,
      "learning_rate": 5e-06,
      "loss": 0.5785,
      "step": 2330
    },
    {
      "epoch": 0.5063838995888336,
      "grad_norm": 0.8049184278922541,
      "learning_rate": 5e-06,
      "loss": 0.5663,
      "step": 2340
    },
    {
      "epoch": 0.5085479333477603,
      "grad_norm": 0.79714737630827,
      "learning_rate": 5e-06,
      "loss": 0.5696,
      "step": 2350
    },
    {
      "epoch": 0.5107119671066869,
      "grad_norm": 0.8079273656180676,
      "learning_rate": 5e-06,
      "loss": 0.5611,
      "step": 2360
    },
    {
      "epoch": 0.5128760008656135,
      "grad_norm": 0.7996864982834264,
      "learning_rate": 5e-06,
      "loss": 0.5634,
      "step": 2370
    },
    {
      "epoch": 0.5150400346245402,
      "grad_norm": 0.7392249752105758,
      "learning_rate": 5e-06,
      "loss": 0.5441,
      "step": 2380
    },
    {
      "epoch": 0.5172040683834668,
      "grad_norm": 0.8235210489123114,
      "learning_rate": 5e-06,
      "loss": 0.5616,
      "step": 2390
    },
    {
      "epoch": 0.5193681021423934,
      "grad_norm": 0.7859990977815138,
      "learning_rate": 5e-06,
      "loss": 0.5738,
      "step": 2400
    },
    {
      "epoch": 0.5215321359013201,
      "grad_norm": 0.8006842059420661,
      "learning_rate": 5e-06,
      "loss": 0.5669,
      "step": 2410
    },
    {
      "epoch": 0.5236961696602467,
      "grad_norm": 0.8301226308008953,
      "learning_rate": 5e-06,
      "loss": 0.5585,
      "step": 2420
    },
    {
      "epoch": 0.5258602034191734,
      "grad_norm": 0.8116242656076512,
      "learning_rate": 5e-06,
      "loss": 0.5575,
      "step": 2430
    },
    {
      "epoch": 0.5280242371781,
      "grad_norm": 0.7951327769075789,
      "learning_rate": 5e-06,
      "loss": 0.5623,
      "step": 2440
    },
    {
      "epoch": 0.5301882709370266,
      "grad_norm": 0.7632924959396983,
      "learning_rate": 5e-06,
      "loss": 0.5557,
      "step": 2450
    },
    {
      "epoch": 0.5323523046959533,
      "grad_norm": 0.7703765895080564,
      "learning_rate": 5e-06,
      "loss": 0.5674,
      "step": 2460
    },
    {
      "epoch": 0.5345163384548799,
      "grad_norm": 0.7519955078038866,
      "learning_rate": 5e-06,
      "loss": 0.5517,
      "step": 2470
    },
    {
      "epoch": 0.5366803722138065,
      "grad_norm": 0.777945289031897,
      "learning_rate": 5e-06,
      "loss": 0.5503,
      "step": 2480
    },
    {
      "epoch": 0.5388444059727332,
      "grad_norm": 0.7925318160395382,
      "learning_rate": 5e-06,
      "loss": 0.5657,
      "step": 2490
    },
    {
      "epoch": 0.5410084397316598,
      "grad_norm": 0.8170912915049142,
      "learning_rate": 5e-06,
      "loss": 0.5666,
      "step": 2500
    },
    {
      "epoch": 0.5431724734905865,
      "grad_norm": 0.7848624682678103,
      "learning_rate": 5e-06,
      "loss": 0.5644,
      "step": 2510
    },
    {
      "epoch": 0.5453365072495131,
      "grad_norm": 0.8305872418987535,
      "learning_rate": 5e-06,
      "loss": 0.5579,
      "step": 2520
    },
    {
      "epoch": 0.5475005410084397,
      "grad_norm": 0.754778805221588,
      "learning_rate": 5e-06,
      "loss": 0.5447,
      "step": 2530
    },
    {
      "epoch": 0.5496645747673664,
      "grad_norm": 0.8433164922898455,
      "learning_rate": 5e-06,
      "loss": 0.577,
      "step": 2540
    },
    {
      "epoch": 0.551828608526293,
      "grad_norm": 0.7364207026132309,
      "learning_rate": 5e-06,
      "loss": 0.5502,
      "step": 2550
    },
    {
      "epoch": 0.5539926422852196,
      "grad_norm": 0.7703003076672548,
      "learning_rate": 5e-06,
      "loss": 0.5391,
      "step": 2560
    },
    {
      "epoch": 0.5561566760441463,
      "grad_norm": 0.7874453563557188,
      "learning_rate": 5e-06,
      "loss": 0.5579,
      "step": 2570
    },
    {
      "epoch": 0.5583207098030729,
      "grad_norm": 0.7946629512850162,
      "learning_rate": 5e-06,
      "loss": 0.5667,
      "step": 2580
    },
    {
      "epoch": 0.5604847435619996,
      "grad_norm": 0.7841342441931893,
      "learning_rate": 5e-06,
      "loss": 0.5552,
      "step": 2590
    },
    {
      "epoch": 0.5626487773209262,
      "grad_norm": 0.8770518493960907,
      "learning_rate": 5e-06,
      "loss": 0.5654,
      "step": 2600
    },
    {
      "epoch": 0.5648128110798528,
      "grad_norm": 0.7612803825837043,
      "learning_rate": 5e-06,
      "loss": 0.5481,
      "step": 2610
    },
    {
      "epoch": 0.5669768448387795,
      "grad_norm": 0.8112957896812497,
      "learning_rate": 5e-06,
      "loss": 0.5697,
      "step": 2620
    },
    {
      "epoch": 0.5691408785977061,
      "grad_norm": 0.8292780888969601,
      "learning_rate": 5e-06,
      "loss": 0.5631,
      "step": 2630
    },
    {
      "epoch": 0.5713049123566327,
      "grad_norm": 0.8534427911951067,
      "learning_rate": 5e-06,
      "loss": 0.5675,
      "step": 2640
    },
    {
      "epoch": 0.5734689461155594,
      "grad_norm": 0.7983159182220968,
      "learning_rate": 5e-06,
      "loss": 0.5651,
      "step": 2650
    },
    {
      "epoch": 0.575632979874486,
      "grad_norm": 0.8286911243296953,
      "learning_rate": 5e-06,
      "loss": 0.5661,
      "step": 2660
    },
    {
      "epoch": 0.5777970136334126,
      "grad_norm": 0.7873673354292886,
      "learning_rate": 5e-06,
      "loss": 0.5568,
      "step": 2670
    },
    {
      "epoch": 0.5799610473923393,
      "grad_norm": 0.8643973853008751,
      "learning_rate": 5e-06,
      "loss": 0.5529,
      "step": 2680
    },
    {
      "epoch": 0.5821250811512659,
      "grad_norm": 0.8297018704245965,
      "learning_rate": 5e-06,
      "loss": 0.5646,
      "step": 2690
    },
    {
      "epoch": 0.5842891149101926,
      "grad_norm": 0.8282391086423111,
      "learning_rate": 5e-06,
      "loss": 0.5704,
      "step": 2700
    },
    {
      "epoch": 0.5864531486691192,
      "grad_norm": 0.8185226749510477,
      "learning_rate": 5e-06,
      "loss": 0.5732,
      "step": 2710
    },
    {
      "epoch": 0.5886171824280458,
      "grad_norm": 0.7837221330454157,
      "learning_rate": 5e-06,
      "loss": 0.5668,
      "step": 2720
    },
    {
      "epoch": 0.5907812161869725,
      "grad_norm": 0.7830075768638615,
      "learning_rate": 5e-06,
      "loss": 0.5614,
      "step": 2730
    },
    {
      "epoch": 0.5929452499458991,
      "grad_norm": 0.7669203988650726,
      "learning_rate": 5e-06,
      "loss": 0.5632,
      "step": 2740
    },
    {
      "epoch": 0.5951092837048257,
      "grad_norm": 0.8064484665186933,
      "learning_rate": 5e-06,
      "loss": 0.5599,
      "step": 2750
    },
    {
      "epoch": 0.5972733174637525,
      "grad_norm": 0.8468294496439435,
      "learning_rate": 5e-06,
      "loss": 0.5451,
      "step": 2760
    },
    {
      "epoch": 0.599437351222679,
      "grad_norm": 0.9077279068821958,
      "learning_rate": 5e-06,
      "loss": 0.5724,
      "step": 2770
    },
    {
      "epoch": 0.6016013849816058,
      "grad_norm": 0.7929898760133751,
      "learning_rate": 5e-06,
      "loss": 0.5616,
      "step": 2780
    },
    {
      "epoch": 0.6037654187405324,
      "grad_norm": 0.7971762324043354,
      "learning_rate": 5e-06,
      "loss": 0.563,
      "step": 2790
    },
    {
      "epoch": 0.605929452499459,
      "grad_norm": 0.8288075182953192,
      "learning_rate": 5e-06,
      "loss": 0.5579,
      "step": 2800
    },
    {
      "epoch": 0.6080934862583857,
      "grad_norm": 0.8498140525184206,
      "learning_rate": 5e-06,
      "loss": 0.564,
      "step": 2810
    },
    {
      "epoch": 0.6102575200173123,
      "grad_norm": 0.8102245476480473,
      "learning_rate": 5e-06,
      "loss": 0.5609,
      "step": 2820
    },
    {
      "epoch": 0.6124215537762389,
      "grad_norm": 0.7890358297228374,
      "learning_rate": 5e-06,
      "loss": 0.5665,
      "step": 2830
    },
    {
      "epoch": 0.6145855875351656,
      "grad_norm": 0.7979304789718528,
      "learning_rate": 5e-06,
      "loss": 0.5485,
      "step": 2840
    },
    {
      "epoch": 0.6167496212940922,
      "grad_norm": 0.8003173415042142,
      "learning_rate": 5e-06,
      "loss": 0.5598,
      "step": 2850
    },
    {
      "epoch": 0.6189136550530189,
      "grad_norm": 0.7767086997147948,
      "learning_rate": 5e-06,
      "loss": 0.5516,
      "step": 2860
    },
    {
      "epoch": 0.6210776888119455,
      "grad_norm": 0.8045021838341775,
      "learning_rate": 5e-06,
      "loss": 0.5368,
      "step": 2870
    },
    {
      "epoch": 0.6232417225708721,
      "grad_norm": 0.7498417279355564,
      "learning_rate": 5e-06,
      "loss": 0.5518,
      "step": 2880
    },
    {
      "epoch": 0.6254057563297988,
      "grad_norm": 0.8253099324603884,
      "learning_rate": 5e-06,
      "loss": 0.5581,
      "step": 2890
    },
    {
      "epoch": 0.6275697900887254,
      "grad_norm": 0.8079450289566134,
      "learning_rate": 5e-06,
      "loss": 0.5488,
      "step": 2900
    },
    {
      "epoch": 0.629733823847652,
      "grad_norm": 0.7952980563003963,
      "learning_rate": 5e-06,
      "loss": 0.5685,
      "step": 2910
    },
    {
      "epoch": 0.6318978576065787,
      "grad_norm": 0.7835163645285544,
      "learning_rate": 5e-06,
      "loss": 0.5477,
      "step": 2920
    },
    {
      "epoch": 0.6340618913655053,
      "grad_norm": 0.8048903722703916,
      "learning_rate": 5e-06,
      "loss": 0.5596,
      "step": 2930
    },
    {
      "epoch": 0.636225925124432,
      "grad_norm": 0.81044303853561,
      "learning_rate": 5e-06,
      "loss": 0.5598,
      "step": 2940
    },
    {
      "epoch": 0.6383899588833586,
      "grad_norm": 0.8174138765074157,
      "learning_rate": 5e-06,
      "loss": 0.5505,
      "step": 2950
    },
    {
      "epoch": 0.6405539926422852,
      "grad_norm": 0.8475108197008772,
      "learning_rate": 5e-06,
      "loss": 0.5424,
      "step": 2960
    },
    {
      "epoch": 0.6427180264012119,
      "grad_norm": 0.8042022600558663,
      "learning_rate": 5e-06,
      "loss": 0.5539,
      "step": 2970
    },
    {
      "epoch": 0.6448820601601385,
      "grad_norm": 0.7700287206840692,
      "learning_rate": 5e-06,
      "loss": 0.5513,
      "step": 2980
    },
    {
      "epoch": 0.6470460939190651,
      "grad_norm": 0.8281997779232809,
      "learning_rate": 5e-06,
      "loss": 0.5467,
      "step": 2990
    },
    {
      "epoch": 0.6492101276779918,
      "grad_norm": 0.813584320341026,
      "learning_rate": 5e-06,
      "loss": 0.5603,
      "step": 3000
    },
    {
      "epoch": 0.6513741614369184,
      "grad_norm": 0.8585543401025539,
      "learning_rate": 5e-06,
      "loss": 0.5701,
      "step": 3010
    },
    {
      "epoch": 0.6535381951958451,
      "grad_norm": 0.836920539360064,
      "learning_rate": 5e-06,
      "loss": 0.5504,
      "step": 3020
    },
    {
      "epoch": 0.6557022289547717,
      "grad_norm": 1.0694200561637055,
      "learning_rate": 5e-06,
      "loss": 0.5536,
      "step": 3030
    },
    {
      "epoch": 0.6578662627136983,
      "grad_norm": 0.7556155594948578,
      "learning_rate": 5e-06,
      "loss": 0.5421,
      "step": 3040
    },
    {
      "epoch": 0.660030296472625,
      "grad_norm": 0.8263966978271399,
      "learning_rate": 5e-06,
      "loss": 0.542,
      "step": 3050
    },
    {
      "epoch": 0.6621943302315516,
      "grad_norm": 0.7959474918960274,
      "learning_rate": 5e-06,
      "loss": 0.5544,
      "step": 3060
    },
    {
      "epoch": 0.6643583639904782,
      "grad_norm": 0.782166640538943,
      "learning_rate": 5e-06,
      "loss": 0.5581,
      "step": 3070
    },
    {
      "epoch": 0.6665223977494049,
      "grad_norm": 0.8119704290024918,
      "learning_rate": 5e-06,
      "loss": 0.5652,
      "step": 3080
    },
    {
      "epoch": 0.6686864315083315,
      "grad_norm": 0.8281560870508221,
      "learning_rate": 5e-06,
      "loss": 0.5634,
      "step": 3090
    },
    {
      "epoch": 0.6708504652672582,
      "grad_norm": 0.8095783840417501,
      "learning_rate": 5e-06,
      "loss": 0.5601,
      "step": 3100
    },
    {
      "epoch": 0.6730144990261848,
      "grad_norm": 0.795068179288108,
      "learning_rate": 5e-06,
      "loss": 0.5678,
      "step": 3110
    },
    {
      "epoch": 0.6751785327851114,
      "grad_norm": 0.7792728010855305,
      "learning_rate": 5e-06,
      "loss": 0.5607,
      "step": 3120
    },
    {
      "epoch": 0.6773425665440381,
      "grad_norm": 0.8179451854658257,
      "learning_rate": 5e-06,
      "loss": 0.5435,
      "step": 3130
    },
    {
      "epoch": 0.6795066003029647,
      "grad_norm": 0.7667511572553716,
      "learning_rate": 5e-06,
      "loss": 0.5596,
      "step": 3140
    },
    {
      "epoch": 0.6816706340618913,
      "grad_norm": 0.8029813372366518,
      "learning_rate": 5e-06,
      "loss": 0.55,
      "step": 3150
    },
    {
      "epoch": 0.683834667820818,
      "grad_norm": 0.7987681487999668,
      "learning_rate": 5e-06,
      "loss": 0.5451,
      "step": 3160
    },
    {
      "epoch": 0.6859987015797446,
      "grad_norm": 0.779467033707979,
      "learning_rate": 5e-06,
      "loss": 0.543,
      "step": 3170
    },
    {
      "epoch": 0.6881627353386713,
      "grad_norm": 0.7628766043647324,
      "learning_rate": 5e-06,
      "loss": 0.549,
      "step": 3180
    },
    {
      "epoch": 0.6903267690975979,
      "grad_norm": 0.7850057643186282,
      "learning_rate": 5e-06,
      "loss": 0.5668,
      "step": 3190
    },
    {
      "epoch": 0.6924908028565245,
      "grad_norm": 0.8407107964724864,
      "learning_rate": 5e-06,
      "loss": 0.5501,
      "step": 3200
    },
    {
      "epoch": 0.6946548366154512,
      "grad_norm": 0.7767999916714133,
      "learning_rate": 5e-06,
      "loss": 0.543,
      "step": 3210
    },
    {
      "epoch": 0.6968188703743778,
      "grad_norm": 0.7996258238832009,
      "learning_rate": 5e-06,
      "loss": 0.5483,
      "step": 3220
    },
    {
      "epoch": 0.6989829041333044,
      "grad_norm": 0.8377643899045712,
      "learning_rate": 5e-06,
      "loss": 0.5487,
      "step": 3230
    },
    {
      "epoch": 0.7011469378922311,
      "grad_norm": 0.7514752154537118,
      "learning_rate": 5e-06,
      "loss": 0.5597,
      "step": 3240
    },
    {
      "epoch": 0.7033109716511577,
      "grad_norm": 0.8416197122709502,
      "learning_rate": 5e-06,
      "loss": 0.5565,
      "step": 3250
    },
    {
      "epoch": 0.7054750054100843,
      "grad_norm": 0.7931722916017413,
      "learning_rate": 5e-06,
      "loss": 0.5546,
      "step": 3260
    },
    {
      "epoch": 0.707639039169011,
      "grad_norm": 0.8217577869277688,
      "learning_rate": 5e-06,
      "loss": 0.5653,
      "step": 3270
    },
    {
      "epoch": 0.7098030729279377,
      "grad_norm": 0.7118453738807855,
      "learning_rate": 5e-06,
      "loss": 0.5402,
      "step": 3280
    },
    {
      "epoch": 0.7119671066868644,
      "grad_norm": 0.7697241307191445,
      "learning_rate": 5e-06,
      "loss": 0.5589,
      "step": 3290
    },
    {
      "epoch": 0.714131140445791,
      "grad_norm": 0.802274974087682,
      "learning_rate": 5e-06,
      "loss": 0.5551,
      "step": 3300
    },
    {
      "epoch": 0.7162951742047176,
      "grad_norm": 0.792023853084678,
      "learning_rate": 5e-06,
      "loss": 0.5493,
      "step": 3310
    },
    {
      "epoch": 0.7184592079636443,
      "grad_norm": 0.7544132296053949,
      "learning_rate": 5e-06,
      "loss": 0.5393,
      "step": 3320
    },
    {
      "epoch": 0.7206232417225709,
      "grad_norm": 0.7980914984239874,
      "learning_rate": 5e-06,
      "loss": 0.5433,
      "step": 3330
    },
    {
      "epoch": 0.7227872754814975,
      "grad_norm": 0.784437501269065,
      "learning_rate": 5e-06,
      "loss": 0.5568,
      "step": 3340
    },
    {
      "epoch": 0.7249513092404242,
      "grad_norm": 0.7614522809136514,
      "learning_rate": 5e-06,
      "loss": 0.5455,
      "step": 3350
    },
    {
      "epoch": 0.7271153429993508,
      "grad_norm": 0.7734683988885462,
      "learning_rate": 5e-06,
      "loss": 0.5501,
      "step": 3360
    },
    {
      "epoch": 0.7292793767582775,
      "grad_norm": 0.7866152633301271,
      "learning_rate": 5e-06,
      "loss": 0.556,
      "step": 3370
    },
    {
      "epoch": 0.7314434105172041,
      "grad_norm": 0.7702246648713589,
      "learning_rate": 5e-06,
      "loss": 0.5497,
      "step": 3380
    },
    {
      "epoch": 0.7336074442761307,
      "grad_norm": 0.7600423928776777,
      "learning_rate": 5e-06,
      "loss": 0.5505,
      "step": 3390
    },
    {
      "epoch": 0.7357714780350574,
      "grad_norm": 0.807615131927307,
      "learning_rate": 5e-06,
      "loss": 0.5634,
      "step": 3400
    },
    {
      "epoch": 0.737935511793984,
      "grad_norm": 0.7922705128025508,
      "learning_rate": 5e-06,
      "loss": 0.558,
      "step": 3410
    },
    {
      "epoch": 0.7400995455529106,
      "grad_norm": 0.8339598322147505,
      "learning_rate": 5e-06,
      "loss": 0.552,
      "step": 3420
    },
    {
      "epoch": 0.7422635793118373,
      "grad_norm": 0.7464262500668064,
      "learning_rate": 5e-06,
      "loss": 0.5483,
      "step": 3430
    },
    {
      "epoch": 0.7444276130707639,
      "grad_norm": 0.7741739625171083,
      "learning_rate": 5e-06,
      "loss": 0.5573,
      "step": 3440
    },
    {
      "epoch": 0.7465916468296906,
      "grad_norm": 0.7453424498568156,
      "learning_rate": 5e-06,
      "loss": 0.5438,
      "step": 3450
    },
    {
      "epoch": 0.7487556805886172,
      "grad_norm": 0.754303311641887,
      "learning_rate": 5e-06,
      "loss": 0.5548,
      "step": 3460
    },
    {
      "epoch": 0.7509197143475438,
      "grad_norm": 0.8689734619851907,
      "learning_rate": 5e-06,
      "loss": 0.556,
      "step": 3470
    },
    {
      "epoch": 0.7530837481064705,
      "grad_norm": 0.7773889781541226,
      "learning_rate": 5e-06,
      "loss": 0.5494,
      "step": 3480
    },
    {
      "epoch": 0.7552477818653971,
      "grad_norm": 0.7849782050400904,
      "learning_rate": 5e-06,
      "loss": 0.5537,
      "step": 3490
    },
    {
      "epoch": 0.7574118156243237,
      "grad_norm": 0.7637907513233742,
      "learning_rate": 5e-06,
      "loss": 0.5559,
      "step": 3500
    },
    {
      "epoch": 0.7595758493832504,
      "grad_norm": 0.7679605914857784,
      "learning_rate": 5e-06,
      "loss": 0.5576,
      "step": 3510
    },
    {
      "epoch": 0.761739883142177,
      "grad_norm": 0.828819439384587,
      "learning_rate": 5e-06,
      "loss": 0.566,
      "step": 3520
    },
    {
      "epoch": 0.7639039169011037,
      "grad_norm": 0.8446403352206237,
      "learning_rate": 5e-06,
      "loss": 0.5583,
      "step": 3530
    },
    {
      "epoch": 0.7660679506600303,
      "grad_norm": 0.8075329194131221,
      "learning_rate": 5e-06,
      "loss": 0.5592,
      "step": 3540
    },
    {
      "epoch": 0.7682319844189569,
      "grad_norm": 0.8442605962725928,
      "learning_rate": 5e-06,
      "loss": 0.5439,
      "step": 3550
    },
    {
      "epoch": 0.7703960181778836,
      "grad_norm": 0.8588068255220394,
      "learning_rate": 5e-06,
      "loss": 0.5587,
      "step": 3560
    },
    {
      "epoch": 0.7725600519368102,
      "grad_norm": 0.8524091917859525,
      "learning_rate": 5e-06,
      "loss": 0.5517,
      "step": 3570
    },
    {
      "epoch": 0.7747240856957368,
      "grad_norm": 0.8244971263231196,
      "learning_rate": 5e-06,
      "loss": 0.561,
      "step": 3580
    },
    {
      "epoch": 0.7768881194546635,
      "grad_norm": 0.7627478595291142,
      "learning_rate": 5e-06,
      "loss": 0.5641,
      "step": 3590
    },
    {
      "epoch": 0.7790521532135901,
      "grad_norm": 0.8086037426148818,
      "learning_rate": 5e-06,
      "loss": 0.5621,
      "step": 3600
    },
    {
      "epoch": 0.7812161869725168,
      "grad_norm": 0.791930301424169,
      "learning_rate": 5e-06,
      "loss": 0.5471,
      "step": 3610
    },
    {
      "epoch": 0.7833802207314434,
      "grad_norm": 0.7878241907870572,
      "learning_rate": 5e-06,
      "loss": 0.5521,
      "step": 3620
    },
    {
      "epoch": 0.78554425449037,
      "grad_norm": 0.7830841775537941,
      "learning_rate": 5e-06,
      "loss": 0.5467,
      "step": 3630
    },
    {
      "epoch": 0.7877082882492967,
      "grad_norm": 0.7623641467313392,
      "learning_rate": 5e-06,
      "loss": 0.5336,
      "step": 3640
    },
    {
      "epoch": 0.7898723220082233,
      "grad_norm": 0.8232677684209438,
      "learning_rate": 5e-06,
      "loss": 0.5651,
      "step": 3650
    },
    {
      "epoch": 0.7920363557671499,
      "grad_norm": 0.761362883743628,
      "learning_rate": 5e-06,
      "loss": 0.5597,
      "step": 3660
    },
    {
      "epoch": 0.7942003895260766,
      "grad_norm": 0.8637874664285103,
      "learning_rate": 5e-06,
      "loss": 0.5602,
      "step": 3670
    },
    {
      "epoch": 0.7963644232850032,
      "grad_norm": 0.8267668782394713,
      "learning_rate": 5e-06,
      "loss": 0.5561,
      "step": 3680
    },
    {
      "epoch": 0.7985284570439299,
      "grad_norm": 0.817446940423478,
      "learning_rate": 5e-06,
      "loss": 0.5575,
      "step": 3690
    },
    {
      "epoch": 0.8006924908028565,
      "grad_norm": 0.7818597023986829,
      "learning_rate": 5e-06,
      "loss": 0.5604,
      "step": 3700
    },
    {
      "epoch": 0.8028565245617831,
      "grad_norm": 0.776499232049182,
      "learning_rate": 5e-06,
      "loss": 0.5643,
      "step": 3710
    },
    {
      "epoch": 0.8050205583207098,
      "grad_norm": 0.8364781671818916,
      "learning_rate": 5e-06,
      "loss": 0.5438,
      "step": 3720
    },
    {
      "epoch": 0.8071845920796364,
      "grad_norm": 0.8397874518563361,
      "learning_rate": 5e-06,
      "loss": 0.5513,
      "step": 3730
    },
    {
      "epoch": 0.809348625838563,
      "grad_norm": 0.7969750238399832,
      "learning_rate": 5e-06,
      "loss": 0.5402,
      "step": 3740
    },
    {
      "epoch": 0.8115126595974897,
      "grad_norm": 0.9068956852645959,
      "learning_rate": 5e-06,
      "loss": 0.5569,
      "step": 3750
    },
    {
      "epoch": 0.8136766933564163,
      "grad_norm": 0.7536728350124055,
      "learning_rate": 5e-06,
      "loss": 0.5419,
      "step": 3760
    },
    {
      "epoch": 0.815840727115343,
      "grad_norm": 0.7676812187895282,
      "learning_rate": 5e-06,
      "loss": 0.547,
