{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.5878606770195464,
  "eval_steps": 375,
  "global_step": 1500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0003919071180130309,
      "grad_norm": 2.4371085166931152,
      "learning_rate": 2e-05,
      "loss": 1.7734,
      "step": 1
    },
    {
      "epoch": 0.0007838142360260618,
      "grad_norm": 2.9669151306152344,
      "learning_rate": 4e-05,
      "loss": 1.4547,
      "step": 2
    },
    {
      "epoch": 0.0011757213540390928,
      "grad_norm": 3.8742661476135254,
      "learning_rate": 6e-05,
      "loss": 1.9772,
      "step": 3
    },
    {
      "epoch": 0.0015676284720521237,
      "grad_norm": 3.457343816757202,
      "learning_rate": 8e-05,
      "loss": 1.5527,
      "step": 4
    },
    {
      "epoch": 0.0019595355900651544,
      "grad_norm": 5.118892669677734,
      "learning_rate": 0.0001,
      "loss": 2.249,
      "step": 5
    },
    {
      "epoch": 0.0023514427080781855,
      "grad_norm": 4.277401447296143,
      "learning_rate": 0.00012,
      "loss": 2.0634,
      "step": 6
    },
    {
      "epoch": 0.0027433498260912162,
      "grad_norm": 4.400008201599121,
      "learning_rate": 0.00014,
      "loss": 1.5298,
      "step": 7
    },
    {
      "epoch": 0.0031352569441042474,
      "grad_norm": 5.6465301513671875,
      "learning_rate": 0.00016,
      "loss": 2.3682,
      "step": 8
    },
    {
      "epoch": 0.003527164062117278,
      "grad_norm": 5.808071136474609,
      "learning_rate": 0.00018,
      "loss": 2.019,
      "step": 9
    },
    {
      "epoch": 0.003919071180130309,
      "grad_norm": 5.150087356567383,
      "learning_rate": 0.0002,
      "loss": 2.092,
      "step": 10
    },
    {
      "epoch": 0.00431097829814334,
      "grad_norm": 5.8296003341674805,
      "learning_rate": 0.00019999977772170748,
      "loss": 2.4594,
      "step": 11
    },
    {
      "epoch": 0.004702885416156371,
      "grad_norm": 6.403761863708496,
      "learning_rate": 0.00019999911088781805,
      "loss": 2.0065,
      "step": 12
    },
    {
      "epoch": 0.005094792534169402,
      "grad_norm": 6.481574058532715,
      "learning_rate": 0.0001999979995012962,
      "loss": 2.3399,
      "step": 13
    },
    {
      "epoch": 0.0054866996521824325,
      "grad_norm": 6.5774617195129395,
      "learning_rate": 0.00019999644356708261,
      "loss": 2.5916,
      "step": 14
    },
    {
      "epoch": 0.005878606770195464,
      "grad_norm": 6.851273059844971,
      "learning_rate": 0.00019999444309209432,
      "loss": 2.1913,
      "step": 15
    },
    {
      "epoch": 0.006270513888208495,
      "grad_norm": 6.6362714767456055,
      "learning_rate": 0.0001999919980852246,
      "loss": 2.2232,
      "step": 16
    },
    {
      "epoch": 0.006662421006221526,
      "grad_norm": 6.993611812591553,
      "learning_rate": 0.00019998910855734288,
      "loss": 2.4459,
      "step": 17
    },
    {
      "epoch": 0.007054328124234556,
      "grad_norm": 6.401386737823486,
      "learning_rate": 0.0001999857745212947,
      "loss": 2.4317,
      "step": 18
    },
    {
      "epoch": 0.007446235242247587,
      "grad_norm": 6.840269565582275,
      "learning_rate": 0.00019998199599190178,
      "loss": 2.4664,
      "step": 19
    },
    {
      "epoch": 0.007838142360260618,
      "grad_norm": 7.5184006690979,
      "learning_rate": 0.0001999777729859618,
      "loss": 2.5693,
      "step": 20
    },
    {
      "epoch": 0.00823004947827365,
      "grad_norm": 6.816198825836182,
      "learning_rate": 0.00019997310552224846,
      "loss": 2.1192,
      "step": 21
    },
    {
      "epoch": 0.00862195659628668,
      "grad_norm": 7.093342304229736,
      "learning_rate": 0.00019996799362151122,
      "loss": 1.9554,
      "step": 22
    },
    {
      "epoch": 0.009013863714299712,
      "grad_norm": 7.271922588348389,
      "learning_rate": 0.00019996243730647538,
      "loss": 2.3086,
      "step": 23
    },
    {
      "epoch": 0.009405770832312742,
      "grad_norm": 7.031599044799805,
      "learning_rate": 0.00019995643660184191,
      "loss": 2.3936,
      "step": 24
    },
    {
      "epoch": 0.009797677950325772,
      "grad_norm": 6.852816581726074,
      "learning_rate": 0.00019994999153428737,
      "loss": 2.1071,
      "step": 25
    },
    {
      "epoch": 0.010189585068338804,
      "grad_norm": 7.014102458953857,
      "learning_rate": 0.00019994310213246368,
      "loss": 2.1758,
      "step": 26
    },
    {
      "epoch": 0.010581492186351835,
      "grad_norm": 7.79160213470459,
      "learning_rate": 0.00019993576842699816,
      "loss": 2.4044,
      "step": 27
    },
    {
      "epoch": 0.010973399304364865,
      "grad_norm": 8.982135772705078,
      "learning_rate": 0.0001999279904504933,
      "loss": 2.6033,
      "step": 28