      "step": 3770
    },
    {
      "epoch": 0.8180047608742697,
      "grad_norm": 0.762833586705676,
      "learning_rate": 5e-06,
      "loss": 0.5521,
      "step": 3780
    },
    {
      "epoch": 0.8201687946331963,
      "grad_norm": 0.756771887345399,
      "learning_rate": 5e-06,
      "loss": 0.5556,
      "step": 3790
    },
    {
      "epoch": 0.822332828392123,
      "grad_norm": 0.8734848583188696,
      "learning_rate": 5e-06,
      "loss": 0.5545,
      "step": 3800
    },
    {
      "epoch": 0.8244968621510496,
      "grad_norm": 0.8004103342026402,
      "learning_rate": 5e-06,
      "loss": 0.5513,
      "step": 3810
    },
    {
      "epoch": 0.8266608959099762,
      "grad_norm": 0.7719483146665875,
      "learning_rate": 5e-06,
      "loss": 0.5446,
      "step": 3820
    },
    {
      "epoch": 0.8288249296689029,
      "grad_norm": 0.7744925484130326,
      "learning_rate": 5e-06,
      "loss": 0.5635,
      "step": 3830
    },
    {
      "epoch": 0.8309889634278295,
      "grad_norm": 0.7562447275301897,
      "learning_rate": 5e-06,
      "loss": 0.5444,
      "step": 3840
    },
    {
      "epoch": 0.8331529971867561,
      "grad_norm": 0.7924604852102323,
      "learning_rate": 5e-06,
      "loss": 0.5421,
      "step": 3850
    },
    {
      "epoch": 0.8353170309456828,
      "grad_norm": 0.8520564532781372,
      "learning_rate": 5e-06,
      "loss": 0.5601,
      "step": 3860
    },
    {
      "epoch": 0.8374810647046094,
      "grad_norm": 0.8746743288593334,
      "learning_rate": 5e-06,
      "loss": 0.5509,
      "step": 3870
    },
    {
      "epoch": 0.8396450984635361,
      "grad_norm": 0.7873244885756802,
      "learning_rate": 5e-06,
      "loss": 0.5472,
      "step": 3880
    },
    {
      "epoch": 0.8418091322224627,
      "grad_norm": 0.8209663148355055,
      "learning_rate": 5e-06,
      "loss": 0.5385,
      "step": 3890
    },
    {
      "epoch": 0.8439731659813893,
      "grad_norm": 0.819629928526103,
      "learning_rate": 5e-06,
      "loss": 0.5392,
      "step": 3900
    },
    {
      "epoch": 0.846137199740316,
      "grad_norm": 0.8584352651397266,
      "learning_rate": 5e-06,
      "loss": 0.5554,
      "step": 3910
    },
    {
      "epoch": 0.8483012334992426,
      "grad_norm": 0.8503960560319187,
      "learning_rate": 5e-06,
      "loss": 0.5363,
      "step": 3920
    },
    {
      "epoch": 0.8504652672581692,
      "grad_norm": 0.8108480185136313,
      "learning_rate": 5e-06,
      "loss": 0.5495,
      "step": 3930
    },
    {
      "epoch": 0.8526293010170959,
      "grad_norm": 0.7373181469886592,
      "learning_rate": 5e-06,
      "loss": 0.5734,
      "step": 3940
    },
    {
      "epoch": 0.8547933347760225,
      "grad_norm": 0.8163364176974437,
      "learning_rate": 5e-06,
      "loss": 0.5599,
      "step": 3950
    },
    {
      "epoch": 0.8569573685349492,
      "grad_norm": 0.7959374117805932,
      "learning_rate": 5e-06,
      "loss": 0.5454,
      "step": 3960
    },
    {
      "epoch": 0.8591214022938758,
      "grad_norm": 0.812330127649017,
      "learning_rate": 5e-06,
      "loss": 0.5471,
      "step": 3970
    },
    {
      "epoch": 0.8612854360528024,
      "grad_norm": 0.7945522211940125,
      "learning_rate": 5e-06,
      "loss": 0.5403,
      "step": 3980
    },
    {
      "epoch": 0.8634494698117291,
      "grad_norm": 0.7696968400441544,
      "learning_rate": 5e-06,
      "loss": 0.5423,
      "step": 3990
    },
    {
      "epoch": 0.8656135035706557,
      "grad_norm": 0.7450513825177912,
      "learning_rate": 5e-06,
      "loss": 0.5405,
      "step": 4000
    },
    {
      "epoch": 0.8677775373295823,
      "grad_norm": 0.8008938478720109,
      "learning_rate": 5e-06,
      "loss": 0.5453,
      "step": 4010
    },
    {
      "epoch": 0.869941571088509,
      "grad_norm": 0.7624217719794756,
      "learning_rate": 5e-06,
      "loss": 0.536,
      "step": 4020
    },
    {
      "epoch": 0.8721056048474356,
      "grad_norm": 0.7876625538590819,
      "learning_rate": 5e-06,
      "loss": 0.55,
      "step": 4030
    },
    {
      "epoch": 0.8742696386063623,
      "grad_norm": 0.7667825013844164,
      "learning_rate": 5e-06,
      "loss": 0.5425,
      "step": 4040
    },
    {
      "epoch": 0.8764336723652889,
      "grad_norm": 0.8391975491875862,
      "learning_rate": 5e-06,
      "loss": 0.5619,
      "step": 4050
    },
    {
      "epoch": 0.8785977061242155,
      "grad_norm": 0.7580142563713078,
      "learning_rate": 5e-06,
      "loss": 0.5395,
      "step": 4060
    },
    {
      "epoch": 0.8807617398831422,
      "grad_norm": 0.873766541325802,
      "learning_rate": 5e-06,
      "loss": 0.5498,
      "step": 4070
    },
    {
      "epoch": 0.8829257736420688,
      "grad_norm": 0.8045714640840378,
      "learning_rate": 5e-06,
      "loss": 0.5565,
      "step": 4080
    },
    {
      "epoch": 0.8850898074009954,
      "grad_norm": 0.7839455981681682,
      "learning_rate": 5e-06,
      "loss": 0.5568,
      "step": 4090
    },
    {
      "epoch": 0.8872538411599221,
      "grad_norm": 0.7507205318651358,
      "learning_rate": 5e-06,
      "loss": 0.5446,
      "step": 4100
    },
    {
      "epoch": 0.8894178749188487,
      "grad_norm": 0.7770875343736735,
      "learning_rate": 5e-06,
      "loss": 0.5478,
      "step": 4110
    },
    {
      "epoch": 0.8915819086777754,
      "grad_norm": 0.739795465311441,
      "learning_rate": 5e-06,
      "loss": 0.5488,
      "step": 4120
    },
    {
      "epoch": 0.893745942436702,
      "grad_norm": 0.8343986949867656,
      "learning_rate": 5e-06,
      "loss": 0.5502,
      "step": 4130
    },
    {
      "epoch": 0.8959099761956286,
      "grad_norm": 0.8072361062581804,
      "learning_rate": 5e-06,
      "loss": 0.5321,
      "step": 4140
    },
    {
      "epoch": 0.8980740099545553,
      "grad_norm": 0.7921719388331905,
      "learning_rate": 5e-06,
      "loss": 0.5469,
      "step": 4150
    },
    {
      "epoch": 0.9002380437134819,
      "grad_norm": 0.7996953628324828,
      "learning_rate": 5e-06,
      "loss": 0.5557,
      "step": 4160
    },
    {
      "epoch": 0.9024020774724085,
      "grad_norm": 0.7272138954178266,
      "learning_rate": 5e-06,
      "loss": 0.5472,
      "step": 4170
    },
    {
      "epoch": 0.9045661112313352,
      "grad_norm": 0.8812123709262756,
      "learning_rate": 5e-06,
      "loss": 0.5349,
      "step": 4180
    },
    {
      "epoch": 0.9067301449902618,
      "grad_norm": 0.8356868824967147,
      "learning_rate": 5e-06,
      "loss": 0.5362,
      "step": 4190
    },
    {
      "epoch": 0.9088941787491885,
      "grad_norm": 0.7972289983410239,
      "learning_rate": 5e-06,
      "loss": 0.5493,
      "step": 4200
    },
    {
      "epoch": 0.9110582125081151,
      "grad_norm": 0.8872937838904918,
      "learning_rate": 5e-06,
      "loss": 0.5406,
      "step": 4210
    },
    {
      "epoch": 0.9132222462670417,
      "grad_norm": 0.7649271319041719,
      "learning_rate": 5e-06,
      "loss": 0.5508,
      "step": 4220
    },
    {
      "epoch": 0.9153862800259684,
      "grad_norm": 0.8237087753008616,
      "learning_rate": 5e-06,
      "loss": 0.5448,
      "step": 4230
    },
    {
      "epoch": 0.917550313784895,
      "grad_norm": 0.7789310787368502,
      "learning_rate": 5e-06,
      "loss": 0.5477,
      "step": 4240
    },
    {
      "epoch": 0.9197143475438216,
      "grad_norm": 0.8216295495680673,
      "learning_rate": 5e-06,
      "loss": 0.562,
      "step": 4250
    },
    {
      "epoch": 0.9218783813027484,
      "grad_norm": 0.8122833746800567,
      "learning_rate": 5e-06,
      "loss": 0.5434,
      "step": 4260
    },
    {
      "epoch": 0.924042415061675,
      "grad_norm": 0.798727376063973,
      "learning_rate": 5e-06,
      "loss": 0.556,
      "step": 4270
    },
    {
      "epoch": 0.9262064488206015,
      "grad_norm": 0.7984668327444104,
      "learning_rate": 5e-06,
      "loss": 0.5397,
      "step": 4280
    },
    {
      "epoch": 0.9283704825795283,
      "grad_norm": 0.7812328766656631,
      "learning_rate": 5e-06,
      "loss": 0.5348,
      "step": 4290
    },
    {
      "epoch": 0.9305345163384549,
      "grad_norm": 0.7893412102379088,
      "learning_rate": 5e-06,
      "loss": 0.549,
      "step": 4300
    },
    {
      "epoch": 0.9326985500973816,
      "grad_norm": 0.7981646202301648,
      "learning_rate": 5e-06,
      "loss": 0.526,
      "step": 4310
    },
    {
      "epoch": 0.9348625838563082,
      "grad_norm": 0.7730443584141122,
      "learning_rate": 5e-06,
      "loss": 0.5339,
      "step": 4320
    },
    {
      "epoch": 0.9370266176152348,
      "grad_norm": 0.844663764946019,
      "learning_rate": 5e-06,
      "loss": 0.5458,
      "step": 4330
    },
    {
      "epoch": 0.9391906513741615,
      "grad_norm": 0.7794979343833512,
      "learning_rate": 5e-06,
      "loss": 0.5505,
      "step": 4340
    },
    {
      "epoch": 0.9413546851330881,
      "grad_norm": 0.7837049408297568,
      "learning_rate": 5e-06,
      "loss": 0.5497,
      "step": 4350
    },
    {
      "epoch": 0.9435187188920147,
      "grad_norm": 0.7854813181023578,
      "learning_rate": 5e-06,
      "loss": 0.5563,
      "step": 4360
    },
    {
      "epoch": 0.9456827526509414,
      "grad_norm": 0.8390893115478139,
      "learning_rate": 5e-06,
      "loss": 0.55,
      "step": 4370
    },
    {
      "epoch": 0.947846786409868,
      "grad_norm": 0.7933589395245781,
      "learning_rate": 5e-06,
      "loss": 0.5375,
      "step": 4380
    },
    {
      "epoch": 0.9500108201687947,
      "grad_norm": 0.7553926376395359,
      "learning_rate": 5e-06,
      "loss": 0.5527,
      "step": 4390
    },
    {
      "epoch": 0.9521748539277213,
      "grad_norm": 0.745261362871144,
      "learning_rate": 5e-06,
      "loss": 0.539,
      "step": 4400
    },
    {
      "epoch": 0.9543388876866479,
      "grad_norm": 0.8391407287157648,
      "learning_rate": 5e-06,
      "loss": 0.5523,
      "step": 4410
    },
    {
      "epoch": 0.9565029214455746,
      "grad_norm": 0.8243378126848461,
      "learning_rate": 5e-06,
      "loss": 0.5445,
      "step": 4420
    },
    {
      "epoch": 0.9586669552045012,
      "grad_norm": 0.8075823666456592,
      "learning_rate": 5e-06,
      "loss": 0.5377,
      "step": 4430
    },
    {
      "epoch": 0.9608309889634278,
      "grad_norm": 0.813436060772594,
      "learning_rate": 5e-06,
      "loss": 0.5519,
      "step": 4440
    },
    {
      "epoch": 0.9629950227223545,
      "grad_norm": 0.7806826995434077,
      "learning_rate": 5e-06,
      "loss": 0.5454,
      "step": 4450
    },
    {
      "epoch": 0.9651590564812811,
      "grad_norm": 0.7670805996089919,
      "learning_rate": 5e-06,
      "loss": 0.537,
      "step": 4460
    },
    {
      "epoch": 0.9673230902402078,
      "grad_norm": 0.82069231371166,
      "learning_rate": 5e-06,
      "loss": 0.5457,
      "step": 4470
    },
    {
      "epoch": 0.9694871239991344,
      "grad_norm": 0.8237227003500424,
      "learning_rate": 5e-06,
      "loss": 0.556,
      "step": 4480
    },
    {
      "epoch": 0.971651157758061,
      "grad_norm": 0.8426211166204151,
      "learning_rate": 5e-06,
      "loss": 0.5497,
      "step": 4490
    },
    {
      "epoch": 0.9738151915169877,
      "grad_norm": 0.7835202417017654,
      "learning_rate": 5e-06,
      "loss": 0.5303,
      "step": 4500
    },
    {
      "epoch": 0.9759792252759143,
      "grad_norm": 0.8392317298621986,
      "learning_rate": 5e-06,
      "loss": 0.5464,
      "step": 4510
    },
    {
      "epoch": 0.9781432590348409,
      "grad_norm": 0.7465598042444795,
      "learning_rate": 5e-06,
      "loss": 0.5419,
      "step": 4520
    },
    {
      "epoch": 0.9803072927937676,
      "grad_norm": 0.7544743383197097,
      "learning_rate": 5e-06,
      "loss": 0.5378,
      "step": 4530
    },
    {
      "epoch": 0.9824713265526942,
      "grad_norm": 0.764919840823264,
      "learning_rate": 5e-06,
      "loss": 0.5329,
      "step": 4540
    },
    {
      "epoch": 0.9846353603116209,
      "grad_norm": 0.7432743151815665,
      "learning_rate": 5e-06,
      "loss": 0.5488,
      "step": 4550
    },
    {
      "epoch": 0.9867993940705475,
      "grad_norm": 0.8962154593923901,
      "learning_rate": 5e-06,
      "loss": 0.544,
      "step": 4560
    },
    {
      "epoch": 0.9889634278294741,
      "grad_norm": 0.8515589470297252,
      "learning_rate": 5e-06,
      "loss": 0.5438,
      "step": 4570
    },
    {
      "epoch": 0.9911274615884008,
      "grad_norm": 0.7539916472553879,
      "learning_rate": 5e-06,
      "loss": 0.5368,
      "step": 4580
    },
    {
      "epoch": 0.9932914953473274,
      "grad_norm": 0.7966250502268061,
      "learning_rate": 5e-06,
      "loss": 0.5516,
      "step": 4590
    },
    {
      "epoch": 0.995455529106254,
      "grad_norm": 0.796275990986384,
      "learning_rate": 5e-06,
      "loss": 0.535,
      "step": 4600
    },
    {
      "epoch": 0.9976195628651807,
      "grad_norm": 0.7572730283592524,
      "learning_rate": 5e-06,
      "loss": 0.5399,
      "step": 4610
    },
    {
      "epoch": 0.9997835966241073,
      "grad_norm": 0.7866241644050972,
      "learning_rate": 5e-06,
      "loss": 0.5396,
      "step": 4620
    },
    {
      "epoch": 1.0,
      "eval_loss": 0.5429847836494446,
      "eval_runtime": 588.4696,
      "eval_samples_per_second": 26.448,
      "eval_steps_per_second": 0.415,
      "step": 4621
    },
    {
      "epoch": 1.001947630383034,
      "grad_norm": 0.8198402400349776,
      "learning_rate": 5e-06,
      "loss": 0.4682,
      "step": 4630
    },
    {
      "epoch": 1.0041116641419605,
      "grad_norm": 0.7801283113390183,
      "learning_rate": 5e-06,
      "loss": 0.4779,
      "step": 4640
    },
    {
      "epoch": 1.0062756979008873,
      "grad_norm": 0.7609745957390824,
      "learning_rate": 5e-06,
      "loss": 0.4627,
      "step": 4650
    },
    {
      "epoch": 1.008439731659814,
      "grad_norm": 0.7821883017630472,
      "learning_rate": 5e-06,
      "loss": 0.4713,
      "step": 4660
    },
    {
      "epoch": 1.0106037654187405,
      "grad_norm": 0.7260403031295831,
      "learning_rate": 5e-06,
      "loss": 0.4684,
      "step": 4670
    },
    {
      "epoch": 1.0127677991776671,
      "grad_norm": 0.7812573688271381,
      "learning_rate": 5e-06,
      "loss": 0.4736,
      "step": 4680
    },
    {
      "epoch": 1.0149318329365937,
      "grad_norm": 0.7822921675866278,
      "learning_rate": 5e-06,
      "loss": 0.4548,
      "step": 4690
    },
    {
      "epoch": 1.0170958666955205,
      "grad_norm": 0.7694090971084214,
      "learning_rate": 5e-06,
      "loss": 0.46,
      "step": 4700
    },
    {
      "epoch": 1.0192599004544471,
      "grad_norm": 0.7459628030829183,
      "learning_rate": 5e-06,
      "loss": 0.464,
      "step": 4710
    },
    {
      "epoch": 1.0214239342133737,
      "grad_norm": 0.715134160634303,
      "learning_rate": 5e-06,
      "loss": 0.4542,
      "step": 4720
    },
    {
      "epoch": 1.0235879679723003,
      "grad_norm": 0.7414824535187435,
      "learning_rate": 5e-06,
      "loss": 0.465,
      "step": 4730
    },
    {
      "epoch": 1.025752001731227,
      "grad_norm": 0.7453860186675011,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 4740
    },
    {
      "epoch": 1.0279160354901538,
      "grad_norm": 0.7718909976941406,
      "learning_rate": 5e-06,
      "loss": 0.4632,
      "step": 4750
    },
    {
      "epoch": 1.0300800692490804,
      "grad_norm": 0.7645671037278959,
      "learning_rate": 5e-06,
      "loss": 0.4736,
      "step": 4760
    },
    {
      "epoch": 1.032244103008007,
      "grad_norm": 0.7718550788244495,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 4770
    },
    {
      "epoch": 1.0344081367669336,
      "grad_norm": 0.7964499026649204,
      "learning_rate": 5e-06,
      "loss": 0.4833,
      "step": 4780
    },
    {
      "epoch": 1.0365721705258601,
      "grad_norm": 0.834083497700352,
      "learning_rate": 5e-06,
      "loss": 0.4766,
      "step": 4790
    },
    {
      "epoch": 1.0387362042847867,
      "grad_norm": 0.7319738782041046,
      "learning_rate": 5e-06,
      "loss": 0.4755,
      "step": 4800
    },
    {
      "epoch": 1.0409002380437136,
      "grad_norm": 0.7462008777592158,
      "learning_rate": 5e-06,
      "loss": 0.4688,
      "step": 4810
    },
    {
      "epoch": 1.0430642718026402,
      "grad_norm": 0.7694025080042929,
      "learning_rate": 5e-06,
      "loss": 0.4731,
      "step": 4820
    },
    {
      "epoch": 1.0452283055615668,
      "grad_norm": 0.7712938874733282,
      "learning_rate": 5e-06,
      "loss": 0.4783,
      "step": 4830
    },
    {
      "epoch": 1.0473923393204934,
      "grad_norm": 0.6854206720325331,
      "learning_rate": 5e-06,
      "loss": 0.4496,
      "step": 4840
    },
    {
      "epoch": 1.04955637307942,
      "grad_norm": 0.7296190822576524,
      "learning_rate": 5e-06,
      "loss": 0.4466,
      "step": 4850
    },
    {
      "epoch": 1.0517204068383468,
      "grad_norm": 0.8044988975714437,
      "learning_rate": 5e-06,
      "loss": 0.4763,
      "step": 4860
    },
    {
      "epoch": 1.0538844405972734,
      "grad_norm": 0.7550484269456027,
      "learning_rate": 5e-06,
      "loss": 0.4631,
      "step": 4870
    },
    {
      "epoch": 1.0560484743562,
      "grad_norm": 0.7716087062170465,
      "learning_rate": 5e-06,
      "loss": 0.4638,
      "step": 4880
    },
    {
      "epoch": 1.0582125081151266,
      "grad_norm": 0.7069074185268656,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 4890
    },
    {
      "epoch": 1.0603765418740532,
      "grad_norm": 0.7406983646764749,
      "learning_rate": 5e-06,
      "loss": 0.4661,
      "step": 4900
    },
    {
      "epoch": 1.0625405756329798,
      "grad_norm": 0.8513739665965723,
      "learning_rate": 5e-06,
      "loss": 0.4779,
      "step": 4910
    },
    {
      "epoch": 1.0647046093919066,
      "grad_norm": 0.7741744083175378,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 4920
    },
    {
      "epoch": 1.0668686431508332,
      "grad_norm": 0.841842230732826,
      "learning_rate": 5e-06,
      "loss": 0.4559,
      "step": 4930
    },
    {
      "epoch": 1.0690326769097598,
      "grad_norm": 0.7561799157279342,
      "learning_rate": 5e-06,
      "loss": 0.4661,
      "step": 4940
    },
    {
      "epoch": 1.0711967106686864,
      "grad_norm": 0.7452446191339013,
      "learning_rate": 5e-06,
      "loss": 0.4796,
      "step": 4950
    },
    {
      "epoch": 1.073360744427613,
      "grad_norm": 0.7219452780323502,
      "learning_rate": 5e-06,
      "loss": 0.4693,
      "step": 4960
    },
    {
      "epoch": 1.0755247781865398,
      "grad_norm": 0.8011050940745716,
      "learning_rate": 5e-06,
      "loss": 0.4625,
      "step": 4970
    },
    {
      "epoch": 1.0776888119454664,
      "grad_norm": 0.794254062662002,
      "learning_rate": 5e-06,
      "loss": 0.4734,
      "step": 4980
    },
    {
      "epoch": 1.079852845704393,
      "grad_norm": 0.7841880358208203,
      "learning_rate": 5e-06,
      "loss": 0.4735,
      "step": 4990
    },
    {
      "epoch": 1.0820168794633196,
      "grad_norm": 0.7541874362266996,
      "learning_rate": 5e-06,
      "loss": 0.4719,
      "step": 5000
    },
    {
      "epoch": 1.0841809132222462,
      "grad_norm": 0.7832163238689167,
      "learning_rate": 5e-06,
      "loss": 0.4781,
      "step": 5010
    },
    {
      "epoch": 1.086344946981173,
      "grad_norm": 0.7612558574099242,
      "learning_rate": 5e-06,
      "loss": 0.4677,
      "step": 5020
    },
    {
      "epoch": 1.0885089807400996,
      "grad_norm": 0.7457279545654795,
      "learning_rate": 5e-06,
      "loss": 0.4568,
      "step": 5030
    },
    {
      "epoch": 1.0906730144990262,
      "grad_norm": 0.7802158241592927,
      "learning_rate": 5e-06,
      "loss": 0.4695,
      "step": 5040
    },
    {
      "epoch": 1.0928370482579528,
      "grad_norm": 0.7523296252008893,
      "learning_rate": 5e-06,
      "loss": 0.459,
      "step": 5050
    },
    {
      "epoch": 1.0950010820168794,
      "grad_norm": 0.7317362216143073,
      "learning_rate": 5e-06,
      "loss": 0.4594,
      "step": 5060
    },
    {
      "epoch": 1.097165115775806,
      "grad_norm": 0.8089179922502759,
      "learning_rate": 5e-06,
      "loss": 0.4701,
      "step": 5070
    },
    {
      "epoch": 1.0993291495347328,
      "grad_norm": 0.7971117700530117,
      "learning_rate": 5e-06,
      "loss": 0.4647,
      "step": 5080
    },
    {
      "epoch": 1.1014931832936594,
      "grad_norm": 0.7797006135536613,
      "learning_rate": 5e-06,
      "loss": 0.4746,
      "step": 5090
    },
    {
      "epoch": 1.103657217052586,
      "grad_norm": 0.8125978920365499,
      "learning_rate": 5e-06,
      "loss": 0.4679,
      "step": 5100
    },
    {
      "epoch": 1.1058212508115126,
      "grad_norm": 0.7785309067797123,
      "learning_rate": 5e-06,
      "loss": 0.4621,
      "step": 5110
    },
    {
      "epoch": 1.1079852845704392,
      "grad_norm": 0.8022112834916565,
      "learning_rate": 5e-06,
      "loss": 0.4547,
      "step": 5120
    },
    {
      "epoch": 1.110149318329366,
      "grad_norm": 0.7644085293828123,
      "learning_rate": 5e-06,
      "loss": 0.4694,
      "step": 5130
    },
    {
      "epoch": 1.1123133520882926,
      "grad_norm": 0.7337804491792407,
      "learning_rate": 5e-06,
      "loss": 0.4594,
      "step": 5140
    },
    {
      "epoch": 1.1144773858472192,
      "grad_norm": 0.7430781593703488,
      "learning_rate": 5e-06,
      "loss": 0.4655,
      "step": 5150
    },
    {
      "epoch": 1.1166414196061458,
      "grad_norm": 0.7666150250001785,
      "learning_rate": 5e-06,
      "loss": 0.4601,
      "step": 5160
    },
    {
      "epoch": 1.1188054533650724,
      "grad_norm": 0.7926818454734849,
      "learning_rate": 5e-06,
      "loss": 0.475,
      "step": 5170
    },
    {
      "epoch": 1.120969487123999,
      "grad_norm": 0.7761288586447196,
      "learning_rate": 5e-06,
      "loss": 0.464,
      "step": 5180
    },
    {
      "epoch": 1.1231335208829258,
      "grad_norm": 0.7715150614179276,
      "learning_rate": 5e-06,
      "loss": 0.4683,
      "step": 5190
    },
    {
      "epoch": 1.1252975546418524,
      "grad_norm": 0.768487598686053,
      "learning_rate": 5e-06,
      "loss": 0.4597,
      "step": 5200
    },
    {
      "epoch": 1.127461588400779,
      "grad_norm": 0.7780154395955231,
      "learning_rate": 5e-06,
      "loss": 0.4815,
      "step": 5210
    },
    {
      "epoch": 1.1296256221597056,
      "grad_norm": 0.7754600935683129,
      "learning_rate": 5e-06,
      "loss": 0.4691,
      "step": 5220
    },
    {
      "epoch": 1.1317896559186322,
      "grad_norm": 0.7168967688429924,
      "learning_rate": 5e-06,
      "loss": 0.4576,
      "step": 5230
    },
    {
      "epoch": 1.133953689677559,
      "grad_norm": 0.7887975954709602,
      "learning_rate": 5e-06,
      "loss": 0.4656,
      "step": 5240
    },
    {
      "epoch": 1.1361177234364856,
      "grad_norm": 0.7711181920453322,
      "learning_rate": 5e-06,
      "loss": 0.4638,
      "step": 5250
    },
    {
      "epoch": 1.1382817571954122,
      "grad_norm": 0.8040928973948708,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 5260
    },
    {
      "epoch": 1.1404457909543388,
      "grad_norm": 0.7480038005652744,
      "learning_rate": 5e-06,
      "loss": 0.4654,
      "step": 5270
    },
    {
      "epoch": 1.1426098247132654,
      "grad_norm": 0.7772151707383336,
      "learning_rate": 5e-06,
      "loss": 0.4728,
      "step": 5280
    },
    {
      "epoch": 1.1447738584721923,
      "grad_norm": 0.7915680198334887,
      "learning_rate": 5e-06,
      "loss": 0.4576,
      "step": 5290
    },
    {
      "epoch": 1.1469378922311189,
      "grad_norm": 0.7591374087955354,
      "learning_rate": 5e-06,
      "loss": 0.4707,
      "step": 5300
    },
    {
      "epoch": 1.1491019259900455,
      "grad_norm": 0.7805726783738297,
      "learning_rate": 5e-06,
      "loss": 0.4656,
      "step": 5310
    },
    {
      "epoch": 1.151265959748972,
      "grad_norm": 0.7721078057930081,
      "learning_rate": 5e-06,
      "loss": 0.4711,
      "step": 5320
    },
    {
      "epoch": 1.1534299935078987,
      "grad_norm": 0.7400720348849441,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 5330
    },
    {
      "epoch": 1.1555940272668255,
      "grad_norm": 0.7435323729236517,
      "learning_rate": 5e-06,
      "loss": 0.4778,
      "step": 5340
    },
    {
      "epoch": 1.157758061025752,
      "grad_norm": 0.7911198407125987,
      "learning_rate": 5e-06,
      "loss": 0.4665,
      "step": 5350
    },
    {
      "epoch": 1.1599220947846787,
      "grad_norm": 0.7610463033369455,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 5360
    },
    {
      "epoch": 1.1620861285436053,
      "grad_norm": 0.7711693447277651,
      "learning_rate": 5e-06,
      "loss": 0.4726,
      "step": 5370
    },
    {
      "epoch": 1.1642501623025319,
      "grad_norm": 0.7859946817663731,
      "learning_rate": 5e-06,
      "loss": 0.4617,
      "step": 5380
    },
    {
      "epoch": 1.1664141960614585,
      "grad_norm": 0.7890418189022593,
      "learning_rate": 5e-06,
      "loss": 0.4713,
      "step": 5390
    },
    {
      "epoch": 1.1685782298203853,
      "grad_norm": 0.7220427522209382,
      "learning_rate": 5e-06,
      "loss": 0.4572,
      "step": 5400
    },
    {
      "epoch": 1.1707422635793119,
      "grad_norm": 0.7713515210815366,
      "learning_rate": 5e-06,
      "loss": 0.4769,
      "step": 5410
    },
    {
      "epoch": 1.1729062973382385,
      "grad_norm": 0.77210689086503,
      "learning_rate": 5e-06,
      "loss": 0.4796,
      "step": 5420
    },
    {
      "epoch": 1.175070331097165,
      "grad_norm": 0.741038220204444,
      "learning_rate": 5e-06,
      "loss": 0.4574,
      "step": 5430
    },
    {
      "epoch": 1.1772343648560917,
      "grad_norm": 0.7831344931379454,
      "learning_rate": 5e-06,
      "loss": 0.4566,
      "step": 5440
    },
    {
      "epoch": 1.1793983986150183,
      "grad_norm": 0.7345331592568521,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 5450
    },
    {
      "epoch": 1.181562432373945,
      "grad_norm": 0.7820999080210016,
      "learning_rate": 5e-06,
      "loss": 0.4578,
      "step": 5460
    },
    {
      "epoch": 1.1837264661328717,
      "grad_norm": 0.7853647666817763,
      "learning_rate": 5e-06,
      "loss": 0.4706,
      "step": 5470
    },
    {
      "epoch": 1.1858904998917983,
      "grad_norm": 0.7893513383734171,
      "learning_rate": 5e-06,
      "loss": 0.4645,
      "step": 5480
    },
    {
      "epoch": 1.1880545336507249,
      "grad_norm": 0.7936708468940181,
      "learning_rate": 5e-06,
      "loss": 0.4609,
      "step": 5490
    },
    {
      "epoch": 1.1902185674096515,
      "grad_norm": 0.7199140376687585,
      "learning_rate": 5e-06,
      "loss": 0.458,
      "step": 5500
    },
    {
      "epoch": 1.1923826011685783,
      "grad_norm": 0.7381096208206134,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 5510
    },
    {
      "epoch": 1.194546634927505,
      "grad_norm": 0.8083844653278189,
      "learning_rate": 5e-06,
      "loss": 0.4683,
      "step": 5520
    },
    {
      "epoch": 1.1967106686864315,
      "grad_norm": 0.7791767856954264,
      "learning_rate": 5e-06,
      "loss": 0.4727,
      "step": 5530
    },
    {
      "epoch": 1.198874702445358,
      "grad_norm": 0.732693234829825,
      "learning_rate": 5e-06,
      "loss": 0.4707,
      "step": 5540
    },
    {
      "epoch": 1.2010387362042847,
      "grad_norm": 0.7467511842772838,
      "learning_rate": 5e-06,
      "loss": 0.4573,
      "step": 5550
    },
    {
      "epoch": 1.2032027699632115,
      "grad_norm": 0.7816698329179276,
      "learning_rate": 5e-06,
      "loss": 0.4775,
      "step": 5560
    },
    {
      "epoch": 1.2053668037221381,
      "grad_norm": 0.7443727351984066,
      "learning_rate": 5e-06,
      "loss": 0.4669,
      "step": 5570
    },
    {
      "epoch": 1.2075308374810647,
      "grad_norm": 0.7425842695291263,
      "learning_rate": 5e-06,
      "loss": 0.4583,
      "step": 5580
    },
    {
      "epoch": 1.2096948712399913,
      "grad_norm": 0.7240789385956823,
      "learning_rate": 5e-06,
      "loss": 0.4576,
      "step": 5590
    },
    {
      "epoch": 1.211858904998918,
      "grad_norm": 0.8030670523927637,
      "learning_rate": 5e-06,
      "loss": 0.4684,
      "step": 5600
    },
    {
      "epoch": 1.2140229387578447,
      "grad_norm": 0.780934768631753,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 5610
    },
    {
      "epoch": 1.2161869725167713,
      "grad_norm": 0.7842677396611177,
      "learning_rate": 5e-06,
      "loss": 0.4641,
      "step": 5620
    },
    {
      "epoch": 1.218351006275698,
      "grad_norm": 0.8130606689249475,
      "learning_rate": 5e-06,
      "loss": 0.466,
      "step": 5630
    },
    {
      "epoch": 1.2205150400346245,
      "grad_norm": 0.7815641772761741,
      "learning_rate": 5e-06,
      "loss": 0.4741,
      "step": 5640
    },
    {
      "epoch": 1.2226790737935511,
      "grad_norm": 0.7490386272049809,
      "learning_rate": 5e-06,
      "loss": 0.4645,
      "step": 5650
    },
    {
      "epoch": 1.224843107552478,
      "grad_norm": 0.8212925702802806,
      "learning_rate": 5e-06,
      "loss": 0.4832,
      "step": 5660
    },
    {
      "epoch": 1.2270071413114045,
      "grad_norm": 0.7630549664820707,
      "learning_rate": 5e-06,
      "loss": 0.4604,
      "step": 5670
    },
    {
      "epoch": 1.2291711750703311,
      "grad_norm": 0.8239583313017402,
      "learning_rate": 5e-06,
      "loss": 0.4708,
      "step": 5680
    },
    {
      "epoch": 1.2313352088292577,
      "grad_norm": 0.760630893549288,
      "learning_rate": 5e-06,
      "loss": 0.4694,
      "step": 5690
    },
    {
      "epoch": 1.2334992425881843,
      "grad_norm": 0.7748580962778236,
      "learning_rate": 5e-06,
      "loss": 0.4676,
      "step": 5700
    },
    {
      "epoch": 1.235663276347111,
      "grad_norm": 0.7601867125563785,
      "learning_rate": 5e-06,
      "loss": 0.4643,
      "step": 5710
    },
    {
      "epoch": 1.2378273101060377,
      "grad_norm": 0.7954825352678104,
      "learning_rate": 5e-06,
      "loss": 0.4609,
      "step": 5720
    },
    {
      "epoch": 1.2399913438649643,
      "grad_norm": 0.7751059458000068,
      "learning_rate": 5e-06,
      "loss": 0.459,
      "step": 5730
    },
    {
      "epoch": 1.242155377623891,
      "grad_norm": 0.7718457396804564,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 5740
    },
    {
      "epoch": 1.2443194113828175,
      "grad_norm": 0.7754578421273243,
      "learning_rate": 5e-06,
      "loss": 0.4638,
      "step": 5750
    },
    {
      "epoch": 1.2464834451417441,
      "grad_norm": 0.7548864065745818,
      "learning_rate": 5e-06,
      "loss": 0.4584,
      "step": 5760
    },
    {
      "epoch": 1.2486474789006707,
      "grad_norm": 0.771889502196166,
      "learning_rate": 5e-06,
      "loss": 0.4717,
      "step": 5770
    },
    {
      "epoch": 1.2508115126595976,
      "grad_norm": 0.7651949593744706,
      "learning_rate": 5e-06,
      "loss": 0.4581,
      "step": 5780
    },
    {
      "epoch": 1.2529755464185242,
      "grad_norm": 0.7561690262824943,
      "learning_rate": 5e-06,
      "loss": 0.4611,
      "step": 5790
    },
    {
      "epoch": 1.2551395801774508,
      "grad_norm": 0.800376700365697,
      "learning_rate": 5e-06,
      "loss": 0.4522,
      "step": 5800
    },
    {
      "epoch": 1.2573036139363774,
      "grad_norm": 0.7501757361110241,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 5810
    },
    {
      "epoch": 1.259467647695304,
      "grad_norm": 0.7926927152467118,
      "learning_rate": 5e-06,
      "loss": 0.4594,
      "step": 5820
    },
    {
      "epoch": 1.2616316814542308,
      "grad_norm": 0.7757167167590527,
      "learning_rate": 5e-06,
      "loss": 0.4698,
      "step": 5830
    },
    {
      "epoch": 1.2637957152131574,
      "grad_norm": 0.7543427971274005,
      "learning_rate": 5e-06,
      "loss": 0.4609,
      "step": 5840
    },
    {
      "epoch": 1.265959748972084,
      "grad_norm": 0.7029065517148244,
      "learning_rate": 5e-06,
      "loss": 0.4576,
      "step": 5850
    },
    {
      "epoch": 1.2681237827310106,
      "grad_norm": 0.7949904996002711,
      "learning_rate": 5e-06,
      "loss": 0.4728,
      "step": 5860
    },
    {
      "epoch": 1.2702878164899372,
      "grad_norm": 0.8061316601164296,
      "learning_rate": 5e-06,
      "loss": 0.4683,
      "step": 5870
    },
    {
      "epoch": 1.272451850248864,
      "grad_norm": 0.7679124071082001,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 5880
    },
    {
      "epoch": 1.2746158840077906,
      "grad_norm": 0.7268154855749079,
      "learning_rate": 5e-06,
      "loss": 0.4613,
      "step": 5890
    },
    {
      "epoch": 1.2767799177667172,
      "grad_norm": 0.7581450065958767,
      "learning_rate": 5e-06,
      "loss": 0.4846,
      "step": 5900
    },
    {
      "epoch": 1.2789439515256438,
      "grad_norm": 0.7728758954937528,
      "learning_rate": 5e-06,
      "loss": 0.4679,
      "step": 5910
    },
    {
      "epoch": 1.2811079852845704,
      "grad_norm": 0.7757132304247395,
      "learning_rate": 5e-06,
      "loss": 0.4735,
      "step": 5920
    },
    {
      "epoch": 1.2832720190434972,
      "grad_norm": 0.7385618172633933,
      "learning_rate": 5e-06,
      "loss": 0.449,
      "step": 5930
    },
    {
      "epoch": 1.2854360528024238,
      "grad_norm": 0.7462092766571883,
      "learning_rate": 5e-06,
      "loss": 0.4607,
      "step": 5940
    },
    {
      "epoch": 1.2876000865613504,
      "grad_norm": 0.7571240627556616,
      "learning_rate": 5e-06,
      "loss": 0.4617,
      "step": 5950
    },
    {
      "epoch": 1.289764120320277,
      "grad_norm": 0.788671204299716,
      "learning_rate": 5e-06,
      "loss": 0.4663,
      "step": 5960
    },
    {
      "epoch": 1.2919281540792036,
      "grad_norm": 0.7494295917052568,
      "learning_rate": 5e-06,
      "loss": 0.4653,
      "step": 5970
    },
    {
      "epoch": 1.2940921878381304,
      "grad_norm": 0.7384360351296172,
      "learning_rate": 5e-06,
      "loss": 0.465,
      "step": 5980
    },
    {
      "epoch": 1.2962562215970568,
      "grad_norm": 0.7352800951816788,
      "learning_rate": 5e-06,
      "loss": 0.4826,
      "step": 5990
    },
    {
      "epoch": 1.2984202553559836,
      "grad_norm": 0.7602237811687234,
      "learning_rate": 5e-06,
      "loss": 0.4567,
      "step": 6000
    },
    {
      "epoch": 1.3005842891149102,
      "grad_norm": 0.7530207047121187,
      "learning_rate": 5e-06,
      "loss": 0.4684,
      "step": 6010
    },
    {
      "epoch": 1.3027483228738368,
      "grad_norm": 0.7356178071902953,
      "learning_rate": 5e-06,
      "loss": 0.4651,
      "step": 6020
    },
    {
      "epoch": 1.3049123566327634,
      "grad_norm": 0.7797947935109668,
      "learning_rate": 5e-06,
      "loss": 0.466,
      "step": 6030
    },
    {
      "epoch": 1.30707639039169,
      "grad_norm": 0.7574370101991291,
      "learning_rate": 5e-06,
      "loss": 0.4711,
      "step": 6040
    },
    {
      "epoch": 1.3092404241506168,
      "grad_norm": 0.7688758966039535,
      "learning_rate": 5e-06,
      "loss": 0.4583,
      "step": 6050
    },
    {
      "epoch": 1.3114044579095434,
      "grad_norm": 0.7745547556636165,
      "learning_rate": 5e-06,
      "loss": 0.4761,
      "step": 6060
    },
    {
      "epoch": 1.31356849166847,
      "grad_norm": 0.7728585651214986,
      "learning_rate": 5e-06,
      "loss": 0.4749,
      "step": 6070
    },
    {
      "epoch": 1.3157325254273966,
      "grad_norm": 0.8170846148828539,
      "learning_rate": 5e-06,
      "loss": 0.4657,
      "step": 6080
    },
    {
      "epoch": 1.3178965591863232,
      "grad_norm": 0.7264085797482112,
      "learning_rate": 5e-06,
      "loss": 0.4582,
      "step": 6090
    },
    {
      "epoch": 1.32006059294525,
      "grad_norm": 0.719801823367032,
      "learning_rate": 5e-06,
      "loss": 0.4538,
      "step": 6100
    },
    {
      "epoch": 1.3222246267041766,
      "grad_norm": 0.7607932021425476,
      "learning_rate": 5e-06,
      "loss": 0.4715,
      "step": 6110
    },
    {
      "epoch": 1.3243886604631032,
      "grad_norm": 0.775171785274517,
      "learning_rate": 5e-06,
      "loss": 0.4647,
      "step": 6120
    },
    {
      "epoch": 1.3265526942220298,
      "grad_norm": 0.7619426266950845,
      "learning_rate": 5e-06,
      "loss": 0.472,
      "step": 6130
    },
    {
      "epoch": 1.3287167279809564,
      "grad_norm": 0.7591540829634035,
      "learning_rate": 5e-06,
      "loss": 0.4571,
      "step": 6140
    },
    {
      "epoch": 1.3308807617398832,
      "grad_norm": 0.7905338255727203,
      "learning_rate": 5e-06,
      "loss": 0.4732,
      "step": 6150
    },
    {
      "epoch": 1.3330447954988098,
      "grad_norm": 0.8612647077572982,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 6160
    },
    {
      "epoch": 1.3352088292577364,
      "grad_norm": 0.7786711376086402,
      "learning_rate": 5e-06,
      "loss": 0.4717,
      "step": 6170
    },
    {
      "epoch": 1.337372863016663,
      "grad_norm": 0.7846310010615345,
      "learning_rate": 5e-06,
      "loss": 0.4625,
      "step": 6180
    },
    {
      "epoch": 1.3395368967755896,
      "grad_norm": 0.8811606228034371,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 6190
    },
    {
      "epoch": 1.3417009305345164,
      "grad_norm": 0.8146235287621797,
      "learning_rate": 5e-06,
      "loss": 0.4596,
      "step": 6200
    },
    {
      "epoch": 1.343864964293443,
      "grad_norm": 0.7533772816124655,
      "learning_rate": 5e-06,
      "loss": 0.4582,
      "step": 6210
    },
    {
      "epoch": 1.3460289980523696,
      "grad_norm": 0.7830512487247953,
      "learning_rate": 5e-06,
      "loss": 0.4736,
      "step": 6220
    },
    {
      "epoch": 1.3481930318112962,
      "grad_norm": 0.7797798896935799,
      "learning_rate": 5e-06,
      "loss": 0.4681,
      "step": 6230
    },
    {
      "epoch": 1.3503570655702228,
      "grad_norm": 0.7932078006037683,
      "learning_rate": 5e-06,
      "loss": 0.4712,
      "step": 6240
    },
    {
      "epoch": 1.3525210993291497,
      "grad_norm": 0.8036931232108603,
      "learning_rate": 5e-06,
      "loss": 0.4672,
      "step": 6250
    },
    {
      "epoch": 1.3546851330880763,
      "grad_norm": 0.7368268217791855,
      "learning_rate": 5e-06,
      "loss": 0.4695,
      "step": 6260
    },
    {
      "epoch": 1.3568491668470029,
      "grad_norm": 0.7359272221589768,
      "learning_rate": 5e-06,
      "loss": 0.466,
      "step": 6270
    },
    {
      "epoch": 1.3590132006059295,
      "grad_norm": 0.8200622976978171,
      "learning_rate": 5e-06,
      "loss": 0.4723,
      "step": 6280
    },
    {
      "epoch": 1.361177234364856,
      "grad_norm": 0.7939699615671225,
      "learning_rate": 5e-06,
      "loss": 0.468,
      "step": 6290
    },
    {
      "epoch": 1.3633412681237829,
      "grad_norm": 0.800957707217982,
      "learning_rate": 5e-06,
      "loss": 0.4529,
      "step": 6300
    },
    {
      "epoch": 1.3655053018827092,
      "grad_norm": 0.80512867222786,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 6310
    },
    {
      "epoch": 1.367669335641636,
      "grad_norm": 0.7360035804063453,
      "learning_rate": 5e-06,
      "loss": 0.4635,
      "step": 6320
    },
    {
      "epoch": 1.3698333694005627,
      "grad_norm": 0.7716321408040111,
      "learning_rate": 5e-06,
      "loss": 0.454,
      "step": 6330
    },
    {
      "epoch": 1.3719974031594893,
      "grad_norm": 0.776027207843901,
      "learning_rate": 5e-06,
      "loss": 0.4613,
      "step": 6340
    },
    {
      "epoch": 1.3741614369184159,
      "grad_norm": 0.7658728763444741,
      "learning_rate": 5e-06,
      "loss": 0.4713,
      "step": 6350
    },
    {
      "epoch": 1.3763254706773425,
      "grad_norm": 0.7884808842340179,
      "learning_rate": 5e-06,
      "loss": 0.4746,
      "step": 6360
    },
    {
      "epoch": 1.3784895044362693,
      "grad_norm": 0.7775593156734237,
      "learning_rate": 5e-06,
      "loss": 0.4677,
      "step": 6370
    },
    {
      "epoch": 1.3806535381951959,
      "grad_norm": 0.7949890338718408,
      "learning_rate": 5e-06,
      "loss": 0.4635,
      "step": 6380
    },
    {
      "epoch": 1.3828175719541225,
      "grad_norm": 0.7828096111334193,
      "learning_rate": 5e-06,
      "loss": 0.462,
      "step": 6390
    },
    {
      "epoch": 1.384981605713049,
      "grad_norm": 0.7814745763059232,
      "learning_rate": 5e-06,
      "loss": 0.4718,
      "step": 6400
    },
    {
      "epoch": 1.3871456394719757,
      "grad_norm": 0.854764057136108,
      "learning_rate": 5e-06,
      "loss": 0.4734,
      "step": 6410
    },
    {
      "epoch": 1.3893096732309025,
      "grad_norm": 0.8266509384735298,
      "learning_rate": 5e-06,
      "loss": 0.4698,
      "step": 6420
    },
    {
      "epoch": 1.391473706989829,
      "grad_norm": 0.7213913037415728,
      "learning_rate": 5e-06,
      "loss": 0.4487,
      "step": 6430
    },
    {
      "epoch": 1.3936377407487557,
      "grad_norm": 0.7231480436825534,
      "learning_rate": 5e-06,
      "loss": 0.459,
      "step": 6440
    },
    {
      "epoch": 1.3958017745076823,
      "grad_norm": 0.7509805363624873,
      "learning_rate": 5e-06,
      "loss": 0.4621,
      "step": 6450
    },
    {
      "epoch": 1.3979658082666089,
      "grad_norm": 0.737834748034084,
      "learning_rate": 5e-06,
      "loss": 0.4695,
      "step": 6460
    },
    {
      "epoch": 1.4001298420255357,
      "grad_norm": 0.7620010469419662,
      "learning_rate": 5e-06,
      "loss": 0.4623,
      "step": 6470
    },
    {
      "epoch": 1.4022938757844623,
      "grad_norm": 0.7323866973353454,
      "learning_rate": 5e-06,
      "loss": 0.4765,
      "step": 6480
    },
    {
      "epoch": 1.404457909543389,
      "grad_norm": 0.7648536342318661,
      "learning_rate": 5e-06,
      "loss": 0.4644,
      "step": 6490
    },
    {
      "epoch": 1.4066219433023155,
      "grad_norm": 0.7674198954299694,
      "learning_rate": 5e-06,
      "loss": 0.4642,
      "step": 6500
    },
    {
      "epoch": 1.408785977061242,
      "grad_norm": 0.746120158345663,
      "learning_rate": 5e-06,
      "loss": 0.4597,
      "step": 6510
    },
    {
      "epoch": 1.410950010820169,
      "grad_norm": 0.7650565264216188,
      "learning_rate": 5e-06,
      "loss": 0.4594,
      "step": 6520
    },
    {
      "epoch": 1.4131140445790955,
      "grad_norm": 0.7369052966841463,
      "learning_rate": 5e-06,
      "loss": 0.4599,
      "step": 6530
    },
    {
      "epoch": 1.415278078338022,
      "grad_norm": 0.7661658293815887,
      "learning_rate": 5e-06,
      "loss": 0.4788,
      "step": 6540
    },
    {
      "epoch": 1.4174421120969487,
      "grad_norm": 0.8245544955088241,
      "learning_rate": 5e-06,
      "loss": 0.4759,
      "step": 6550
    },
    {
      "epoch": 1.4196061458558753,
      "grad_norm": 0.7739803616267428,
      "learning_rate": 5e-06,
      "loss": 0.4634,
      "step": 6560
    },
    {
      "epoch": 1.4217701796148021,
      "grad_norm": 0.7539199622313375,
      "learning_rate": 5e-06,
      "loss": 0.4675,
      "step": 6570
    },
    {
      "epoch": 1.4239342133737285,
      "grad_norm": 0.7650706870481234,
      "learning_rate": 5e-06,
      "loss": 0.4635,
      "step": 6580
    },
    {
      "epoch": 1.4260982471326553,
      "grad_norm": 0.7303061857380676,
      "learning_rate": 5e-06,
      "loss": 0.4528,
      "step": 6590
    },
    {
      "epoch": 1.428262280891582,
      "grad_norm": 0.7448893505433495,
      "learning_rate": 5e-06,
      "loss": 0.4664,
      "step": 6600
    },
    {
      "epoch": 1.4304263146505085,
      "grad_norm": 0.8093720993778148,
      "learning_rate": 5e-06,
      "loss": 0.4847,
      "step": 6610
    },
    {
      "epoch": 1.432590348409435,
      "grad_norm": 0.7534159752837184,
      "learning_rate": 5e-06,
      "loss": 0.4787,
      "step": 6620
    },
    {
      "epoch": 1.4347543821683617,
      "grad_norm": 0.7650400003197116,
      "learning_rate": 5e-06,
      "loss": 0.4682,
      "step": 6630
    },
    {
      "epoch": 1.4369184159272885,
      "grad_norm": 0.7965801032383258,
      "learning_rate": 5e-06,
      "loss": 0.4766,
      "step": 6640
    },
    {
      "epoch": 1.4390824496862151,
      "grad_norm": 0.8096153226598906,
      "learning_rate": 5e-06,
      "loss": 0.4708,
      "step": 6650
    },
    {
      "epoch": 1.4412464834451417,
      "grad_norm": 0.7872971777972805,
      "learning_rate": 5e-06,
      "loss": 0.4651,
      "step": 6660
    },
    {
      "epoch": 1.4434105172040683,
      "grad_norm": 0.7309523643397555,
      "learning_rate": 5e-06,
      "loss": 0.448,
      "step": 6670
    },
    {
      "epoch": 1.445574550962995,
      "grad_norm": 0.7739858235533356,
      "learning_rate": 5e-06,
      "loss": 0.463,
      "step": 6680
    },
    {
      "epoch": 1.4477385847219217,
      "grad_norm": 0.7215324022117396,
      "learning_rate": 5e-06,
      "loss": 0.4534,
      "step": 6690
    },
    {
      "epoch": 1.4499026184808483,
      "grad_norm": 0.8295795696846178,
      "learning_rate": 5e-06,
      "loss": 0.455,
      "step": 6700
    },
    {
      "epoch": 1.452066652239775,
      "grad_norm": 0.796699047578508,
      "learning_rate": 5e-06,
      "loss": 0.4675,
      "step": 6710
    },
    {
      "epoch": 1.4542306859987015,
      "grad_norm": 0.7661917101852191,
      "learning_rate": 5e-06,
      "loss": 0.4612,
      "step": 6720
    },
    {
      "epoch": 1.4563947197576281,
      "grad_norm": 0.7292322135619679,
      "learning_rate": 5e-06,
      "loss": 0.4723,
      "step": 6730
    },
    {
      "epoch": 1.458558753516555,
      "grad_norm": 0.7274092722777359,
      "learning_rate": 5e-06,
      "loss": 0.4668,
      "step": 6740
    },
    {
      "epoch": 1.4607227872754815,
      "grad_norm": 0.7908003272383125,
      "learning_rate": 5e-06,
      "loss": 0.4707,
      "step": 6750
    },
    {
      "epoch": 1.4628868210344081,
      "grad_norm": 0.7645468887204627,
      "learning_rate": 5e-06,
      "loss": 0.4681,
      "step": 6760
    },
    {
      "epoch": 1.4650508547933347,
      "grad_norm": 0.770270828639935,
      "learning_rate": 5e-06,
      "loss": 0.471,
      "step": 6770
    },
    {
      "epoch": 1.4672148885522613,
      "grad_norm": 0.7506554829741943,
      "learning_rate": 5e-06,
      "loss": 0.4792,
      "step": 6780
    },
    {
      "epoch": 1.4693789223111882,
      "grad_norm": 0.8131214331089905,
      "learning_rate": 5e-06,
      "loss": 0.4661,
      "step": 6790
    },
    {
      "epoch": 1.4715429560701148,
      "grad_norm": 0.7015838324010227,
      "learning_rate": 5e-06,
      "loss": 0.4711,
      "step": 6800
    },
    {
      "epoch": 1.4737069898290414,
      "grad_norm": 0.7861814435987762,
      "learning_rate": 5e-06,
      "loss": 0.4613,
      "step": 6810
    },
    {
      "epoch": 1.475871023587968,
      "grad_norm": 0.7591866835488571,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 6820
    },
    {
      "epoch": 1.4780350573468946,
      "grad_norm": 0.7404595294517535,
      "learning_rate": 5e-06,
      "loss": 0.461,
      "step": 6830
    },
    {
      "epoch": 1.4801990911058214,
      "grad_norm": 0.7697305030650266,
      "learning_rate": 5e-06,
      "loss": 0.4598,
      "step": 6840
    },
    {
      "epoch": 1.482363124864748,
      "grad_norm": 0.784141929033681,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 6850
    },
    {
      "epoch": 1.4845271586236746,
      "grad_norm": 0.8136152308153741,
      "learning_rate": 5e-06,
      "loss": 0.4552,
      "step": 6860
    },
    {
      "epoch": 1.4866911923826012,
      "grad_norm": 0.7345552940901736,
      "learning_rate": 5e-06,
      "loss": 0.4767,
      "step": 6870
    },
    {
      "epoch": 1.4888552261415278,
      "grad_norm": 0.7776472050120381,
      "learning_rate": 5e-06,
      "loss": 0.4744,
      "step": 6880
    },
    {
      "epoch": 1.4910192599004546,
      "grad_norm": 0.751123599938991,
      "learning_rate": 5e-06,
      "loss": 0.4678,
      "step": 6890
    },
    {
      "epoch": 1.493183293659381,
      "grad_norm": 0.7437167540645744,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 6900
    },
    {
      "epoch": 1.4953473274183078,
      "grad_norm": 0.7721598263875197,
      "learning_rate": 5e-06,
      "loss": 0.482,
      "step": 6910
    },
    {
      "epoch": 1.4975113611772344,
      "grad_norm": 0.7817667866593803,
      "learning_rate": 5e-06,
      "loss": 0.4666,
      "step": 6920
    },
    {
      "epoch": 1.499675394936161,
      "grad_norm": 0.7694139195197549,
      "learning_rate": 5e-06,
      "loss": 0.466,
      "step": 6930
    },
    {
      "epoch": 1.5018394286950878,
      "grad_norm": 0.7795215699875335,
      "learning_rate": 5e-06,
      "loss": 0.4632,
      "step": 6940
    },
    {
      "epoch": 1.5040034624540142,
      "grad_norm": 0.7378131555326808,
      "learning_rate": 5e-06,
      "loss": 0.4703,
      "step": 6950
    },
    {
      "epoch": 1.506167496212941,
      "grad_norm": 0.7502895598759765,
      "learning_rate": 5e-06,
      "loss": 0.4675,
      "step": 6960
    },
    {
      "epoch": 1.5083315299718676,
      "grad_norm": 0.8323332633475088,
      "learning_rate": 5e-06,
      "loss": 0.4803,
      "step": 6970
    },
    {
      "epoch": 1.5104955637307942,
      "grad_norm": 0.7892580411151743,
      "learning_rate": 5e-06,
      "loss": 0.4765,
      "step": 6980
    },
    {
      "epoch": 1.5126595974897208,
      "grad_norm": 0.7875592663001918,
      "learning_rate": 5e-06,
      "loss": 0.4679,
      "step": 6990
    },
    {
      "epoch": 1.5148236312486474,
      "grad_norm": 0.7634571975953766,
      "learning_rate": 5e-06,
      "loss": 0.47,
      "step": 7000
    },
    {
      "epoch": 1.5169876650075742,
      "grad_norm": 0.7756829723486675,
      "learning_rate": 5e-06,
      "loss": 0.464,
      "step": 7010
    },
    {
      "epoch": 1.5191516987665008,
      "grad_norm": 0.7581561828919042,
      "learning_rate": 5e-06,
      "loss": 0.4561,
      "step": 7020
    },
    {
      "epoch": 1.5213157325254274,
      "grad_norm": 0.8138028391909014,
      "learning_rate": 5e-06,
      "loss": 0.4738,
      "step": 7030
    },
    {
      "epoch": 1.523479766284354,
      "grad_norm": 0.7713503967397953,
      "learning_rate": 5e-06,
      "loss": 0.4536,
      "step": 7040
    },
    {
      "epoch": 1.5256438000432806,
      "grad_norm": 0.7721407676831082,
      "learning_rate": 5e-06,
      "loss": 0.477,
      "step": 7050
    },
    {
      "epoch": 1.5278078338022074,
      "grad_norm": 0.7835072836836937,
      "learning_rate": 5e-06,
      "loss": 0.4781,
      "step": 7060
    },
    {
      "epoch": 1.5299718675611338,
      "grad_norm": 0.8062247813326742,
      "learning_rate": 5e-06,
      "loss": 0.4729,
      "step": 7070
    },
    {
      "epoch": 1.5321359013200606,
      "grad_norm": 0.7136619017556197,
      "learning_rate": 5e-06,
      "loss": 0.4531,
      "step": 7080
    },
    {
      "epoch": 1.5342999350789872,
      "grad_norm": 0.78043707583408,
      "learning_rate": 5e-06,
      "loss": 0.465,
      "step": 7090
    },
    {
      "epoch": 1.5364639688379138,
      "grad_norm": 0.7933315893100612,
      "learning_rate": 5e-06,
      "loss": 0.4655,
      "step": 7100
    },
    {
      "epoch": 1.5386280025968406,
      "grad_norm": 0.7476138813332562,
      "learning_rate": 5e-06,
      "loss": 0.4642,
      "step": 7110
    },
    {
      "epoch": 1.540792036355767,
      "grad_norm": 0.791596305147064,
      "learning_rate": 5e-06,
      "loss": 0.4656,
      "step": 7120
    },
    {
      "epoch": 1.5429560701146938,
      "grad_norm": 0.7765122295933571,
      "learning_rate": 5e-06,
      "loss": 0.4578,
      "step": 7130
    },
    {
      "epoch": 1.5451201038736204,
      "grad_norm": 0.7721038031057484,
      "learning_rate": 5e-06,
      "loss": 0.4724,
      "step": 7140
    },
    {
      "epoch": 1.547284137632547,
      "grad_norm": 0.7889869463184822,
      "learning_rate": 5e-06,
      "loss": 0.472,
      "step": 7150
    },
    {
      "epoch": 1.5494481713914738,
      "grad_norm": 0.7326495041461524,
      "learning_rate": 5e-06,
      "loss": 0.4587,
      "step": 7160
    },
    {
      "epoch": 1.5516122051504002,
      "grad_norm": 0.7713670879576949,
      "learning_rate": 5e-06,
      "loss": 0.472,
      "step": 7170
    },
    {
      "epoch": 1.553776238909327,
      "grad_norm": 0.7864345761195741,
      "learning_rate": 5e-06,
      "loss": 0.4708,
      "step": 7180
    },
    {
      "epoch": 1.5559402726682536,
      "grad_norm": 0.7669143468828357,
      "learning_rate": 5e-06,
      "loss": 0.4622,
      "step": 7190
    },
    {
      "epoch": 1.5581043064271802,
      "grad_norm": 0.8036141108986897,
      "learning_rate": 5e-06,
      "loss": 0.4779,
      "step": 7200
    },
    {
      "epoch": 1.560268340186107,
      "grad_norm": 0.7606518551283948,
      "learning_rate": 5e-06,
      "loss": 0.4677,
      "step": 7210
    },
    {
      "epoch": 1.5624323739450334,
      "grad_norm": 0.7411431196268208,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 7220
    },
    {
      "epoch": 1.5645964077039602,
      "grad_norm": 0.7463278194705449,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 7230
    },
    {
      "epoch": 1.5667604414628868,
      "grad_norm": 0.7916035934870831,
      "learning_rate": 5e-06,
      "loss": 0.4844,
      "step": 7240
    },
    {
      "epoch": 1.5689244752218134,
      "grad_norm": 0.8240660319984418,
      "learning_rate": 5e-06,
      "loss": 0.4683,
      "step": 7250
    },
    {
      "epoch": 1.5710885089807403,
      "grad_norm": 0.800318824146685,
      "learning_rate": 5e-06,
      "loss": 0.4633,
      "step": 7260
    },
    {
      "epoch": 1.5732525427396666,
      "grad_norm": 0.8094080756414094,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 7270
    },
    {
      "epoch": 1.5754165764985935,
      "grad_norm": 0.774844371732236,
      "learning_rate": 5e-06,
      "loss": 0.4587,
      "step": 7280
    },
    {
      "epoch": 1.57758061025752,
      "grad_norm": 0.8210197993957168,
      "learning_rate": 5e-06,
      "loss": 0.4676,
      "step": 7290
    },
    {
      "epoch": 1.5797446440164467,
      "grad_norm": 0.7926225329081739,
      "learning_rate": 5e-06,
      "loss": 0.4746,
      "step": 7300
    },
    {
      "epoch": 1.5819086777753733,
      "grad_norm": 0.7714683122387773,
      "learning_rate": 5e-06,
      "loss": 0.4696,
      "step": 7310
    },
    {
      "epoch": 1.5840727115342998,
      "grad_norm": 0.7668988486606838,
      "learning_rate": 5e-06,
      "loss": 0.4637,
      "step": 7320
    },
    {
      "epoch": 1.5862367452932267,
      "grad_norm": 0.8068671779651346,
      "learning_rate": 5e-06,
      "loss": 0.4625,
      "step": 7330
    },
    {
      "epoch": 1.5884007790521533,
      "grad_norm": 0.7503714168675623,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 7340
    },
    {
      "epoch": 1.5905648128110799,
      "grad_norm": 0.766762311923028,
      "learning_rate": 5e-06,
      "loss": 0.4607,
      "step": 7350
    },
    {
      "epoch": 1.5927288465700065,
      "grad_norm": 0.7400989288703183,
      "learning_rate": 5e-06,
      "loss": 0.4586,
      "step": 7360
    },
    {
      "epoch": 1.594892880328933,
      "grad_norm": 0.7420053327580863,
      "learning_rate": 5e-06,
      "loss": 0.4611,
      "step": 7370
    },
    {
      "epoch": 1.5970569140878599,
      "grad_norm": 0.7848315895928035,
      "learning_rate": 5e-06,
      "loss": 0.4622,
      "step": 7380
    },
    {
      "epoch": 1.5992209478467863,
      "grad_norm": 0.7844109009233191,
      "learning_rate": 5e-06,
      "loss": 0.4726,
      "step": 7390
    },
    {
      "epoch": 1.601384981605713,
      "grad_norm": 0.717574518313775,
      "learning_rate": 5e-06,
      "loss": 0.4768,
      "step": 7400
    },
    {
      "epoch": 1.6035490153646397,
      "grad_norm": 0.7684187592976545,
      "learning_rate": 5e-06,
      "loss": 0.4696,
      "step": 7410
    },
    {
      "epoch": 1.6057130491235663,
      "grad_norm": 0.7309659679104782,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 7420
    },
    {
      "epoch": 1.607877082882493,
      "grad_norm": 0.7644663148759375,
      "learning_rate": 5e-06,
      "loss": 0.4772,
      "step": 7430
    },
    {
      "epoch": 1.6100411166414195,
      "grad_norm": 0.7764874598382566,
      "learning_rate": 5e-06,
      "loss": 0.4646,
      "step": 7440
    },
    {
      "epoch": 1.6122051504003463,
      "grad_norm": 0.7620772054300151,
      "learning_rate": 5e-06,
      "loss": 0.4586,
      "step": 7450
    },
    {
      "epoch": 1.6143691841592729,
      "grad_norm": 0.7820555180817107,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 7460
    },
    {
      "epoch": 1.6165332179181995,
      "grad_norm": 0.7494625835808725,
      "learning_rate": 5e-06,
      "loss": 0.4585,
      "step": 7470
    },
    {
      "epoch": 1.6186972516771263,
      "grad_norm": 0.7521807522464258,
      "learning_rate": 5e-06,
      "loss": 0.4584,
      "step": 7480
    },
    {
      "epoch": 1.6208612854360527,
      "grad_norm": 0.7581980433304982,
      "learning_rate": 5e-06,
      "loss": 0.4588,
      "step": 7490
    },
    {
      "epoch": 1.6230253191949795,
      "grad_norm": 0.7627156140971617,
      "learning_rate": 5e-06,
      "loss": 0.4741,
      "step": 7500
    },
    {
      "epoch": 1.625189352953906,
      "grad_norm": 0.7905461292081447,
      "learning_rate": 5e-06,
      "loss": 0.4658,
      "step": 7510
    },
    {
      "epoch": 1.6273533867128327,
      "grad_norm": 0.8124704291753462,
      "learning_rate": 5e-06,
      "loss": 0.4735,
      "step": 7520
    },
    {
      "epoch": 1.6295174204717595,
      "grad_norm": 0.749724335823781,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 7530
    },
    {
      "epoch": 1.631681454230686,
      "grad_norm": 0.7544817597703268,
      "learning_rate": 5e-06,
      "loss": 0.4664,
      "step": 7540
    },
    {
      "epoch": 1.6338454879896127,
      "grad_norm": 0.7703639179989715,
      "learning_rate": 5e-06,
      "loss": 0.4514,
      "step": 7550
    },
    {
      "epoch": 1.6360095217485393,
      "grad_norm": 0.790414972968879,
      "learning_rate": 5e-06,
      "loss": 0.4647,
      "step": 7560
    },
    {
      "epoch": 1.638173555507466,
      "grad_norm": 0.777968424055043,
      "learning_rate": 5e-06,
      "loss": 0.4687,
      "step": 7570
    },
    {
      "epoch": 1.6403375892663925,
      "grad_norm": 0.8168837011418277,
      "learning_rate": 5e-06,
      "loss": 0.4748,
      "step": 7580
    },
    {
      "epoch": 1.642501623025319,
      "grad_norm": 0.7880218866379387,
      "learning_rate": 5e-06,
      "loss": 0.463,
      "step": 7590
    },
    {
      "epoch": 1.644665656784246,
      "grad_norm": 0.7622801106079738,
      "learning_rate": 5e-06,
      "loss": 0.4687,
      "step": 7600
    },
    {
      "epoch": 1.6468296905431725,
      "grad_norm": 0.7536846157018885,
      "learning_rate": 5e-06,
      "loss": 0.4646,
      "step": 7610
    },
    {
      "epoch": 1.6489937243020991,
      "grad_norm": 0.7798435120288673,
      "learning_rate": 5e-06,
      "loss": 0.4709,
      "step": 7620
    },
    {
      "epoch": 1.6511577580610257,
      "grad_norm": 0.7550086843977464,
      "learning_rate": 5e-06,
      "loss": 0.4641,
      "step": 7630
    },
    {
      "epoch": 1.6533217918199523,
      "grad_norm": 0.7528156940459134,
      "learning_rate": 5e-06,
      "loss": 0.4668,
      "step": 7640
    },
    {
      "epoch": 1.6554858255788791,
      "grad_norm": 0.7718809368052605,
      "learning_rate": 5e-06,
      "loss": 0.4616,
      "step": 7650
    },
    {
      "epoch": 1.6576498593378055,
      "grad_norm": 0.7784564880148578,
      "learning_rate": 5e-06,
      "loss": 0.4667,
      "step": 7660
    },
    {
      "epoch": 1.6598138930967323,
      "grad_norm": 0.7397166601008868,
      "learning_rate": 5e-06,
      "loss": 0.4691,
      "step": 7670
    },
    {
      "epoch": 1.661977926855659,
      "grad_norm": 0.7881087574473704,
      "learning_rate": 5e-06,
      "loss": 0.4562,
      "step": 7680
    },
    {
      "epoch": 1.6641419606145855,
      "grad_norm": 0.7536684489497238,
      "learning_rate": 5e-06,
      "loss": 0.4581,
      "step": 7690
    },
    {
      "epoch": 1.6663059943735123,
      "grad_norm": 0.7981078552903653,
      "learning_rate": 5e-06,
      "loss": 0.4644,
      "step": 7700
    },
    {
      "epoch": 1.6684700281324387,
      "grad_norm": 0.7019669743910957,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 7710
    },
    {
      "epoch": 1.6706340618913655,
      "grad_norm": 0.7977655257782118,
      "learning_rate": 5e-06,
      "loss": 0.4721,
      "step": 7720
    },
    {
      "epoch": 1.6727980956502921,
      "grad_norm": 0.8017386100927005,
      "learning_rate": 5e-06,
      "loss": 0.469,
      "step": 7730
    },
    {
      "epoch": 1.6749621294092187,
      "grad_norm": 0.771451526799465,
      "learning_rate": 5e-06,
      "loss": 0.4596,
      "step": 7740
    },
    {
      "epoch": 1.6771261631681456,
      "grad_norm": 0.7227699836985734,
      "learning_rate": 5e-06,
      "loss": 0.4546,
      "step": 7750
    },
    {
      "epoch": 1.679290196927072,
      "grad_norm": 0.7676454649666475,
      "learning_rate": 5e-06,
      "loss": 0.4663,
      "step": 7760
    },
    {
      "epoch": 1.6814542306859988,
      "grad_norm": 0.8226610282436956,
      "learning_rate": 5e-06,
      "loss": 0.4745,
      "step": 7770
    },
    {
      "epoch": 1.6836182644449253,
      "grad_norm": 0.7771996492106517,
      "learning_rate": 5e-06,
      "loss": 0.459,
      "step": 7780
    },
    {
      "epoch": 1.685782298203852,
      "grad_norm": 0.7845408388142724,
      "learning_rate": 5e-06,
      "loss": 0.4662,
      "step": 7790
    },
    {
      "epoch": 1.6879463319627788,
      "grad_norm": 0.8358661114189557,
      "learning_rate": 5e-06,
      "loss": 0.4616,
      "step": 7800
    },
    {
      "epoch": 1.6901103657217051,
      "grad_norm": 0.745445671430633,
      "learning_rate": 5e-06,
      "loss": 0.4633,
      "step": 7810
    },
    {
      "epoch": 1.692274399480632,
      "grad_norm": 0.7806093109649215,
      "learning_rate": 5e-06,
      "loss": 0.4713,
      "step": 7820
    },
    {
      "epoch": 1.6944384332395586,
      "grad_norm": 0.7931013296476084,
      "learning_rate": 5e-06,
      "loss": 0.4518,
      "step": 7830
    },
    {
      "epoch": 1.6966024669984852,
      "grad_norm": 0.7523743415278165,
      "learning_rate": 5e-06,
      "loss": 0.468,
      "step": 7840
    },
    {
      "epoch": 1.698766500757412,
      "grad_norm": 0.7657925402229067,
      "learning_rate": 5e-06,
      "loss": 0.4653,
      "step": 7850
    },
    {
      "epoch": 1.7009305345163384,
      "grad_norm": 0.7623478102612719,
      "learning_rate": 5e-06,
      "loss": 0.4714,
      "step": 7860
    },
    {
      "epoch": 1.7030945682752652,
      "grad_norm": 0.7981733209185972,
      "learning_rate": 5e-06,
      "loss": 0.4699,
      "step": 7870
    },
    {
      "epoch": 1.7052586020341918,
      "grad_norm": 0.7781587550663129,
      "learning_rate": 5e-06,
      "loss": 0.4765,
      "step": 7880
    },
    {
      "epoch": 1.7074226357931184,
      "grad_norm": 0.7587942982310993,
      "learning_rate": 5e-06,
      "loss": 0.4771,
      "step": 7890
    },
    {
      "epoch": 1.709586669552045,
      "grad_norm": 0.7698688590200006,
      "learning_rate": 5e-06,
      "loss": 0.4647,
      "step": 7900
    },
    {
      "epoch": 1.7117507033109716,
      "grad_norm": 0.804203999005989,
      "learning_rate": 5e-06,
      "loss": 0.4686,
      "step": 7910
    },
    {
      "epoch": 1.7139147370698984,
      "grad_norm": 0.7911772144270678,
      "learning_rate": 5e-06,
      "loss": 0.4639,
      "step": 7920
    },
    {
      "epoch": 1.716078770828825,
      "grad_norm": 0.7836618578593015,
      "learning_rate": 5e-06,
      "loss": 0.4666,
      "step": 7930
    },
    {
      "epoch": 1.7182428045877516,
      "grad_norm": 0.788575508265406,
      "learning_rate": 5e-06,
      "loss": 0.4729,
      "step": 7940
    },
    {
      "epoch": 1.7204068383466782,
      "grad_norm": 0.7861934781471758,
      "learning_rate": 5e-06,
      "loss": 0.4555,
      "step": 7950
    },
    {
      "epoch": 1.7225708721056048,
      "grad_norm": 0.7536239150523527,
      "learning_rate": 5e-06,
      "loss": 0.4694,
      "step": 7960
    },
    {
      "epoch": 1.7247349058645316,
      "grad_norm": 0.7316152360900233,
      "learning_rate": 5e-06,
      "loss": 0.4654,
      "step": 7970
    },
    {
      "epoch": 1.726898939623458,
      "grad_norm": 0.7959312951820343,
      "learning_rate": 5e-06,
      "loss": 0.4742,
      "step": 7980
    },
    {
      "epoch": 1.7290629733823848,
      "grad_norm": 0.7328251440184733,
      "learning_rate": 5e-06,
      "loss": 0.4636,
      "step": 7990
    },
    {
      "epoch": 1.7312270071413114,
      "grad_norm": 0.8721623540661089,
      "learning_rate": 5e-06,
      "loss": 0.4727,
      "step": 8000
    },
    {
      "epoch": 1.733391040900238,
      "grad_norm": 0.7788156040311068,
      "learning_rate": 5e-06,
      "loss": 0.479,
      "step": 8010
    },
    {
      "epoch": 1.7355550746591648,
      "grad_norm": 0.7352632852244723,
      "learning_rate": 5e-06,
      "loss": 0.4694,
      "step": 8020
    },
    {
      "epoch": 1.7377191084180912,
      "grad_norm": 0.8294281676756476,
      "learning_rate": 5e-06,
      "loss": 0.4756,
      "step": 8030
    },
    {
      "epoch": 1.739883142177018,
      "grad_norm": 0.7872766681208978,
      "learning_rate": 5e-06,
      "loss": 0.4714,
      "step": 8040
    },
    {
      "epoch": 1.7420471759359446,
      "grad_norm": 0.7549853443648625,
      "learning_rate": 5e-06,
      "loss": 0.4793,
      "step": 8050
    },
    {
      "epoch": 1.7442112096948712,
      "grad_norm": 0.7752211889605723,
      "learning_rate": 5e-06,
      "loss": 0.4568,
      "step": 8060
    },
    {
      "epoch": 1.746375243453798,
      "grad_norm": 0.7656375809085874,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 8070
    },
    {
      "epoch": 1.7485392772127244,
      "grad_norm": 0.7758127700797092,
      "learning_rate": 5e-06,
      "loss": 0.4583,
      "step": 8080
    },
    {
      "epoch": 1.7507033109716512,
      "grad_norm": 0.7634613626649636,
      "learning_rate": 5e-06,
      "loss": 0.4595,
      "step": 8090
    },
    {
      "epoch": 1.7528673447305778,
      "grad_norm": 0.7698077132883445,
      "learning_rate": 5e-06,
      "loss": 0.4717,
      "step": 8100
    },
    {
      "epoch": 1.7550313784895044,
      "grad_norm": 0.7383242906768342,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 8110
    },
    {
      "epoch": 1.7571954122484312,
      "grad_norm": 0.823790637914916,
      "learning_rate": 5e-06,
      "loss": 0.4617,
      "step": 8120
    },
    {
      "epoch": 1.7593594460073576,
      "grad_norm": 0.7581731295515508,
      "learning_rate": 5e-06,
      "loss": 0.4599,
      "step": 8130
    },
    {
      "epoch": 1.7615234797662844,
      "grad_norm": 0.7863922041923643,
      "learning_rate": 5e-06,
      "loss": 0.4633,
      "step": 8140
    },
    {
      "epoch": 1.763687513525211,
      "grad_norm": 0.7258530440371272,
      "learning_rate": 5e-06,
      "loss": 0.4695,
      "step": 8150
    },
    {
      "epoch": 1.7658515472841376,
      "grad_norm": 0.7617292499788662,
      "learning_rate": 5e-06,
      "loss": 0.4584,
      "step": 8160
    },
    {
      "epoch": 1.7680155810430642,
      "grad_norm": 0.7835216384518215,
      "learning_rate": 5e-06,
      "loss": 0.4776,
      "step": 8170
    },
    {
      "epoch": 1.7701796148019908,
      "grad_norm": 0.7905903337960178,
      "learning_rate": 5e-06,
      "loss": 0.4575,
      "step": 8180
    },
    {
      "epoch": 1.7723436485609176,
      "grad_norm": 0.7470273698032939,
      "learning_rate": 5e-06,
      "loss": 0.4519,
      "step": 8190
    },
    {
      "epoch": 1.7745076823198442,
      "grad_norm": 0.7672368207649455,
      "learning_rate": 5e-06,
      "loss": 0.4648,
      "step": 8200
    },
    {
      "epoch": 1.7766717160787708,
      "grad_norm": 0.7316674700923429,
      "learning_rate": 5e-06,
      "loss": 0.4724,
      "step": 8210
    },
    {
      "epoch": 1.7788357498376974,
      "grad_norm": 0.7360630882150441,
      "learning_rate": 5e-06,
      "loss": 0.4638,
      "step": 8220
    },
    {
      "epoch": 1.780999783596624,
      "grad_norm": 0.781016234417425,
      "learning_rate": 5e-06,
      "loss": 0.4728,
      "step": 8230
    },
    {
      "epoch": 1.7831638173555509,
      "grad_norm": 0.7876473046986651,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 8240
    },
    {
      "epoch": 1.7853278511144772,
      "grad_norm": 0.7404645309296104,
      "learning_rate": 5e-06,
      "loss": 0.4624,
      "step": 8250
    },
    {
      "epoch": 1.787491884873404,
      "grad_norm": 0.7684119788472853,
      "learning_rate": 5e-06,
      "loss": 0.4635,
      "step": 8260
    },
    {
      "epoch": 1.7896559186323306,
      "grad_norm": 0.7663695712294315,
      "learning_rate": 5e-06,
      "loss": 0.465,
      "step": 8270
    },
    {
      "epoch": 1.7918199523912572,
      "grad_norm": 0.7590055968711897,
      "learning_rate": 5e-06,
      "loss": 0.4643,
      "step": 8280
    },
    {
      "epoch": 1.793983986150184,
      "grad_norm": 0.7895074154532604,
      "learning_rate": 5e-06,
      "loss": 0.4671,
      "step": 8290
    },
    {
      "epoch": 1.7961480199091104,
      "grad_norm": 0.7962575230042065,
      "learning_rate": 5e-06,
      "loss": 0.472,
      "step": 8300
    },
    {
      "epoch": 1.7983120536680373,
      "grad_norm": 0.7478573445116543,
      "learning_rate": 5e-06,
      "loss": 0.4626,
      "step": 8310
    },
    {
      "epoch": 1.8004760874269639,
      "grad_norm": 0.762202720740558,
      "learning_rate": 5e-06,
      "loss": 0.4515,
      "step": 8320
    },
    {
      "epoch": 1.8026401211858905,
      "grad_norm": 0.8404740937407859,
      "learning_rate": 5e-06,
      "loss": 0.4674,
      "step": 8330
    },
    {
      "epoch": 1.8048041549448173,
      "grad_norm": 0.7952968170552206,
      "learning_rate": 5e-06,
      "loss": 0.4619,
      "step": 8340
    },
    {
      "epoch": 1.8069681887037436,
      "grad_norm": 0.7820220237908069,
      "learning_rate": 5e-06,
      "loss": 0.4667,
      "step": 8350
    },
    {
      "epoch": 1.8091322224626705,
      "grad_norm": 0.7888823174892542,
      "learning_rate": 5e-06,
      "loss": 0.4732,
      "step": 8360
    },
    {
      "epoch": 1.811296256221597,
      "grad_norm": 0.7350957126888091,
      "learning_rate": 5e-06,
      "loss": 0.4532,
      "step": 8370
    },
    {
      "epoch": 1.8134602899805237,
      "grad_norm": 0.6957708222610044,
      "learning_rate": 5e-06,
      "loss": 0.4497,
      "step": 8380
    },
    {
      "epoch": 1.8156243237394505,
      "grad_norm": 0.8053515701021636,
      "learning_rate": 5e-06,
      "loss": 0.4785,
      "step": 8390
    },
    {
      "epoch": 1.8177883574983769,
      "grad_norm": 0.6995897457091669,
      "learning_rate": 5e-06,
      "loss": 0.4551,
      "step": 8400
    },
    {
      "epoch": 1.8199523912573037,
      "grad_norm": 0.7753814167641445,
      "learning_rate": 5e-06,
      "loss": 0.4621,
      "step": 8410
    },
    {
      "epoch": 1.8221164250162303,
      "grad_norm": 0.7661880273668203,
      "learning_rate": 5e-06,
      "loss": 0.4609,
      "step": 8420
    },
    {
      "epoch": 1.8242804587751569,
      "grad_norm": 0.7609799960654735,
      "learning_rate": 5e-06,
      "loss": 0.4736,
      "step": 8430
    },
    {
      "epoch": 1.8264444925340837,
      "grad_norm": 0.7841320073323229,
      "learning_rate": 5e-06,
      "loss": 0.4575,
      "step": 8440
    },
    {
      "epoch": 1.82860852629301,
      "grad_norm": 0.8063573574026484,
      "learning_rate": 5e-06,
      "loss": 0.4617,
      "step": 8450
    },
    {
      "epoch": 1.830772560051937,
      "grad_norm": 0.7854293405760782,
      "learning_rate": 5e-06,
      "loss": 0.4777,
      "step": 8460
    },
    {
      "epoch": 1.8329365938108635,
      "grad_norm": 0.7408682039776677,
      "learning_rate": 5e-06,
      "loss": 0.4614,
      "step": 8470
    },
    {
      "epoch": 1.83510062756979,
      "grad_norm": 0.7231058234976815,
      "learning_rate": 5e-06,
      "loss": 0.4624,
      "step": 8480
    },
    {
      "epoch": 1.8372646613287167,
      "grad_norm": 0.800682041278424,
      "learning_rate": 5e-06,
      "loss": 0.4641,
      "step": 8490
    },
    {
      "epoch": 1.8394286950876433,
      "grad_norm": 0.7673472834532682,
      "learning_rate": 5e-06,
      "loss": 0.4606,
      "step": 8500
    },
    {
      "epoch": 1.84159272884657,
      "grad_norm": 0.7619493537629672,
      "learning_rate": 5e-06,
      "loss": 0.4706,
      "step": 8510
    },
    {
      "epoch": 1.8437567626054965,
      "grad_norm": 0.7972147582688511,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 8520
    },
    {
      "epoch": 1.8459207963644233,
      "grad_norm": 0.7625120492458264,
      "learning_rate": 5e-06,
      "loss": 0.4571,
      "step": 8530
    },
    {
      "epoch": 1.84808483012335,
      "grad_norm": 0.8283056055978331,
      "learning_rate": 5e-06,
      "loss": 0.4739,
      "step": 8540
    },
    {
      "epoch": 1.8502488638822765,
      "grad_norm": 0.7573851435355785,
      "learning_rate": 5e-06,
      "loss": 0.4671,
      "step": 8550
    },
    {
      "epoch": 1.8524128976412033,
      "grad_norm": 0.8293614574747807,
      "learning_rate": 5e-06,
      "loss": 0.4652,
      "step": 8560
    },
    {
      "epoch": 1.8545769314001297,
      "grad_norm": 0.7412074084724187,
      "learning_rate": 5e-06,
      "loss": 0.4592,
      "step": 8570
    },
    {
      "epoch": 1.8567409651590565,
      "grad_norm": 0.8475011905897842,
      "learning_rate": 5e-06,
      "loss": 0.4588,
      "step": 8580
    },
    {
      "epoch": 1.858904998917983,
      "grad_norm": 0.7431837931733989,
      "learning_rate": 5e-06,
      "loss": 0.453,
      "step": 8590
    },
    {
      "epoch": 1.8610690326769097,
      "grad_norm": 0.743103879188485,
      "learning_rate": 5e-06,
      "loss": 0.4685,
      "step": 8600
    },
    {
      "epoch": 1.8632330664358365,
      "grad_norm": 0.7512142776663842,
      "learning_rate": 5e-06,
      "loss": 0.4685,
      "step": 8610
    },
    {
      "epoch": 1.865397100194763,
      "grad_norm": 0.733739499974568,
      "learning_rate": 5e-06,
      "loss": 0.4625,
      "step": 8620
    },
    {
      "epoch": 1.8675611339536897,
      "grad_norm": 0.7976859851306687,
      "learning_rate": 5e-06,
      "loss": 0.4676,
      "step": 8630
    },
    {
      "epoch": 1.8697251677126163,
      "grad_norm": 0.7661212684653242,
      "learning_rate": 5e-06,
      "loss": 0.475,
      "step": 8640
    },
    {
      "epoch": 1.871889201471543,
      "grad_norm": 0.771734215920023,
      "learning_rate": 5e-06,
      "loss": 0.4621,
      "step": 8650
    },
    {
      "epoch": 1.8740532352304697,
      "grad_norm": 0.766268946678346,
      "learning_rate": 5e-06,
      "loss": 0.4531,
      "step": 8660
    },
    {
      "epoch": 1.8762172689893961,
      "grad_norm": 0.7754183196315589,
      "learning_rate": 5e-06,
      "loss": 0.4673,
      "step": 8670
    },
    {
      "epoch": 1.878381302748323,
      "grad_norm": 0.7534048500045719,
      "learning_rate": 5e-06,
      "loss": 0.467,
      "step": 8680
    },
    {
      "epoch": 1.8805453365072495,
      "grad_norm": 0.7764377429907858,
      "learning_rate": 5e-06,
      "loss": 0.4579,
      "step": 8690
    },
    {
      "epoch": 1.8827093702661761,
      "grad_norm": 0.7698958247598046,
      "learning_rate": 5e-06,
      "loss": 0.4786,
      "step": 8700
    },
    {
      "epoch": 1.884873404025103,
      "grad_norm": 0.7651418118029404,
      "learning_rate": 5e-06,
      "loss": 0.4594,
      "step": 8710
    },
    {
      "epoch": 1.8870374377840293,
      "grad_norm": 0.7590832586194605,
      "learning_rate": 5e-06,
      "loss": 0.46,
      "step": 8720
    },
    {
      "epoch": 1.8892014715429561,
      "grad_norm": 0.7731735855097023,
      "learning_rate": 5e-06,
      "loss": 0.4727,
      "step": 8730
    },
    {
      "epoch": 1.8913655053018827,
      "grad_norm": 0.7856091416226185,
      "learning_rate": 5e-06,
      "loss": 0.4503,
      "step": 8740
    },
    {
      "epoch": 1.8935295390608093,
      "grad_norm": 0.7722457110486987,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 8750
    },
    {
      "epoch": 1.895693572819736,
      "grad_norm": 0.8198508677971693,
      "learning_rate": 5e-06,
      "loss": 0.4673,
      "step": 8760
    },
    {
      "epoch": 1.8978576065786625,
      "grad_norm": 0.7910065087434878,
      "learning_rate": 5e-06,
      "loss": 0.4619,
      "step": 8770
    },
    {
      "epoch": 1.9000216403375894,
      "grad_norm": 0.7485212969421532,
      "learning_rate": 5e-06,
      "loss": 0.4687,
      "step": 8780
    },
    {
      "epoch": 1.902185674096516,
      "grad_norm": 0.8215859586315127,
      "learning_rate": 5e-06,
      "loss": 0.4555,
      "step": 8790
    },
    {
      "epoch": 1.9043497078554426,
      "grad_norm": 0.7909926394009462,
      "learning_rate": 5e-06,
      "loss": 0.4666,
      "step": 8800
    },
    {
      "epoch": 1.9065137416143692,
      "grad_norm": 0.7403806845841657,
      "learning_rate": 5e-06,
      "loss": 0.4722,
      "step": 8810
    },
    {
      "epoch": 1.9086777753732957,
      "grad_norm": 0.7617013904099731,
      "learning_rate": 5e-06,
      "loss": 0.4553,
      "step": 8820
    },
    {
      "epoch": 1.9108418091322226,
      "grad_norm": 0.7849518527403264,
      "learning_rate": 5e-06,
      "loss": 0.4612,
      "step": 8830
    },
    {
      "epoch": 1.913005842891149,
      "grad_norm": 0.7899431737849835,
      "learning_rate": 5e-06,
      "loss": 0.4795,
      "step": 8840
    },
    {
      "epoch": 1.9151698766500758,
      "grad_norm": 0.7232707414732422,
      "learning_rate": 5e-06,
      "loss": 0.4661,
      "step": 8850
    },
    {
      "epoch": 1.9173339104090024,
      "grad_norm": 0.7715859939569867,
      "learning_rate": 5e-06,
      "loss": 0.4729,
      "step": 8860
    },
    {
      "epoch": 1.919497944167929,
      "grad_norm": 0.778365711933746,
      "learning_rate": 5e-06,
      "loss": 0.466,
      "step": 8870
    },
    {
      "epoch": 1.9216619779268558,
      "grad_norm": 0.7315745037821393,
      "learning_rate": 5e-06,
      "loss": 0.458,
      "step": 8880
    },
    {
      "epoch": 1.9238260116857822,
      "grad_norm": 0.7850569825810849,
      "learning_rate": 5e-06,
      "loss": 0.4657,
      "step": 8890
    },
    {
      "epoch": 1.925990045444709,
      "grad_norm": 0.7935430629873113,
      "learning_rate": 5e-06,
      "loss": 0.4704,
      "step": 8900
    },
    {
      "epoch": 1.9281540792036356,
      "grad_norm": 0.8082828639182642,
      "learning_rate": 5e-06,
      "loss": 0.4694,
      "step": 8910
    },
    {
      "epoch": 1.9303181129625622,
      "grad_norm": 0.7982975795262971,
      "learning_rate": 5e-06,
      "loss": 0.4595,
      "step": 8920
    },
    {
      "epoch": 1.932482146721489,
      "grad_norm": 0.7278908896393382,
      "learning_rate": 5e-06,
      "loss": 0.4705,
      "step": 8930
    },
    {
      "epoch": 1.9346461804804154,
      "grad_norm": 0.7593550485177655,
      "learning_rate": 5e-06,
      "loss": 0.4769,
      "step": 8940
    },
    {
      "epoch": 1.9368102142393422,
      "grad_norm": 0.7560080244005302,
      "learning_rate": 5e-06,
      "loss": 0.4508,
      "step": 8950
    },
    {
      "epoch": 1.9389742479982688,
      "grad_norm": 0.7523700953082769,
      "learning_rate": 5e-06,
      "loss": 0.4583,
      "step": 8960
    },
    {
      "epoch": 1.9411382817571954,
      "grad_norm": 0.7729932814097218,
      "learning_rate": 5e-06,
      "loss": 0.46,
      "step": 8970
    },
    {
      "epoch": 1.9433023155161222,
      "grad_norm": 0.7859581817688281,
      "learning_rate": 5e-06,
      "loss": 0.4657,
      "step": 8980
    },
    {
      "epoch": 1.9454663492750486,
      "grad_norm": 0.7649621801554861,
      "learning_rate": 5e-06,
      "loss": 0.4673,
      "step": 8990
    },
    {
      "epoch": 1.9476303830339754,
      "grad_norm": 0.7697491440260914,
      "learning_rate": 5e-06,
      "loss": 0.4635,
      "step": 9000
    },
    {
      "epoch": 1.949794416792902,
      "grad_norm": 0.783274985348169,
      "learning_rate": 5e-06,
      "loss": 0.4634,
      "step": 9010
    },
    {
      "epoch": 1.9519584505518286,
      "grad_norm": 0.715096846477804,
      "learning_rate": 5e-06,
      "loss": 0.4693,
      "step": 9020
    },
    {
      "epoch": 1.9541224843107552,
      "grad_norm": 0.8477280634175208,
      "learning_rate": 5e-06,
      "loss": 0.4597,
      "step": 9030
    },
    {
      "epoch": 1.9562865180696818,
      "grad_norm": 0.7592259661678222,
      "learning_rate": 5e-06,
      "loss": 0.459,
      "step": 9040
    },
    {
      "epoch": 1.9584505518286086,
      "grad_norm": 0.855762642211244,
      "learning_rate": 5e-06,
      "loss": 0.4552,
      "step": 9050
    },
    {
      "epoch": 1.9606145855875352,
      "grad_norm": 0.7724079364010088,
      "learning_rate": 5e-06,
      "loss": 0.4585,
      "step": 9060
    },
    {
      "epoch": 1.9627786193464618,
      "grad_norm": 0.7472414181252647,
      "learning_rate": 5e-06,
      "loss": 0.4575,
      "step": 9070
    },
    {
      "epoch": 1.9649426531053884,
      "grad_norm": 0.77667179344492,
      "learning_rate": 5e-06,
      "loss": 0.4702,
      "step": 9080
    },
    {
      "epoch": 1.967106686864315,
      "grad_norm": 0.7738412500478078,
      "learning_rate": 5e-06,
      "loss": 0.4548,
      "step": 9090
    },
    {
      "epoch": 1.9692707206232418,
      "grad_norm": 0.6873551934444759,
      "learning_rate": 5e-06,
      "loss": 0.4618,
      "step": 9100
    },
    {
      "epoch": 1.9714347543821682,
      "grad_norm": 0.7616982926840125,
      "learning_rate": 5e-06,
      "loss": 0.4633,
      "step": 9110
    },
    {
      "epoch": 1.973598788141095,
      "grad_norm": 0.7359315671968003,
      "learning_rate": 5e-06,
      "loss": 0.4507,
      "step": 9120
    },
    {
      "epoch": 1.9757628219000216,
      "grad_norm": 0.7862965542736634,
      "learning_rate": 5e-06,
      "loss": 0.4752,
      "step": 9130
    },
    {
      "epoch": 1.9779268556589482,
      "grad_norm": 0.7833594304960272,
      "learning_rate": 5e-06,
      "loss": 0.4665,
      "step": 9140
    },
    {
      "epoch": 1.980090889417875,
      "grad_norm": 0.7749375306947249,
      "learning_rate": 5e-06,
      "loss": 0.4606,
      "step": 9150
    },
    {
      "epoch": 1.9822549231768014,
      "grad_norm": 0.7755483335751571,
      "learning_rate": 5e-06,
      "loss": 0.46,
      "step": 9160
    },
    {
      "epoch": 1.9844189569357282,
      "grad_norm": 0.7318259540031191,
      "learning_rate": 5e-06,
      "loss": 0.4557,
      "step": 9170
    },
    {
      "epoch": 1.9865829906946548,
      "grad_norm": 0.7415991714178316,
      "learning_rate": 5e-06,
      "loss": 0.4699,
      "step": 9180
    },
    {
      "epoch": 1.9887470244535814,
      "grad_norm": 0.8918970620944271,
      "learning_rate": 5e-06,
      "loss": 0.4584,
      "step": 9190
    },
    {
      "epoch": 1.9909110582125082,
      "grad_norm": 0.7854257183159034,
      "learning_rate": 5e-06,
      "loss": 0.4659,
      "step": 9200
    },
    {
      "epoch": 1.9930750919714346,
      "grad_norm": 0.743885959515272,
      "learning_rate": 5e-06,
      "loss": 0.4614,
      "step": 9210
    },
    {
      "epoch": 1.9952391257303614,
      "grad_norm": 0.7564617143226638,
      "learning_rate": 5e-06,
      "loss": 0.4629,
      "step": 9220
    },
    {
      "epoch": 1.997403159489288,
      "grad_norm": 0.7956935106295134,
      "learning_rate": 5e-06,
      "loss": 0.46,
      "step": 9230
    },
    {
      "epoch": 1.9995671932482146,
      "grad_norm": 0.7902840252478065,
      "learning_rate": 5e-06,
      "loss": 0.4605,
      "step": 9240
    },
    {
      "epoch": 2.0,
      "eval_loss": 0.5312886238098145,
      "eval_runtime": 585.8565,
      "eval_samples_per_second": 26.566,
      "eval_steps_per_second": 0.416,
      "step": 9242
    },
    {
      "epoch": 2.0017312270071415,
      "grad_norm": 0.8374428329689654,
      "learning_rate": 5e-06,
      "loss": 0.3932,
      "step": 9250
    },
    {
      "epoch": 2.003895260766068,
      "grad_norm": 0.7293070552754123,
      "learning_rate": 5e-06,
      "loss": 0.3703,
      "step": 9260
    },
    {
      "epoch": 2.0060592945249947,
      "grad_norm": 0.7267288150501345,
      "learning_rate": 5e-06,
      "loss": 0.3791,
      "step": 9270
    },
    {
      "epoch": 2.008223328283921,
      "grad_norm": 0.7328996332443125,
      "learning_rate": 5e-06,
      "loss": 0.3682,
      "step": 9280
    },
    {
      "epoch": 2.010387362042848,
      "grad_norm": 0.769240951148092,
      "learning_rate": 5e-06,
      "loss": 0.3771,
      "step": 9290
    },
    {
      "epoch": 2.0125513958017747,
      "grad_norm": 0.782608875094357,
      "learning_rate": 5e-06,
      "loss": 0.3835,
      "step": 9300
    },
    {
      "epoch": 2.014715429560701,
      "grad_norm": 0.7343404957266344,
      "learning_rate": 5e-06,
      "loss": 0.3672,
      "step": 9310
    },
    {
      "epoch": 2.016879463319628,
      "grad_norm": 0.7361975290672562,
      "learning_rate": 5e-06,
      "loss": 0.3638,
      "step": 9320
    },
    {
      "epoch": 2.0190434970785542,
      "grad_norm": 0.8170669306278748,
      "learning_rate": 5e-06,
      "loss": 0.3833,
      "step": 9330
    },
    {
      "epoch": 2.021207530837481,
      "grad_norm": 0.7599248424010396,
      "learning_rate": 5e-06,
      "loss": 0.3799,
      "step": 9340
    },
    {
      "epoch": 2.023371564596408,
      "grad_norm": 0.7854230438584365,
      "learning_rate": 5e-06,
      "loss": 0.3774,
      "step": 9350
    },
    {
      "epoch": 2.0255355983553343,
      "grad_norm": 0.747309033166393,
      "learning_rate": 5e-06,
      "loss": 0.3837,
      "step": 9360
    },
    {
      "epoch": 2.027699632114261,
      "grad_norm": 0.7554391399636565,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 9370
    },
    {
      "epoch": 2.0298636658731875,
      "grad_norm": 0.7369701594292731,
      "learning_rate": 5e-06,
      "loss": 0.373,
      "step": 9380
    },
    {
      "epoch": 2.0320276996321143,
      "grad_norm": 0.7920170859213356,
      "learning_rate": 5e-06,
      "loss": 0.3829,
      "step": 9390
    },
    {
      "epoch": 2.034191733391041,
      "grad_norm": 0.7385616800337693,
      "learning_rate": 5e-06,
      "loss": 0.3732,
      "step": 9400
    },
    {
      "epoch": 2.0363557671499675,
      "grad_norm": 0.7370778661512036,
      "learning_rate": 5e-06,
      "loss": 0.3634,
      "step": 9410
    },
    {
      "epoch": 2.0385198009088943,
      "grad_norm": 0.7064048496785564,
      "learning_rate": 5e-06,
      "loss": 0.3725,
      "step": 9420
    },
    {
      "epoch": 2.0406838346678207,
      "grad_norm": 0.7588040574820386,
      "learning_rate": 5e-06,
      "loss": 0.3803,
      "step": 9430
    },
    {
      "epoch": 2.0428478684267475,
      "grad_norm": 0.6837725115579112,
      "learning_rate": 5e-06,
      "loss": 0.3711,
      "step": 9440
    },
    {
      "epoch": 2.0450119021856743,
      "grad_norm": 0.7072237203949913,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 9450
    },
    {
      "epoch": 2.0471759359446007,
      "grad_norm": 0.7715770726086181,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 9460
    },
    {
      "epoch": 2.0493399697035275,
      "grad_norm": 0.7603839434200408,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 9470
    },
    {
      "epoch": 2.051504003462454,
      "grad_norm": 0.7811463267122977,
      "learning_rate": 5e-06,
      "loss": 0.3755,
      "step": 9480
    },
    {
      "epoch": 2.0536680372213807,
      "grad_norm": 0.7223273231471339,
      "learning_rate": 5e-06,
      "loss": 0.3775,
      "step": 9490
    },
    {
      "epoch": 2.0558320709803075,
      "grad_norm": 0.7489458096255904,
      "learning_rate": 5e-06,
      "loss": 0.3752,
      "step": 9500
    },
    {
      "epoch": 2.057996104739234,
      "grad_norm": 0.7169025560301021,
      "learning_rate": 5e-06,
      "loss": 0.3838,
      "step": 9510
    },
    {
      "epoch": 2.0601601384981607,
      "grad_norm": 0.8241052771764841,
      "learning_rate": 5e-06,
      "loss": 0.3763,
      "step": 9520
    },
    {
      "epoch": 2.062324172257087,
      "grad_norm": 0.7344203339500558,
      "learning_rate": 5e-06,
      "loss": 0.3759,
      "step": 9530
    },
    {
      "epoch": 2.064488206016014,
      "grad_norm": 0.7469451548032326,
      "learning_rate": 5e-06,
      "loss": 0.3768,
      "step": 9540
    },
    {
      "epoch": 2.0666522397749403,
      "grad_norm": 0.7084397676411069,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 9550
    },
    {
      "epoch": 2.068816273533867,
      "grad_norm": 0.7938766312720807,
      "learning_rate": 5e-06,
      "loss": 0.383,
      "step": 9560
    },
    {
      "epoch": 2.070980307292794,
      "grad_norm": 0.766097768652044,
      "learning_rate": 5e-06,
      "loss": 0.376,
      "step": 9570
    },
    {
      "epoch": 2.0731443410517203,
      "grad_norm": 0.7626098304896327,
      "learning_rate": 5e-06,
      "loss": 0.3735,
      "step": 9580
    },
    {
      "epoch": 2.075308374810647,
      "grad_norm": 0.7506050168232735,
      "learning_rate": 5e-06,
      "loss": 0.3763,
      "step": 9590
    },
    {
      "epoch": 2.0774724085695735,
      "grad_norm": 0.7505245897537008,
      "learning_rate": 5e-06,
      "loss": 0.3781,
      "step": 9600
    },
    {
      "epoch": 2.0796364423285003,
      "grad_norm": 0.7826239773634133,
      "learning_rate": 5e-06,
      "loss": 0.3756,
      "step": 9610
    },
    {
      "epoch": 2.081800476087427,
      "grad_norm": 0.7278522309360076,
      "learning_rate": 5e-06,
      "loss": 0.3752,
      "step": 9620
    },
    {
      "epoch": 2.0839645098463535,
      "grad_norm": 0.7756181784692793,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 9630
    },
    {
      "epoch": 2.0861285436052803,
      "grad_norm": 0.7359055323127612,
      "learning_rate": 5e-06,
      "loss": 0.3722,
      "step": 9640
    },
    {
      "epoch": 2.0882925773642067,
      "grad_norm": 0.7577112116476528,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 9650
    },
    {
      "epoch": 2.0904566111231335,
      "grad_norm": 0.8136878823112847,
      "learning_rate": 5e-06,
      "loss": 0.3897,
      "step": 9660
    },
    {
      "epoch": 2.0926206448820603,
      "grad_norm": 0.7427782161695468,
      "learning_rate": 5e-06,
      "loss": 0.3782,
      "step": 9670
    },
    {
      "epoch": 2.0947846786409867,
      "grad_norm": 0.8424273390908451,
      "learning_rate": 5e-06,
      "loss": 0.3879,
      "step": 9680
    },
    {
      "epoch": 2.0969487123999135,
      "grad_norm": 0.7658654614761568,
      "learning_rate": 5e-06,
      "loss": 0.3809,
      "step": 9690
    },
    {
      "epoch": 2.09911274615884,
      "grad_norm": 0.7636850827656342,
      "learning_rate": 5e-06,
      "loss": 0.3802,
      "step": 9700
    },
    {
      "epoch": 2.1012767799177667,
      "grad_norm": 0.7406666012902577,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 9710
    },
    {
      "epoch": 2.1034408136766936,
      "grad_norm": 0.731920134607451,
      "learning_rate": 5e-06,
      "loss": 0.3668,
      "step": 9720
    },
    {
      "epoch": 2.10560484743562,
      "grad_norm": 0.7932449176977977,
      "learning_rate": 5e-06,
      "loss": 0.3793,
      "step": 9730
    },
    {
      "epoch": 2.1077688811945468,
      "grad_norm": 0.7397705079775352,
      "learning_rate": 5e-06,
      "loss": 0.379,
      "step": 9740
    },
    {
      "epoch": 2.109932914953473,
      "grad_norm": 0.8144870875402731,
      "learning_rate": 5e-06,
      "loss": 0.3831,
      "step": 9750
    },
    {
      "epoch": 2.1120969487124,
      "grad_norm": 0.7451950427223708,
      "learning_rate": 5e-06,
      "loss": 0.3762,
      "step": 9760
    },
    {
      "epoch": 2.1142609824713268,
      "grad_norm": 0.7355015739740769,
      "learning_rate": 5e-06,
      "loss": 0.3778,
      "step": 9770
    },
    {
      "epoch": 2.116425016230253,
      "grad_norm": 0.7536357069669827,
      "learning_rate": 5e-06,
      "loss": 0.3758,
      "step": 9780
    },
    {
      "epoch": 2.11858904998918,
      "grad_norm": 0.7708756762306975,
      "learning_rate": 5e-06,
      "loss": 0.3738,
      "step": 9790
    },
    {
      "epoch": 2.1207530837481063,
      "grad_norm": 0.7626855696919476,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 9800
    },
    {
      "epoch": 2.122917117507033,
      "grad_norm": 0.7158584629999891,
      "learning_rate": 5e-06,
      "loss": 0.3753,
      "step": 9810
    },
    {
      "epoch": 2.1250811512659595,
      "grad_norm": 0.7466487488136114,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 9820
    },
    {
      "epoch": 2.1272451850248864,
      "grad_norm": 0.7404349219412076,
      "learning_rate": 5e-06,
      "loss": 0.3721,
      "step": 9830
    },
    {
      "epoch": 2.129409218783813,
      "grad_norm": 0.7668568690340375,
      "learning_rate": 5e-06,
      "loss": 0.3768,
      "step": 9840
    },
    {
      "epoch": 2.1315732525427395,
      "grad_norm": 0.7268904811888163,
      "learning_rate": 5e-06,
      "loss": 0.3712,
      "step": 9850
    },
    {
      "epoch": 2.1337372863016664,
      "grad_norm": 0.7719283305769358,
      "learning_rate": 5e-06,
      "loss": 0.3902,
      "step": 9860
    },
    {
      "epoch": 2.1359013200605927,
      "grad_norm": 0.7079289824393394,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 9870
    },
    {
      "epoch": 2.1380653538195196,
      "grad_norm": 0.7571867101155465,
      "learning_rate": 5e-06,
      "loss": 0.3836,
      "step": 9880
    },
    {
      "epoch": 2.1402293875784464,
      "grad_norm": 0.7549119684648362,
      "learning_rate": 5e-06,
      "loss": 0.3851,
      "step": 9890
    },
    {
      "epoch": 2.1423934213373728,
      "grad_norm": 0.7941986493329641,
      "learning_rate": 5e-06,
      "loss": 0.3776,
      "step": 9900
    },
    {
      "epoch": 2.1445574550962996,
      "grad_norm": 0.7402941631803586,
      "learning_rate": 5e-06,
      "loss": 0.376,
      "step": 9910
    },
    {
      "epoch": 2.146721488855226,
      "grad_norm": 0.7853141015981866,
      "learning_rate": 5e-06,
      "loss": 0.3757,
      "step": 9920
    },
    {
      "epoch": 2.1488855226141528,
      "grad_norm": 0.7502860009884765,
      "learning_rate": 5e-06,
      "loss": 0.3818,
      "step": 9930
    },
    {
      "epoch": 2.1510495563730796,
      "grad_norm": 0.760008223381561,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 9940
    },
    {
      "epoch": 2.153213590132006,
      "grad_norm": 0.7940769908926166,
      "learning_rate": 5e-06,
      "loss": 0.3804,
      "step": 9950
    },
    {
      "epoch": 2.155377623890933,
      "grad_norm": 0.739081181426962,
      "learning_rate": 5e-06,
      "loss": 0.3773,
      "step": 9960
    },
    {
      "epoch": 2.157541657649859,
      "grad_norm": 0.7503458217565563,
      "learning_rate": 5e-06,
      "loss": 0.3712,
      "step": 9970
    },
    {
      "epoch": 2.159705691408786,
      "grad_norm": 0.7598931433539674,
      "learning_rate": 5e-06,
      "loss": 0.3844,
      "step": 9980
    },
    {
      "epoch": 2.161869725167713,
      "grad_norm": 0.7494704002208226,
      "learning_rate": 5e-06,
      "loss": 0.3895,
      "step": 9990
    },
    {
      "epoch": 2.164033758926639,
      "grad_norm": 0.7853450199962944,
      "learning_rate": 5e-06,
      "loss": 0.382,
      "step": 10000
    },
    {
      "epoch": 2.166197792685566,
      "grad_norm": 0.7578503693676106,
      "learning_rate": 5e-06,
      "loss": 0.383,
      "step": 10010
    },
    {
      "epoch": 2.1683618264444924,
      "grad_norm": 0.7484549962239924,
      "learning_rate": 5e-06,
      "loss": 0.3855,
      "step": 10020
    },
    {
      "epoch": 2.170525860203419,
      "grad_norm": 0.7614013434918991,
      "learning_rate": 5e-06,
      "loss": 0.3696,
      "step": 10030
    },
    {
      "epoch": 2.172689893962346,
      "grad_norm": 0.7392030074985807,
      "learning_rate": 5e-06,
      "loss": 0.3729,
      "step": 10040
    },
    {
      "epoch": 2.1748539277212724,
      "grad_norm": 0.7415769297940891,
      "learning_rate": 5e-06,
      "loss": 0.3767,
      "step": 10050
    },
    {
      "epoch": 2.177017961480199,
      "grad_norm": 0.744537273333133,
      "learning_rate": 5e-06,
      "loss": 0.376,
      "step": 10060
    },
    {
      "epoch": 2.1791819952391256,
      "grad_norm": 0.7295839586667225,
      "learning_rate": 5e-06,
      "loss": 0.3824,
      "step": 10070
    },
    {
      "epoch": 2.1813460289980524,
      "grad_norm": 0.749316196787019,
      "learning_rate": 5e-06,
      "loss": 0.385,
      "step": 10080
    },
    {
      "epoch": 2.1835100627569792,
      "grad_norm": 0.8144715088789032,
      "learning_rate": 5e-06,
      "loss": 0.3869,
      "step": 10090
    },
    {
      "epoch": 2.1856740965159056,
      "grad_norm": 0.7376466691462915,
      "learning_rate": 5e-06,
      "loss": 0.382,
      "step": 10100
    },
    {
      "epoch": 2.1878381302748324,
      "grad_norm": 0.7553394493000064,
      "learning_rate": 5e-06,
      "loss": 0.3831,
      "step": 10110
    },
    {
      "epoch": 2.190002164033759,
      "grad_norm": 0.7571271156204371,
      "learning_rate": 5e-06,
      "loss": 0.3851,
      "step": 10120
    },
    {
      "epoch": 2.1921661977926856,
      "grad_norm": 0.7574287608767191,
      "learning_rate": 5e-06,
      "loss": 0.3783,
      "step": 10130
    },
    {
      "epoch": 2.194330231551612,
      "grad_norm": 0.7883853379655534,
      "learning_rate": 5e-06,
      "loss": 0.3862,
      "step": 10140
    },
    {
      "epoch": 2.196494265310539,
      "grad_norm": 0.7491644336765493,
      "learning_rate": 5e-06,
      "loss": 0.3824,
      "step": 10150
    },
    {
      "epoch": 2.1986582990694656,
      "grad_norm": 0.8176175667235511,
      "learning_rate": 5e-06,
      "loss": 0.382,
      "step": 10160
    },
    {
      "epoch": 2.200822332828392,
      "grad_norm": 0.7397237145396862,
      "learning_rate": 5e-06,
      "loss": 0.3821,
      "step": 10170
    },
    {
      "epoch": 2.202986366587319,
      "grad_norm": 0.6965521397762487,
      "learning_rate": 5e-06,
      "loss": 0.3696,
      "step": 10180
    },
    {
      "epoch": 2.205150400346245,
      "grad_norm": 0.7532643826436776,
      "learning_rate": 5e-06,
      "loss": 0.3794,
      "step": 10190
    },
    {
      "epoch": 2.207314434105172,
      "grad_norm": 0.7654434113655321,
      "learning_rate": 5e-06,
      "loss": 0.3777,
      "step": 10200
    },
    {
      "epoch": 2.209478467864099,
      "grad_norm": 0.7712364055070057,
      "learning_rate": 5e-06,
      "loss": 0.3776,
      "step": 10210
    },
    {
      "epoch": 2.2116425016230252,
      "grad_norm": 0.731992016899879,
      "learning_rate": 5e-06,
      "loss": 0.3867,
      "step": 10220
    },
    {
      "epoch": 2.213806535381952,
      "grad_norm": 0.7463982586618078,
      "learning_rate": 5e-06,
      "loss": 0.3816,
      "step": 10230
    },
    {
      "epoch": 2.2159705691408784,
      "grad_norm": 0.7624502219381054,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 10240
    },
    {
      "epoch": 2.2181346028998052,
      "grad_norm": 0.7686079427165029,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 10250
    },
    {
      "epoch": 2.220298636658732,
      "grad_norm": 0.7572457495264667,
      "learning_rate": 5e-06,
      "loss": 0.3823,
      "step": 10260
    },
    {
      "epoch": 2.2224626704176584,
      "grad_norm": 0.8311579714029802,
      "learning_rate": 5e-06,
      "loss": 0.3894,
      "step": 10270
    },
    {
      "epoch": 2.2246267041765853,
      "grad_norm": 0.753836212037088,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 10280
    },
    {
      "epoch": 2.2267907379355116,
      "grad_norm": 0.7777095091483516,
      "learning_rate": 5e-06,
      "loss": 0.3745,
      "step": 10290
    },
    {
      "epoch": 2.2289547716944385,
      "grad_norm": 0.7345531886014162,
      "learning_rate": 5e-06,
      "loss": 0.3814,
      "step": 10300
    },
    {
      "epoch": 2.2311188054533653,
      "grad_norm": 0.7492651509334102,
      "learning_rate": 5e-06,
      "loss": 0.3895,
      "step": 10310
    },
    {
      "epoch": 2.2332828392122916,
      "grad_norm": 0.7734399829479766,
      "learning_rate": 5e-06,
      "loss": 0.3899,
      "step": 10320
    },
    {
      "epoch": 2.2354468729712185,
      "grad_norm": 0.8170059562572255,
      "learning_rate": 5e-06,
      "loss": 0.3721,
      "step": 10330
    },
    {
      "epoch": 2.237610906730145,
      "grad_norm": 0.7418910648043867,
      "learning_rate": 5e-06,
      "loss": 0.3679,
      "step": 10340
    },
    {
      "epoch": 2.2397749404890717,
      "grad_norm": 0.7707008155582549,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 10350
    },
    {
      "epoch": 2.241938974247998,
      "grad_norm": 0.7508948653468445,
      "learning_rate": 5e-06,
      "loss": 0.3858,
      "step": 10360
    },
    {
      "epoch": 2.244103008006925,
      "grad_norm": 0.7280052352796699,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 10370
    },
    {
      "epoch": 2.2462670417658517,
      "grad_norm": 0.7361553110496917,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 10380
    },
    {
      "epoch": 2.248431075524778,
      "grad_norm": 0.8110758388068879,
      "learning_rate": 5e-06,
      "loss": 0.3859,
      "step": 10390
    },
    {
      "epoch": 2.250595109283705,
      "grad_norm": 0.7585969507359466,
      "learning_rate": 5e-06,
      "loss": 0.37,
      "step": 10400
    },
    {
      "epoch": 2.2527591430426313,
      "grad_norm": 0.7530247958679976,
      "learning_rate": 5e-06,
      "loss": 0.387,
      "step": 10410
    },
    {
      "epoch": 2.254923176801558,
      "grad_norm": 0.7793241812111504,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 10420
    },
    {
      "epoch": 2.257087210560485,
      "grad_norm": 0.7359589759600907,
      "learning_rate": 5e-06,
      "loss": 0.3818,
      "step": 10430
    },
    {
      "epoch": 2.2592512443194113,
      "grad_norm": 0.7074654913334124,
      "learning_rate": 5e-06,
      "loss": 0.3814,
      "step": 10440
    },
    {
      "epoch": 2.261415278078338,
      "grad_norm": 0.8029074288686185,
      "learning_rate": 5e-06,
      "loss": 0.3939,
      "step": 10450
    },
    {
      "epoch": 2.2635793118372645,
      "grad_norm": 0.7773602608387922,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 10460
    },
    {
      "epoch": 2.2657433455961913,
      "grad_norm": 0.7803799314229609,
      "learning_rate": 5e-06,
      "loss": 0.3861,
      "step": 10470
    },
    {
      "epoch": 2.267907379355118,
      "grad_norm": 0.7317655840742071,
      "learning_rate": 5e-06,
      "loss": 0.3846,
      "step": 10480
    },
    {
      "epoch": 2.2700714131140445,
      "grad_norm": 0.7535304850406851,
      "learning_rate": 5e-06,
      "loss": 0.3842,
      "step": 10490
    },
    {
      "epoch": 2.2722354468729713,
      "grad_norm": 0.818389549198698,
      "learning_rate": 5e-06,
      "loss": 0.3864,
      "step": 10500
    },
    {
      "epoch": 2.2743994806318977,
      "grad_norm": 0.7972883371387612,
      "learning_rate": 5e-06,
      "loss": 0.3764,
      "step": 10510
    },
    {
      "epoch": 2.2765635143908245,
      "grad_norm": 0.7684174991828006,
      "learning_rate": 5e-06,
      "loss": 0.3923,
      "step": 10520
    },
    {
      "epoch": 2.2787275481497513,
      "grad_norm": 0.7272379216618123,
      "learning_rate": 5e-06,
      "loss": 0.3703,
      "step": 10530
    },
    {
      "epoch": 2.2808915819086777,
      "grad_norm": 0.7527601200579448,
      "learning_rate": 5e-06,
      "loss": 0.3822,
      "step": 10540
    },
    {
      "epoch": 2.2830556156676045,
      "grad_norm": 0.7661450606626169,
      "learning_rate": 5e-06,
      "loss": 0.3836,
      "step": 10550
    },
    {
      "epoch": 2.285219649426531,
      "grad_norm": 0.779962904036604,
      "learning_rate": 5e-06,
      "loss": 0.3833,
      "step": 10560
    },
    {
      "epoch": 2.2873836831854577,
      "grad_norm": 0.7826624741492424,
      "learning_rate": 5e-06,
      "loss": 0.3858,
      "step": 10570
    },
    {
      "epoch": 2.2895477169443845,
      "grad_norm": 0.745367739386281,
      "learning_rate": 5e-06,
      "loss": 0.3814,
      "step": 10580
    },
    {
      "epoch": 2.291711750703311,
      "grad_norm": 0.730641892603881,
      "learning_rate": 5e-06,
      "loss": 0.3772,
      "step": 10590
    },
    {
      "epoch": 2.2938757844622377,
      "grad_norm": 0.7669396242881692,
      "learning_rate": 5e-06,
      "loss": 0.3839,
      "step": 10600
    },
    {
      "epoch": 2.296039818221164,
      "grad_norm": 0.7319224793844378,
      "learning_rate": 5e-06,
      "loss": 0.3711,
      "step": 10610
    },
    {
      "epoch": 2.298203851980091,
      "grad_norm": 0.7954183700469217,
      "learning_rate": 5e-06,
      "loss": 0.3861,
      "step": 10620
    },
    {
      "epoch": 2.3003678857390177,
      "grad_norm": 0.7564733111942853,
      "learning_rate": 5e-06,
      "loss": 0.3716,
      "step": 10630
    },
    {
      "epoch": 2.302531919497944,
      "grad_norm": 0.7563469175859392,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 10640
    },
    {
      "epoch": 2.304695953256871,
      "grad_norm": 0.740370173672384,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 10650
    },
    {
      "epoch": 2.3068599870157973,
      "grad_norm": 0.8008691481523956,
      "learning_rate": 5e-06,
      "loss": 0.3869,
      "step": 10660
    },
    {
      "epoch": 2.309024020774724,
      "grad_norm": 0.757170756724089,
      "learning_rate": 5e-06,
      "loss": 0.3883,
      "step": 10670
    },
    {
      "epoch": 2.311188054533651,
      "grad_norm": 0.7445447389513807,
      "learning_rate": 5e-06,
      "loss": 0.3739,
      "step": 10680
    },
    {
      "epoch": 2.3133520882925773,
      "grad_norm": 0.7614800896552029,
      "learning_rate": 5e-06,
      "loss": 0.3835,
      "step": 10690
    },
    {
      "epoch": 2.315516122051504,
      "grad_norm": 0.7507557413618469,
      "learning_rate": 5e-06,
      "loss": 0.3776,
      "step": 10700
    },
    {
      "epoch": 2.3176801558104305,
      "grad_norm": 0.7721474736234565,
      "learning_rate": 5e-06,
      "loss": 0.3773,
      "step": 10710
    },
    {
      "epoch": 2.3198441895693573,
      "grad_norm": 0.7313755931401409,
      "learning_rate": 5e-06,
      "loss": 0.3895,
      "step": 10720
    },
    {
      "epoch": 2.322008223328284,
      "grad_norm": 0.7814799741219387,
      "learning_rate": 5e-06,
      "loss": 0.3889,
      "step": 10730
    },
    {
      "epoch": 2.3241722570872105,
      "grad_norm": 0.7774874848897156,
      "learning_rate": 5e-06,
      "loss": 0.3791,
      "step": 10740
    },
    {
      "epoch": 2.3263362908461374,
      "grad_norm": 0.7708201057745674,
      "learning_rate": 5e-06,
      "loss": 0.3888,
      "step": 10750
    },
    {
      "epoch": 2.3285003246050637,
      "grad_norm": 0.7317774627158627,
      "learning_rate": 5e-06,
      "loss": 0.3715,
      "step": 10760
    },
    {
      "epoch": 2.3306643583639906,
      "grad_norm": 0.7271265535921656,
      "learning_rate": 5e-06,
      "loss": 0.3776,
      "step": 10770
    },
    {
      "epoch": 2.332828392122917,
      "grad_norm": 0.8285018373906101,
      "learning_rate": 5e-06,
      "loss": 0.3891,
      "step": 10780
    },
    {
      "epoch": 2.3349924258818437,
      "grad_norm": 0.8019035513607989,
      "learning_rate": 5e-06,
      "loss": 0.3841,
      "step": 10790
    },
    {
      "epoch": 2.3371564596407706,
      "grad_norm": 0.7544365537673512,
      "learning_rate": 5e-06,
      "loss": 0.386,
      "step": 10800
    },
    {
      "epoch": 2.339320493399697,
      "grad_norm": 0.774623922213998,
      "learning_rate": 5e-06,
      "loss": 0.3872,
      "step": 10810
    },
    {
      "epoch": 2.3414845271586238,
      "grad_norm": 0.7287036153983298,
      "learning_rate": 5e-06,
      "loss": 0.3851,
      "step": 10820
    },
    {
      "epoch": 2.34364856091755,
      "grad_norm": 0.7831306302240653,
      "learning_rate": 5e-06,
      "loss": 0.3852,
      "step": 10830
    },
    {
      "epoch": 2.345812594676477,
      "grad_norm": 0.7455400541087714,
      "learning_rate": 5e-06,
      "loss": 0.3832,
      "step": 10840
    },
    {
      "epoch": 2.347976628435404,
      "grad_norm": 0.7911713615620226,
      "learning_rate": 5e-06,
      "loss": 0.3942,
      "step": 10850
    },
    {
      "epoch": 2.35014066219433,
      "grad_norm": 0.8028233557037938,
      "learning_rate": 5e-06,
      "loss": 0.388,
      "step": 10860
    },
    {
      "epoch": 2.352304695953257,
      "grad_norm": 0.7864179290759362,
      "learning_rate": 5e-06,
      "loss": 0.3824,
      "step": 10870
    },
    {
      "epoch": 2.3544687297121834,
      "grad_norm": 0.7554701896856139,
      "learning_rate": 5e-06,
      "loss": 0.393,
      "step": 10880
    },
    {
      "epoch": 2.35663276347111,
      "grad_norm": 0.723649579527871,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 10890
    },
    {
      "epoch": 2.3587967972300365,
      "grad_norm": 0.7542043146491088,
      "learning_rate": 5e-06,
      "loss": 0.3772,
      "step": 10900
    },
    {
      "epoch": 2.3609608309889634,
      "grad_norm": 0.7831233351849556,
      "learning_rate": 5e-06,
      "loss": 0.3796,
      "step": 10910
    },
    {
      "epoch": 2.36312486474789,
      "grad_norm": 0.7789860410163189,
      "learning_rate": 5e-06,
      "loss": 0.3829,
      "step": 10920
    },
    {
      "epoch": 2.3652888985068166,
      "grad_norm": 0.7371403610126779,
      "learning_rate": 5e-06,
      "loss": 0.3893,
      "step": 10930
    },
    {
      "epoch": 2.3674529322657434,
      "grad_norm": 0.7725505898139313,
      "learning_rate": 5e-06,
      "loss": 0.3884,
      "step": 10940
    },
    {
      "epoch": 2.3696169660246698,
      "grad_norm": 0.7347104158979803,
      "learning_rate": 5e-06,
      "loss": 0.3905,
      "step": 10950
    },
    {
      "epoch": 2.3717809997835966,
      "grad_norm": 0.7657492814282947,
      "learning_rate": 5e-06,
      "loss": 0.3879,
      "step": 10960
    },
    {
      "epoch": 2.3739450335425234,
      "grad_norm": 0.7535828749091847,
      "learning_rate": 5e-06,
      "loss": 0.3875,
      "step": 10970
    },
    {
      "epoch": 2.3761090673014498,
      "grad_norm": 0.7577252254306687,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 10980
    },
    {
      "epoch": 2.3782731010603766,
      "grad_norm": 0.754449061571296,
      "learning_rate": 5e-06,
      "loss": 0.384,
      "step": 10990
    },
    {
      "epoch": 2.380437134819303,
      "grad_norm": 0.7286408029795997,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 11000
    },
    {
      "epoch": 2.38260116857823,
      "grad_norm": 0.7630893647158181,
      "learning_rate": 5e-06,
      "loss": 0.381,
      "step": 11010
    },
    {
      "epoch": 2.3847652023371566,
      "grad_norm": 0.7962306496261992,
      "learning_rate": 5e-06,
      "loss": 0.3876,
      "step": 11020
    },
    {
      "epoch": 2.386929236096083,
      "grad_norm": 0.7677708510965319,
      "learning_rate": 5e-06,
      "loss": 0.3845,
      "step": 11030
    },
    {
      "epoch": 2.38909326985501,
      "grad_norm": 0.7570281432515957,
      "learning_rate": 5e-06,
      "loss": 0.3867,
      "step": 11040
    },
    {
      "epoch": 2.391257303613936,
      "grad_norm": 0.8132940105852546,
      "learning_rate": 5e-06,
      "loss": 0.3782,
      "step": 11050
    },
    {
      "epoch": 2.393421337372863,
      "grad_norm": 0.739800914008478,
      "learning_rate": 5e-06,
      "loss": 0.3823,
      "step": 11060
    },
    {
      "epoch": 2.39558537113179,
      "grad_norm": 0.7714600709258795,
      "learning_rate": 5e-06,
      "loss": 0.3934,
      "step": 11070
    },
    {
      "epoch": 2.397749404890716,
      "grad_norm": 0.8296198735762226,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 11080
    },
    {
      "epoch": 2.399913438649643,
      "grad_norm": 0.7783027532344778,
      "learning_rate": 5e-06,
      "loss": 0.3898,
      "step": 11090
    },
    {
      "epoch": 2.4020774724085694,
      "grad_norm": 0.7605774127832313,
      "learning_rate": 5e-06,
      "loss": 0.3786,
      "step": 11100
    },
    {
      "epoch": 2.404241506167496,
      "grad_norm": 0.7161203998165129,
      "learning_rate": 5e-06,
      "loss": 0.3729,
      "step": 11110
    },
    {
      "epoch": 2.406405539926423,
      "grad_norm": 0.8001326738731909,
      "learning_rate": 5e-06,
      "loss": 0.3895,
      "step": 11120
    },
    {
      "epoch": 2.4085695736853494,
      "grad_norm": 0.7852408542137601,
      "learning_rate": 5e-06,
      "loss": 0.3921,
      "step": 11130
    },
    {
      "epoch": 2.4107336074442762,
      "grad_norm": 0.7641378077706924,
      "learning_rate": 5e-06,
      "loss": 0.3881,
      "step": 11140
    },
    {
      "epoch": 2.4128976412032026,
      "grad_norm": 0.7369344114660852,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 11150
    },
    {
      "epoch": 2.4150616749621294,
      "grad_norm": 0.7728142277616102,
      "learning_rate": 5e-06,
      "loss": 0.388,
      "step": 11160
    },
    {
      "epoch": 2.4172257087210562,
      "grad_norm": 0.7938596540542148,
      "learning_rate": 5e-06,
      "loss": 0.3754,
      "step": 11170
    },
    {
      "epoch": 2.4193897424799826,
      "grad_norm": 0.7277868215488438,
      "learning_rate": 5e-06,
      "loss": 0.3803,
      "step": 11180
    },
    {
      "epoch": 2.4215537762389094,
      "grad_norm": 0.7682514933014444,
      "learning_rate": 5e-06,
      "loss": 0.3789,
      "step": 11190
    },
    {
      "epoch": 2.423717809997836,
      "grad_norm": 0.769584672945386,
      "learning_rate": 5e-06,
      "loss": 0.3844,
      "step": 11200
    },
    {
      "epoch": 2.4258818437567626,
      "grad_norm": 0.7295995976083359,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 11210
    },
    {
      "epoch": 2.4280458775156895,
      "grad_norm": 0.7547865237203139,
      "learning_rate": 5e-06,
      "loss": 0.3828,
      "step": 11220
    },
    {
      "epoch": 2.430209911274616,
      "grad_norm": 0.8150532849289923,
      "learning_rate": 5e-06,
      "loss": 0.3858,
      "step": 11230
    },
    {
      "epoch": 2.4323739450335427,
      "grad_norm": 0.8095778619727545,
      "learning_rate": 5e-06,
      "loss": 0.3924,
      "step": 11240
    },
    {
      "epoch": 2.434537978792469,
      "grad_norm": 0.7612807753325169,
      "learning_rate": 5e-06,
      "loss": 0.3825,
      "step": 11250
    },
    {
      "epoch": 2.436702012551396,
      "grad_norm": 0.84684876401696,
      "learning_rate": 5e-06,
      "loss": 0.3912,
      "step": 11260
    },
    {
      "epoch": 2.4388660463103227,
      "grad_norm": 0.7505318974427521,
      "learning_rate": 5e-06,
      "loss": 0.3821,
      "step": 11270
    },
    {
      "epoch": 2.441030080069249,
      "grad_norm": 0.7463667758343082,
      "learning_rate": 5e-06,
      "loss": 0.3779,
      "step": 11280
    },
    {
      "epoch": 2.443194113828176,
      "grad_norm": 0.7930997256650184,
      "learning_rate": 5e-06,
      "loss": 0.3824,
      "step": 11290
    },
    {
      "epoch": 2.4453581475871022,
      "grad_norm": 0.7232751240140999,
      "learning_rate": 5e-06,
      "loss": 0.3712,
      "step": 11300
    },
    {
      "epoch": 2.447522181346029,
      "grad_norm": 0.7452902861534266,
      "learning_rate": 5e-06,
      "loss": 0.3829,
      "step": 11310
    },
    {
      "epoch": 2.449686215104956,
      "grad_norm": 0.7894533038001047,
      "learning_rate": 5e-06,
      "loss": 0.3901,
      "step": 11320
    },
    {
      "epoch": 2.4518502488638823,
      "grad_norm": 0.7822422085940205,
      "learning_rate": 5e-06,
      "loss": 0.3894,
      "step": 11330
    },
    {
      "epoch": 2.454014282622809,
      "grad_norm": 0.7790104348646221,
      "learning_rate": 5e-06,
      "loss": 0.3772,
      "step": 11340
    },
    {
      "epoch": 2.4561783163817354,
      "grad_norm": 0.7674597626994477,
      "learning_rate": 5e-06,
      "loss": 0.3952,
      "step": 11350
    },
    {
      "epoch": 2.4583423501406623,
      "grad_norm": 0.7389051490728133,
      "learning_rate": 5e-06,
      "loss": 0.3886,
      "step": 11360
    },
    {
      "epoch": 2.4605063838995886,
      "grad_norm": 0.7648071653406447,
      "learning_rate": 5e-06,
      "loss": 0.3785,
      "step": 11370
    },
    {
      "epoch": 2.4626704176585155,
      "grad_norm": 0.7485263164164858,
      "learning_rate": 5e-06,
      "loss": 0.3834,
      "step": 11380
    },
    {
      "epoch": 2.4648344514174423,
      "grad_norm": 0.7571813819723854,
      "learning_rate": 5e-06,
      "loss": 0.3891,
      "step": 11390
    },
    {
      "epoch": 2.4669984851763687,
      "grad_norm": 0.79013779876267,
      "learning_rate": 5e-06,
      "loss": 0.3787,
      "step": 11400
    },
    {
      "epoch": 2.4691625189352955,
      "grad_norm": 0.7899502403657093,
      "learning_rate": 5e-06,
      "loss": 0.3792,
      "step": 11410
    },
    {
      "epoch": 2.471326552694222,
      "grad_norm": 0.79065022776973,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 11420
    },
    {
      "epoch": 2.4734905864531487,
      "grad_norm": 0.7579456308475134,
      "learning_rate": 5e-06,
      "loss": 0.3906,
      "step": 11430
    },
    {
      "epoch": 2.4756546202120755,
      "grad_norm": 0.7684812839256481,
      "learning_rate": 5e-06,
      "loss": 0.3846,
      "step": 11440
    },
    {
      "epoch": 2.477818653971002,
      "grad_norm": 0.7541428841884413,
      "learning_rate": 5e-06,
      "loss": 0.3949,
      "step": 11450
    },
    {
      "epoch": 2.4799826877299287,
      "grad_norm": 0.773906564919006,
      "learning_rate": 5e-06,
      "loss": 0.3906,
      "step": 11460
    },
    {
      "epoch": 2.482146721488855,
      "grad_norm": 0.8020255649891496,
      "learning_rate": 5e-06,
      "loss": 0.3925,
      "step": 11470
    },
    {
      "epoch": 2.484310755247782,
      "grad_norm": 0.7615441009856345,
      "learning_rate": 5e-06,
      "loss": 0.4048,
      "step": 11480
    },
    {
      "epoch": 2.4864747890067083,
      "grad_norm": 0.7653863910560466,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 11490
    },
    {
      "epoch": 2.488638822765635,
      "grad_norm": 0.7982102560926431,
      "learning_rate": 5e-06,
      "loss": 0.3791,
      "step": 11500
    },
    {
      "epoch": 2.490802856524562,
      "grad_norm": 0.7646849476306503,
      "learning_rate": 5e-06,
      "loss": 0.3915,
      "step": 11510
    },
    {
      "epoch": 2.4929668902834883,
      "grad_norm": 0.7926740161568403,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 11520
    },
    {
      "epoch": 2.495130924042415,
      "grad_norm": 0.7857256749815332,
      "learning_rate": 5e-06,
      "loss": 0.3901,
      "step": 11530
    },
    {
      "epoch": 2.4972949578013415,
      "grad_norm": 0.7811301407036569,
      "learning_rate": 5e-06,
      "loss": 0.3822,
      "step": 11540
    },
    {
      "epoch": 2.4994589915602683,
      "grad_norm": 0.7612258678572895,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 11550
    },
    {
      "epoch": 2.501623025319195,
      "grad_norm": 0.7510835321762522,
      "learning_rate": 5e-06,
      "loss": 0.3907,
      "step": 11560
    },
    {
      "epoch": 2.5037870590781215,
      "grad_norm": 0.7672276862765546,
      "learning_rate": 5e-06,
      "loss": 0.3828,
      "step": 11570
    },
    {
      "epoch": 2.5059510928370483,
      "grad_norm": 0.7628691005049446,
      "learning_rate": 5e-06,
      "loss": 0.3837,
      "step": 11580
    },
    {
      "epoch": 2.5081151265959747,
      "grad_norm": 0.7471986630619987,
      "learning_rate": 5e-06,
      "loss": 0.3863,
      "step": 11590
    },
    {
      "epoch": 2.5102791603549015,
      "grad_norm": 0.7706154307172854,
      "learning_rate": 5e-06,
      "loss": 0.3896,
      "step": 11600
    },
    {
      "epoch": 2.5124431941138283,
      "grad_norm": 0.7343855290129636,
      "learning_rate": 5e-06,
      "loss": 0.387,
      "step": 11610
    },
    {
      "epoch": 2.5146072278727547,
      "grad_norm": 0.7503993070592432,
      "learning_rate": 5e-06,
      "loss": 0.3947,
      "step": 11620
    },
    {
      "epoch": 2.5167712616316815,
      "grad_norm": 0.7894715279800384,
      "learning_rate": 5e-06,
      "loss": 0.3896,
      "step": 11630
    },
    {
      "epoch": 2.518935295390608,
      "grad_norm": 0.7565826940704703,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 11640
    },
    {
      "epoch": 2.5210993291495347,
      "grad_norm": 0.7680776687871832,
      "learning_rate": 5e-06,
      "loss": 0.3938,
      "step": 11650
    },
    {
      "epoch": 2.5232633629084615,
      "grad_norm": 0.7896385741117732,
      "learning_rate": 5e-06,
      "loss": 0.379,
      "step": 11660
    },
    {
      "epoch": 2.525427396667388,
      "grad_norm": 0.7482445016179938,
      "learning_rate": 5e-06,
      "loss": 0.387,
      "step": 11670
    },
    {
      "epoch": 2.5275914304263147,
      "grad_norm": 0.7788787667528324,
      "learning_rate": 5e-06,
      "loss": 0.3912,
      "step": 11680
    },
    {
      "epoch": 2.529755464185241,
      "grad_norm": 0.7802367273135542,
      "learning_rate": 5e-06,
      "loss": 0.391,
      "step": 11690
    },
    {
      "epoch": 2.531919497944168,
      "grad_norm": 0.7907791607502596,
      "learning_rate": 5e-06,
      "loss": 0.3889,
      "step": 11700
    },
    {
      "epoch": 2.5340835317030947,
      "grad_norm": 0.735159844361493,
      "learning_rate": 5e-06,
      "loss": 0.3866,
      "step": 11710
    },
    {
      "epoch": 2.536247565462021,
      "grad_norm": 0.8002646416305854,
      "learning_rate": 5e-06,
      "loss": 0.3797,
      "step": 11720
    },
    {
      "epoch": 2.538411599220948,
      "grad_norm": 0.744137757018372,
      "learning_rate": 5e-06,
      "loss": 0.3851,
      "step": 11730
    },
    {
      "epoch": 2.5405756329798743,
      "grad_norm": 0.8089884009703747,
      "learning_rate": 5e-06,
      "loss": 0.3876,
      "step": 11740
    },
    {
      "epoch": 2.542739666738801,
      "grad_norm": 0.7409059443870979,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 11750
    },
    {
      "epoch": 2.544903700497728,
      "grad_norm": 0.7474137046476967,
      "learning_rate": 5e-06,
      "loss": 0.3809,
      "step": 11760
    },
    {
      "epoch": 2.5470677342566543,
      "grad_norm": 0.7328688526405038,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 11770
    },
    {
      "epoch": 2.549231768015581,
      "grad_norm": 0.7309917622922488,
      "learning_rate": 5e-06,
      "loss": 0.3821,
      "step": 11780
    },
    {
      "epoch": 2.5513958017745075,
      "grad_norm": 0.760296512703151,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 11790
    },
    {
      "epoch": 2.5535598355334344,
      "grad_norm": 0.7675760208277899,
      "learning_rate": 5e-06,
      "loss": 0.3784,
      "step": 11800
    },
    {
      "epoch": 2.555723869292361,
      "grad_norm": 0.7239541839213967,
      "learning_rate": 5e-06,
      "loss": 0.3893,
      "step": 11810
    },
    {
      "epoch": 2.5578879030512875,
      "grad_norm": 0.7575635250945256,
      "learning_rate": 5e-06,
      "loss": 0.3866,
      "step": 11820
    },
    {
      "epoch": 2.5600519368102144,
      "grad_norm": 0.8205016916935968,
      "learning_rate": 5e-06,
      "loss": 0.3884,
      "step": 11830
    },
    {
      "epoch": 2.5622159705691407,
      "grad_norm": 0.7630887493655477,
      "learning_rate": 5e-06,
      "loss": 0.3822,
      "step": 11840
    },
    {
      "epoch": 2.5643800043280676,
      "grad_norm": 0.7483344691244171,
      "learning_rate": 5e-06,
      "loss": 0.3861,
      "step": 11850
    },
    {
      "epoch": 2.5665440380869944,
      "grad_norm": 0.7836573802364927,
      "learning_rate": 5e-06,
      "loss": 0.3859,
      "step": 11860
    },
    {
      "epoch": 2.5687080718459208,
      "grad_norm": 0.7758135442923644,
      "learning_rate": 5e-06,
      "loss": 0.389,
      "step": 11870
    },
    {
      "epoch": 2.5708721056048476,
      "grad_norm": 0.8156880128058609,
      "learning_rate": 5e-06,
      "loss": 0.3815,
      "step": 11880
    },
    {
      "epoch": 2.573036139363774,
      "grad_norm": 0.769477713621125,
      "learning_rate": 5e-06,
      "loss": 0.3887,
      "step": 11890
    },
    {
      "epoch": 2.5752001731227008,
      "grad_norm": 0.7539503843193486,
      "learning_rate": 5e-06,
      "loss": 0.3746,
      "step": 11900
    },
    {
      "epoch": 2.5773642068816276,
      "grad_norm": 0.7665614539752887,
      "learning_rate": 5e-06,
      "loss": 0.383,
      "step": 11910
    },
    {
      "epoch": 2.579528240640554,
      "grad_norm": 0.78863002604713,
      "learning_rate": 5e-06,
      "loss": 0.3932,
      "step": 11920
    },
    {
      "epoch": 2.581692274399481,
      "grad_norm": 0.7860997141249391,
      "learning_rate": 5e-06,
      "loss": 0.3822,
      "step": 11930
    },
    {
      "epoch": 2.583856308158407,
      "grad_norm": 0.7450886855994443,
      "learning_rate": 5e-06,
      "loss": 0.3847,
      "step": 11940
    },
    {
      "epoch": 2.586020341917334,
      "grad_norm": 0.7184291709731961,
      "learning_rate": 5e-06,
      "loss": 0.4006,
      "step": 11950
    },
    {
      "epoch": 2.588184375676261,
      "grad_norm": 0.7395095877340896,
      "learning_rate": 5e-06,
      "loss": 0.3805,
      "step": 11960
    },
    {
      "epoch": 2.590348409435187,
      "grad_norm": 0.7326541647113222,
      "learning_rate": 5e-06,
      "loss": 0.3926,
      "step": 11970
    },
    {
      "epoch": 2.5925124431941136,
      "grad_norm": 0.7590091456087876,
      "learning_rate": 5e-06,
      "loss": 0.39,
      "step": 11980
    },
    {
      "epoch": 2.5946764769530404,
      "grad_norm": 0.7514373484529209,
      "learning_rate": 5e-06,
      "loss": 0.3805,
      "step": 11990
    },
    {
      "epoch": 2.596840510711967,
      "grad_norm": 0.7974717819972812,
      "learning_rate": 5e-06,
      "loss": 0.3814,
      "step": 12000
    },
    {
      "epoch": 2.599004544470894,
      "grad_norm": 0.7663442947904777,
      "learning_rate": 5e-06,
      "loss": 0.3919,
      "step": 12010
    },
    {
      "epoch": 2.6011685782298204,
      "grad_norm": 0.7869861761124946,
      "learning_rate": 5e-06,
      "loss": 0.3883,
      "step": 12020
    },
    {
      "epoch": 2.6033326119887468,
      "grad_norm": 0.7503773304052421,
      "learning_rate": 5e-06,
      "loss": 0.3877,
      "step": 12030
    },
    {
      "epoch": 2.6054966457476736,
      "grad_norm": 0.776303816000796,
      "learning_rate": 5e-06,
      "loss": 0.3985,
      "step": 12040
    },
    {
      "epoch": 2.6076606795066004,
      "grad_norm": 0.759982286533669,
      "learning_rate": 5e-06,
      "loss": 0.3899,
      "step": 12050
    },
    {
      "epoch": 2.609824713265527,
      "grad_norm": 0.8131552609328792,
      "learning_rate": 5e-06,
      "loss": 0.3884,
      "step": 12060
    },
    {
      "epoch": 2.6119887470244536,
      "grad_norm": 0.7476915162359377,
      "learning_rate": 5e-06,
      "loss": 0.3857,
      "step": 12070
    },
    {
      "epoch": 2.61415278078338,
      "grad_norm": 0.8030188195536405,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 12080
    },
    {
      "epoch": 2.616316814542307,
      "grad_norm": 0.7550164235383711,
      "learning_rate": 5e-06,
      "loss": 0.3903,
      "step": 12090
    },
    {
      "epoch": 2.6184808483012336,
      "grad_norm": 0.759412823543429,
      "learning_rate": 5e-06,
      "loss": 0.3832,
      "step": 12100
    },
    {
      "epoch": 2.62064488206016,
      "grad_norm": 0.7595851928907493,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 12110
    },
    {
      "epoch": 2.622808915819087,
      "grad_norm": 0.7557311993407309,
      "learning_rate": 5e-06,
      "loss": 0.3927,
      "step": 12120
    },
    {
      "epoch": 2.624972949578013,
      "grad_norm": 0.7935821634398855,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 12130
    },
    {
      "epoch": 2.62713698333694,
      "grad_norm": 0.7407643308725581,
      "learning_rate": 5e-06,
      "loss": 0.3871,
      "step": 12140
    },
    {
      "epoch": 2.629301017095867,
      "grad_norm": 0.729366375580239,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 12150
    },
    {
      "epoch": 2.631465050854793,
      "grad_norm": 0.7561366246431421,
      "learning_rate": 5e-06,
      "loss": 0.3822,
      "step": 12160
    },
    {
      "epoch": 2.63362908461372,
      "grad_norm": 0.7667634155030516,
      "learning_rate": 5e-06,
      "loss": 0.3809,
      "step": 12170
    },
    {
      "epoch": 2.6357931183726464,
      "grad_norm": 0.7885773621343677,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 12180
    },
    {
      "epoch": 2.6379571521315732,
      "grad_norm": 0.7440441002401869,
      "learning_rate": 5e-06,
      "loss": 0.3987,
      "step": 12190
    },
    {
      "epoch": 2.6401211858905,
      "grad_norm": 0.7810250085742799,
      "learning_rate": 5e-06,
      "loss": 0.3782,
      "step": 12200
    },
    {
      "epoch": 2.6422852196494264,
      "grad_norm": 0.8150657736573159,
      "learning_rate": 5e-06,
      "loss": 0.3896,
      "step": 12210
    },
    {
      "epoch": 2.6444492534083532,
      "grad_norm": 0.7279096250776349,
      "learning_rate": 5e-06,
      "loss": 0.3821,
      "step": 12220
    },
    {
      "epoch": 2.6466132871672796,
      "grad_norm": 0.766036376483687,
      "learning_rate": 5e-06,
      "loss": 0.3853,
      "step": 12230
    },
    {
      "epoch": 2.6487773209262064,
      "grad_norm": 0.7629756533767561,
      "learning_rate": 5e-06,
      "loss": 0.3848,
      "step": 12240
    },
    {
      "epoch": 2.6509413546851333,
      "grad_norm": 0.7258995252608367,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 12250
    },
    {
      "epoch": 2.6531053884440596,
      "grad_norm": 0.7487644012773005,
      "learning_rate": 5e-06,
      "loss": 0.389,
      "step": 12260
    },
    {
      "epoch": 2.6552694222029865,
      "grad_norm": 0.7730940322471979,
      "learning_rate": 5e-06,
      "loss": 0.3919,
      "step": 12270
    },
    {
      "epoch": 2.657433455961913,
      "grad_norm": 0.7261832821091123,
      "learning_rate": 5e-06,
      "loss": 0.3772,
      "step": 12280
    },
    {
      "epoch": 2.6595974897208396,
      "grad_norm": 0.7841634979405185,
      "learning_rate": 5e-06,
      "loss": 0.3955,
      "step": 12290
    },
    {
      "epoch": 2.6617615234797665,
      "grad_norm": 0.7443154389593502,
      "learning_rate": 5e-06,
      "loss": 0.3998,
      "step": 12300
    },
    {
      "epoch": 2.663925557238693,
      "grad_norm": 0.792617400623899,
      "learning_rate": 5e-06,
      "loss": 0.3838,
      "step": 12310
    },
    {
      "epoch": 2.6660895909976197,
      "grad_norm": 0.7430565465776605,
      "learning_rate": 5e-06,
      "loss": 0.3926,
      "step": 12320
    },
    {
      "epoch": 2.668253624756546,
      "grad_norm": 0.7601315336074875,
      "learning_rate": 5e-06,
      "loss": 0.3935,
      "step": 12330
    },
    {
      "epoch": 2.670417658515473,
      "grad_norm": 0.7349512479796254,
      "learning_rate": 5e-06,
      "loss": 0.386,
      "step": 12340
    },
    {
      "epoch": 2.6725816922743997,
      "grad_norm": 0.7492793091950313,
      "learning_rate": 5e-06,
      "loss": 0.3939,
      "step": 12350
    },
    {
      "epoch": 2.674745726033326,
      "grad_norm": 0.8020327135634121,
      "learning_rate": 5e-06,
      "loss": 0.3933,
      "step": 12360
    },
    {
      "epoch": 2.676909759792253,
      "grad_norm": 0.7821248575887879,
      "learning_rate": 5e-06,
      "loss": 0.3889,
      "step": 12370
    },
    {
      "epoch": 2.6790737935511793,
      "grad_norm": 0.7764574267508418,
      "learning_rate": 5e-06,
      "loss": 0.3834,
      "step": 12380
    },
    {
      "epoch": 2.681237827310106,
      "grad_norm": 0.7581668979481178,
      "learning_rate": 5e-06,
      "loss": 0.3894,
      "step": 12390
    },
    {
      "epoch": 2.683401861069033,
      "grad_norm": 0.7312153550603386,
      "learning_rate": 5e-06,
      "loss": 0.3871,
      "step": 12400
    },
    {
      "epoch": 2.6855658948279593,
      "grad_norm": 0.8290124241375181,
      "learning_rate": 5e-06,
      "loss": 0.3778,
      "step": 12410
    },
    {
      "epoch": 2.687729928586886,
      "grad_norm": 0.7317732999168243,
      "learning_rate": 5e-06,
      "loss": 0.3863,
      "step": 12420
    },
    {
      "epoch": 2.6898939623458125,
      "grad_norm": 0.7638469327765998,
      "learning_rate": 5e-06,
      "loss": 0.3829,
      "step": 12430
    },
    {
      "epoch": 2.6920579961047393,
      "grad_norm": 0.7949708235476729,
      "learning_rate": 5e-06,
      "loss": 0.3926,
      "step": 12440
    },
    {
      "epoch": 2.694222029863666,
      "grad_norm": 0.7550935663260143,
      "learning_rate": 5e-06,
      "loss": 0.3783,
      "step": 12450
    },
    {
      "epoch": 2.6963860636225925,
      "grad_norm": 0.8313055045861929,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 12460
    },
    {
      "epoch": 2.6985500973815193,
      "grad_norm": 0.8005354523317457,
      "learning_rate": 5e-06,
      "loss": 0.3868,
      "step": 12470
    },
    {
      "epoch": 2.7007141311404457,
      "grad_norm": 0.7670070573325029,
      "learning_rate": 5e-06,
      "loss": 0.3815,
      "step": 12480
    },
    {
      "epoch": 2.7028781648993725,
      "grad_norm": 0.7508083437598002,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 12490
    },
    {
      "epoch": 2.7050421986582993,
      "grad_norm": 0.7787188976232684,
      "learning_rate": 5e-06,
      "loss": 0.3872,
      "step": 12500
    },
    {
      "epoch": 2.7072062324172257,
      "grad_norm": 0.7683661547819274,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 12510
    },
    {
      "epoch": 2.7093702661761525,
      "grad_norm": 0.7391247866655715,
      "learning_rate": 5e-06,
      "loss": 0.3806,
      "step": 12520
    },
    {
      "epoch": 2.711534299935079,
      "grad_norm": 0.7565014644999606,
      "learning_rate": 5e-06,
      "loss": 0.3733,
      "step": 12530
    },
    {
      "epoch": 2.7136983336940057,
      "grad_norm": 0.8085171262046021,
      "learning_rate": 5e-06,
      "loss": 0.3737,
      "step": 12540
    },
    {
      "epoch": 2.7158623674529325,
      "grad_norm": 0.7858515305447751,
      "learning_rate": 5e-06,
      "loss": 0.3875,
      "step": 12550
    },
    {
      "epoch": 2.718026401211859,
      "grad_norm": 0.7365212017152488,
      "learning_rate": 5e-06,
      "loss": 0.3908,
      "step": 12560
    },
    {
      "epoch": 2.7201904349707853,
      "grad_norm": 0.7744330424464411,
      "learning_rate": 5e-06,
      "loss": 0.385,
      "step": 12570
    },
    {
      "epoch": 2.722354468729712,
      "grad_norm": 0.7564513841846556,
      "learning_rate": 5e-06,
      "loss": 0.3867,
      "step": 12580
    },
    {
      "epoch": 2.724518502488639,
      "grad_norm": 0.7750108310504817,
      "learning_rate": 5e-06,
      "loss": 0.3892,
      "step": 12590
    },
    {
      "epoch": 2.7266825362475657,
      "grad_norm": 0.7782061698199593,
      "learning_rate": 5e-06,
      "loss": 0.3952,
      "step": 12600
    },
    {
      "epoch": 2.728846570006492,
      "grad_norm": 0.7822075760319362,
      "learning_rate": 5e-06,
      "loss": 0.3981,
      "step": 12610
    },
    {
      "epoch": 2.7310106037654185,
      "grad_norm": 0.7764669704006216,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 12620
    },
    {
      "epoch": 2.7331746375243453,
      "grad_norm": 0.746571146392982,
      "learning_rate": 5e-06,
      "loss": 0.3857,
      "step": 12630
    },
    {
      "epoch": 2.735338671283272,
      "grad_norm": 0.7727176603555808,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 12640
    },
    {
      "epoch": 2.7375027050421985,
      "grad_norm": 0.7716153482496758,
      "learning_rate": 5e-06,
      "loss": 0.3942,
      "step": 12650
    },
    {
      "epoch": 2.7396667388011253,
      "grad_norm": 0.7506221057021802,
      "learning_rate": 5e-06,
      "loss": 0.3828,
      "step": 12660
    },
    {
      "epoch": 2.7418307725600517,
      "grad_norm": 0.765605207662551,
      "learning_rate": 5e-06,
      "loss": 0.3896,
      "step": 12670
    },
    {
      "epoch": 2.7439948063189785,
      "grad_norm": 0.7936436590710892,
      "learning_rate": 5e-06,
      "loss": 0.384,
      "step": 12680
    },
    {
      "epoch": 2.7461588400779053,
      "grad_norm": 0.7417315221453376,
      "learning_rate": 5e-06,
      "loss": 0.3864,
      "step": 12690
    },
    {
      "epoch": 2.7483228738368317,
      "grad_norm": 0.7638299083549244,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 12700
    },
    {
      "epoch": 2.7504869075957585,
      "grad_norm": 0.7401370791792725,
      "learning_rate": 5e-06,
      "loss": 0.3736,
      "step": 12710
    },
    {
      "epoch": 2.752650941354685,
      "grad_norm": 0.7604282539682512,
      "learning_rate": 5e-06,
      "loss": 0.386,
      "step": 12720
    },
    {
      "epoch": 2.7548149751136117,
      "grad_norm": 0.7593327654847288,
      "learning_rate": 5e-06,
      "loss": 0.3874,
      "step": 12730
    },
    {
      "epoch": 2.7569790088725386,
      "grad_norm": 0.762010718640477,
      "learning_rate": 5e-06,
      "loss": 0.3728,
      "step": 12740
    },
    {
      "epoch": 2.759143042631465,
      "grad_norm": 0.7464411042463334,
      "learning_rate": 5e-06,
      "loss": 0.3853,
      "step": 12750
    },
    {
      "epoch": 2.7613070763903917,
      "grad_norm": 0.7527404376235197,
      "learning_rate": 5e-06,
      "loss": 0.3805,
      "step": 12760
    },
    {
      "epoch": 2.763471110149318,
      "grad_norm": 0.7413553870174003,
      "learning_rate": 5e-06,
      "loss": 0.3827,
      "step": 12770
    },
    {
      "epoch": 2.765635143908245,
      "grad_norm": 0.8112331079474291,
      "learning_rate": 5e-06,
      "loss": 0.3912,
      "step": 12780
    },
    {
      "epoch": 2.7677991776671718,
      "grad_norm": 0.7606524264570311,
      "learning_rate": 5e-06,
      "loss": 0.384,
      "step": 12790
    },
    {
      "epoch": 2.769963211426098,
      "grad_norm": 0.7688787180412004,
      "learning_rate": 5e-06,
      "loss": 0.386,
      "step": 12800
    },
    {
      "epoch": 2.772127245185025,
      "grad_norm": 0.7709908015790655,
      "learning_rate": 5e-06,
      "loss": 0.3911,
      "step": 12810
    },
    {
      "epoch": 2.7742912789439513,
      "grad_norm": 0.8256088144670327,
      "learning_rate": 5e-06,
      "loss": 0.394,
      "step": 12820
    },
    {
      "epoch": 2.776455312702878,
      "grad_norm": 0.76732285875986,
      "learning_rate": 5e-06,
      "loss": 0.3825,
      "step": 12830
    },
    {
      "epoch": 2.778619346461805,
      "grad_norm": 0.7623048216172926,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 12840
    },
    {
      "epoch": 2.7807833802207313,
      "grad_norm": 0.7770120662585185,
      "learning_rate": 5e-06,
      "loss": 0.3825,
      "step": 12850
    },
    {
      "epoch": 2.782947413979658,
      "grad_norm": 0.7659653162936821,
      "learning_rate": 5e-06,
      "loss": 0.3802,
      "step": 12860
    },
    {
      "epoch": 2.7851114477385845,
      "grad_norm": 0.7708274852623006,
      "learning_rate": 5e-06,
      "loss": 0.4,
      "step": 12870
    },
    {
      "epoch": 2.7872754814975114,
      "grad_norm": 0.7743267205110804,
      "learning_rate": 5e-06,
      "loss": 0.3833,
      "step": 12880
    },
    {
      "epoch": 2.789439515256438,
      "grad_norm": 0.7749387611421825,
      "learning_rate": 5e-06,
      "loss": 0.3953,
      "step": 12890
    },
    {
      "epoch": 2.7916035490153646,
      "grad_norm": 0.7586126347403827,
      "learning_rate": 5e-06,
      "loss": 0.3853,
      "step": 12900
    },
    {
      "epoch": 2.7937675827742914,
      "grad_norm": 0.7500835141793593,
      "learning_rate": 5e-06,
      "loss": 0.3929,
      "step": 12910
    },
    {
      "epoch": 2.7959316165332178,
      "grad_norm": 0.74663392656643,
      "learning_rate": 5e-06,
      "loss": 0.3795,
      "step": 12920
    },
    {
      "epoch": 2.7980956502921446,
      "grad_norm": 0.7279989402913795,
      "learning_rate": 5e-06,
      "loss": 0.3934,
      "step": 12930
    },
    {
      "epoch": 2.8002596840510714,
      "grad_norm": 0.782757718266062,
      "learning_rate": 5e-06,
      "loss": 0.3902,
      "step": 12940
    },
    {
      "epoch": 2.8024237178099978,
      "grad_norm": 0.7541794890470066,
      "learning_rate": 5e-06,
      "loss": 0.3905,
      "step": 12950
    },
    {
      "epoch": 2.8045877515689246,
      "grad_norm": 0.7511833194905094,
      "learning_rate": 5e-06,
      "loss": 0.3803,
      "step": 12960
    },
    {
      "epoch": 2.806751785327851,
      "grad_norm": 0.7345651454318538,
      "learning_rate": 5e-06,
      "loss": 0.3838,
      "step": 12970
    },
    {
      "epoch": 2.808915819086778,
      "grad_norm": 0.7430756885822685,
      "learning_rate": 5e-06,
      "loss": 0.375,
      "step": 12980
    },
    {
      "epoch": 2.8110798528457046,
      "grad_norm": 0.7503404811162424,
      "learning_rate": 5e-06,
      "loss": 0.3899,
      "step": 12990
    },
    {
      "epoch": 2.813243886604631,
      "grad_norm": 0.7839538272866499,
      "learning_rate": 5e-06,
      "loss": 0.4,
      "step": 13000
    },
    {
      "epoch": 2.815407920363558,
      "grad_norm": 0.8075664166536066,
      "learning_rate": 5e-06,
      "loss": 0.3869,
      "step": 13010
    },
    {
      "epoch": 2.817571954122484,
      "grad_norm": 0.8026133483602639,
      "learning_rate": 5e-06,
      "loss": 0.3957,
      "step": 13020
    },
    {
      "epoch": 2.819735987881411,
      "grad_norm": 0.744545263944405,
      "learning_rate": 5e-06,
      "loss": 0.386,
      "step": 13030
    },
    {
      "epoch": 2.821900021640338,
      "grad_norm": 0.7535839428276422,
      "learning_rate": 5e-06,
      "loss": 0.3808,
      "step": 13040
    },
    {
      "epoch": 2.824064055399264,
      "grad_norm": 0.7957853276701218,
      "learning_rate": 5e-06,
      "loss": 0.3793,
      "step": 13050
    },
    {
      "epoch": 2.826228089158191,
      "grad_norm": 0.7665463998428926,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 13060
    },
    {
      "epoch": 2.8283921229171174,
      "grad_norm": 0.7747068764554864,
      "learning_rate": 5e-06,
      "loss": 0.3899,
      "step": 13070
    },
    {
      "epoch": 2.830556156676044,
      "grad_norm": 0.7529729147926931,
      "learning_rate": 5e-06,
      "loss": 0.3986,
      "step": 13080
    },
    {
      "epoch": 2.832720190434971,
      "grad_norm": 0.7714949651199653,
      "learning_rate": 5e-06,
      "loss": 0.3849,
      "step": 13090
    },
    {
      "epoch": 2.8348842241938974,
      "grad_norm": 0.757669054423436,
      "learning_rate": 5e-06,
      "loss": 0.3973,
      "step": 13100
    },
    {
      "epoch": 2.8370482579528242,
      "grad_norm": 0.7109714152621212,
      "learning_rate": 5e-06,
      "loss": 0.3951,
      "step": 13110
    },
    {
      "epoch": 2.8392122917117506,
      "grad_norm": 0.7676176881192389,
      "learning_rate": 5e-06,
      "loss": 0.3878,
      "step": 13120
    },
    {
      "epoch": 2.8413763254706774,
      "grad_norm": 0.7850663677878515,
      "learning_rate": 5e-06,
      "loss": 0.3932,
      "step": 13130
    },
    {
      "epoch": 2.8435403592296042,
      "grad_norm": 0.7225327094498046,
      "learning_rate": 5e-06,
      "loss": 0.3831,
      "step": 13140
    },
    {
      "epoch": 2.8457043929885306,
      "grad_norm": 0.7817333715491703,
      "learning_rate": 5e-06,
      "loss": 0.3895,
      "step": 13150
    },
    {
      "epoch": 2.847868426747457,
      "grad_norm": 0.7709262647174362,
      "learning_rate": 5e-06,
      "loss": 0.3903,
      "step": 13160
    },
    {
      "epoch": 2.850032460506384,
      "grad_norm": 0.7521690368016893,
      "learning_rate": 5e-06,
      "loss": 0.3841,
      "step": 13170
    },
    {
      "epoch": 2.8521964942653106,
      "grad_norm": 0.7654288882047556,
      "learning_rate": 5e-06,
      "loss": 0.3965,
      "step": 13180
    },
    {
      "epoch": 2.8543605280242375,
      "grad_norm": 0.7797805693245492,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 13190
    },
    {
      "epoch": 2.856524561783164,
      "grad_norm": 0.767159401606924,
      "learning_rate": 5e-06,
      "loss": 0.3967,
      "step": 13200
    },
    {
      "epoch": 2.85868859554209,
      "grad_norm": 0.7700593597456216,
      "learning_rate": 5e-06,
      "loss": 0.3818,
      "step": 13210
    },
    {
      "epoch": 2.860852629301017,
      "grad_norm": 0.7699670336942503,
      "learning_rate": 5e-06,
      "loss": 0.3981,
      "step": 13220
    },
    {
      "epoch": 2.863016663059944,
      "grad_norm": 0.7532067312686156,
      "learning_rate": 5e-06,
      "loss": 0.3845,
      "step": 13230
    },
    {
      "epoch": 2.86518069681887,
      "grad_norm": 0.7700933480079477,
      "learning_rate": 5e-06,
      "loss": 0.3916,
      "step": 13240
    },
    {
      "epoch": 2.867344730577797,
      "grad_norm": 0.7466578230564417,
      "learning_rate": 5e-06,
      "loss": 0.3894,
      "step": 13250
    },
    {
      "epoch": 2.8695087643367234,
      "grad_norm": 0.7937501563778528,
      "learning_rate": 5e-06,
      "loss": 0.3816,
      "step": 13260
    },
    {
      "epoch": 2.8716727980956502,
      "grad_norm": 0.7389001187669237,
      "learning_rate": 5e-06,
      "loss": 0.3728,
      "step": 13270
    },
    {
      "epoch": 2.873836831854577,
      "grad_norm": 0.7578905999176232,
      "learning_rate": 5e-06,
      "loss": 0.3812,
      "step": 13280
    },
    {
      "epoch": 2.8760008656135034,
      "grad_norm": 0.8151694337796392,
      "learning_rate": 5e-06,
      "loss": 0.3861,
      "step": 13290
    },
    {
      "epoch": 2.8781648993724303,
      "grad_norm": 0.7471342559011176,
      "learning_rate": 5e-06,
      "loss": 0.3888,
      "step": 13300
    },
    {
      "epoch": 2.8803289331313566,
      "grad_norm": 0.7422252462621732,
      "learning_rate": 5e-06,
      "loss": 0.3922,
      "step": 13310
    },
    {
      "epoch": 2.8824929668902834,
      "grad_norm": 0.7615451869859242,
      "learning_rate": 5e-06,
      "loss": 0.3829,
      "step": 13320
    },
    {
      "epoch": 2.8846570006492103,
      "grad_norm": 0.8256789814798213,
      "learning_rate": 5e-06,
      "loss": 0.3952,
      "step": 13330
    },
    {
      "epoch": 2.8868210344081366,
      "grad_norm": 0.76680451948221,
      "learning_rate": 5e-06,
      "loss": 0.3956,
      "step": 13340
    },
    {
      "epoch": 2.8889850681670635,
      "grad_norm": 0.7439841068189852,
      "learning_rate": 5e-06,
      "loss": 0.3776,
      "step": 13350
    },
    {
      "epoch": 2.89114910192599,
      "grad_norm": 0.7398497595894248,
      "learning_rate": 5e-06,
      "loss": 0.3877,
      "step": 13360
    },
    {
      "epoch": 2.8933131356849167,
      "grad_norm": 0.7739248139645774,
      "learning_rate": 5e-06,
      "loss": 0.3929,
      "step": 13370
    },
    {
      "epoch": 2.8954771694438435,
      "grad_norm": 0.8440378872399906,
      "learning_rate": 5e-06,
      "loss": 0.3876,
      "step": 13380
    },
    {
      "epoch": 2.89764120320277,
      "grad_norm": 0.7705375133560814,
      "learning_rate": 5e-06,
      "loss": 0.3871,
      "step": 13390
    },
    {
      "epoch": 2.8998052369616967,
      "grad_norm": 0.7344474907711162,
      "learning_rate": 5e-06,
      "loss": 0.3988,
      "step": 13400
    },
    {
      "epoch": 2.901969270720623,
      "grad_norm": 0.7691718865505679,
      "learning_rate": 5e-06,
      "loss": 0.392,
      "step": 13410
    },
    {
      "epoch": 2.90413330447955,
      "grad_norm": 0.7491038351612413,
      "learning_rate": 5e-06,
      "loss": 0.3927,
      "step": 13420
    },
    {
      "epoch": 2.9062973382384767,
      "grad_norm": 0.7519583645050334,
      "learning_rate": 5e-06,
      "loss": 0.3845,
      "step": 13430
    },
    {
      "epoch": 2.908461371997403,
      "grad_norm": 0.8244800954009015,
      "learning_rate": 5e-06,
      "loss": 0.3832,
      "step": 13440
    },
    {
      "epoch": 2.91062540575633,
      "grad_norm": 0.8157251210241125,
      "learning_rate": 5e-06,
      "loss": 0.3805,
      "step": 13450
    },
    {
      "epoch": 2.9127894395152563,
      "grad_norm": 0.7721623213908909,
      "learning_rate": 5e-06,
      "loss": 0.3942,
      "step": 13460
    },
    {
      "epoch": 2.914953473274183,
      "grad_norm": 0.7726737654084024,
      "learning_rate": 5e-06,
      "loss": 0.3931,
      "step": 13470
    },
    {
      "epoch": 2.91711750703311,
      "grad_norm": 0.7550960174025876,
      "learning_rate": 5e-06,
      "loss": 0.3885,
      "step": 13480
    },
    {
      "epoch": 2.9192815407920363,
      "grad_norm": 0.7679327991153526,
      "learning_rate": 5e-06,
      "loss": 0.3964,
      "step": 13490
    },
    {
      "epoch": 2.921445574550963,
      "grad_norm": 0.7823759384711927,
      "learning_rate": 5e-06,
      "loss": 0.3809,
      "step": 13500
    },
    {
      "epoch": 2.9236096083098895,
      "grad_norm": 0.7512641558224188,
      "learning_rate": 5e-06,
      "loss": 0.3842,
      "step": 13510
    },
    {
      "epoch": 2.9257736420688163,
      "grad_norm": 0.782519582204221,
      "learning_rate": 5e-06,
      "loss": 0.3942,
      "step": 13520
    },
    {
      "epoch": 2.927937675827743,
      "grad_norm": 0.7935222646576868,
      "learning_rate": 5e-06,
      "loss": 0.3871,
      "step": 13530
    },
    {
      "epoch": 2.9301017095866695,
      "grad_norm": 0.7490260485361179,
      "learning_rate": 5e-06,
      "loss": 0.3924,
      "step": 13540
    },
    {
      "epoch": 2.9322657433455963,
      "grad_norm": 0.7829574236017007,
      "learning_rate": 5e-06,
      "loss": 0.387,
      "step": 13550
    },
    {
      "epoch": 2.9344297771045227,
      "grad_norm": 0.8053438714777432,
      "learning_rate": 5e-06,
      "loss": 0.3997,
      "step": 13560
    },
    {
      "epoch": 2.9365938108634495,
      "grad_norm": 0.789271548816451,
      "learning_rate": 5e-06,
      "loss": 0.3921,
      "step": 13570
    },
    {
      "epoch": 2.9387578446223763,
      "grad_norm": 0.7620667953803887,
      "learning_rate": 5e-06,
      "loss": 0.3945,
      "step": 13580
    },
    {
      "epoch": 2.9409218783813027,
      "grad_norm": 0.8157019776942496,
      "learning_rate": 5e-06,
      "loss": 0.3973,
      "step": 13590
    },
    {
      "epoch": 2.9430859121402295,
      "grad_norm": 0.7566983087966317,
      "learning_rate": 5e-06,
      "loss": 0.3973,
      "step": 13600
    },
    {
      "epoch": 2.945249945899156,
      "grad_norm": 0.7850341888405364,
      "learning_rate": 5e-06,
      "loss": 0.3852,
      "step": 13610
    },
    {
      "epoch": 2.9474139796580827,
      "grad_norm": 0.7756606362123769,
      "learning_rate": 5e-06,
      "loss": 0.3873,
      "step": 13620
    },
    {
      "epoch": 2.9495780134170095,
      "grad_norm": 0.7582467951052595,
      "learning_rate": 5e-06,
      "loss": 0.4002,
      "step": 13630
    },
    {
      "epoch": 2.951742047175936,
      "grad_norm": 0.7922689379335354,
      "learning_rate": 5e-06,
      "loss": 0.3948,
      "step": 13640
    },
    {
      "epoch": 2.9539060809348627,
      "grad_norm": 0.7435887161247465,
      "learning_rate": 5e-06,
      "loss": 0.3934,
      "step": 13650
    },
    {
      "epoch": 2.956070114693789,
      "grad_norm": 0.7677386746970113,
      "learning_rate": 5e-06,
      "loss": 0.3842,
      "step": 13660
    },
    {
      "epoch": 2.958234148452716,
      "grad_norm": 0.7621270826776456,
      "learning_rate": 5e-06,
      "loss": 0.3882,
      "step": 13670
    },
    {
      "epoch": 2.9603981822116427,
      "grad_norm": 0.7115127400494038,
      "learning_rate": 5e-06,
      "loss": 0.3869,
      "step": 13680
    },
    {
      "epoch": 2.962562215970569,
      "grad_norm": 0.7875693326081818,
      "learning_rate": 5e-06,
      "loss": 0.3818,
      "step": 13690
    },
    {
      "epoch": 2.964726249729496,
      "grad_norm": 0.7236217251683649,
      "learning_rate": 5e-06,
      "loss": 0.3817,
      "step": 13700
    },
    {
      "epoch": 2.9668902834884223,
      "grad_norm": 0.8125742954355849,
      "learning_rate": 5e-06,
      "loss": 0.3854,
      "step": 13710
    },
    {
      "epoch": 2.969054317247349,
      "grad_norm": 0.7524427643623132,
      "learning_rate": 5e-06,
      "loss": 0.3903,
      "step": 13720
    },
    {
      "epoch": 2.971218351006276,
      "grad_norm": 0.7390063903999405,
      "learning_rate": 5e-06,
      "loss": 0.3808,
      "step": 13730
    },
    {
      "epoch": 2.9733823847652023,
      "grad_norm": 0.7511379236710563,
      "learning_rate": 5e-06,
      "loss": 0.3826,
      "step": 13740
    },
    {
      "epoch": 2.9755464185241287,
      "grad_norm": 0.8210016741195086,
      "learning_rate": 5e-06,
      "loss": 0.3935,
      "step": 13750
    },
    {
      "epoch": 2.9777104522830555,
      "grad_norm": 0.7700571385479135,
      "learning_rate": 5e-06,
      "loss": 0.3947,
      "step": 13760
    },
    {
      "epoch": 2.9798744860419824,
      "grad_norm": 0.7901788530732445,
      "learning_rate": 5e-06,
      "loss": 0.3918,
      "step": 13770
    },
    {
      "epoch": 2.982038519800909,
      "grad_norm": 0.790326497189608,
      "learning_rate": 5e-06,
      "loss": 0.395,
      "step": 13780
    },
    {
      "epoch": 2.9842025535598355,
      "grad_norm": 0.7970458568075771,
      "learning_rate": 5e-06,
      "loss": 0.3838,
      "step": 13790
    },
    {
      "epoch": 2.986366587318762,
      "grad_norm": 0.7937921564780092,
      "learning_rate": 5e-06,
      "loss": 0.3871,
      "step": 13800
    },
    {
      "epoch": 2.9885306210776887,
      "grad_norm": 0.7869992321634875,
      "learning_rate": 5e-06,
      "loss": 0.3861,
      "step": 13810
    },
    {
      "epoch": 2.9906946548366156,
      "grad_norm": 0.7507057758028739,
      "learning_rate": 5e-06,
      "loss": 0.3863,
      "step": 13820
    },
    {
      "epoch": 2.992858688595542,
      "grad_norm": 0.8051759990871201,
      "learning_rate": 5e-06,
      "loss": 0.3811,
      "step": 13830
    },
    {
      "epoch": 2.9950227223544688,
      "grad_norm": 0.7596345710239549,
      "learning_rate": 5e-06,
      "loss": 0.3904,
      "step": 13840
    },
    {
      "epoch": 2.997186756113395,
      "grad_norm": 0.7917910564755264,
      "learning_rate": 5e-06,
      "loss": 0.3977,
      "step": 13850
    },
    {
      "epoch": 2.999350789872322,
      "grad_norm": 0.732648425467274,
      "learning_rate": 5e-06,
      "loss": 0.3911,
      "step": 13860
    },
    {
      "epoch": 3.0,
      "eval_loss": 0.546510636806488,
      "eval_runtime": 589.5092,
      "eval_samples_per_second": 26.402,
      "eval_steps_per_second": 0.414,
      "step": 13863
    },
    {
      "epoch": 3.0,
      "step": 13863,
      "total_flos": 3633869689454592.0,
      "train_loss": 0.474967997638056,
      "train_runtime": 94524.6869,
      "train_samples_per_second": 9.385,
      "train_steps_per_second": 0.147
    }
  ],
  "logging_steps": 10,
  "max_steps": 13863,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3633869689454592.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}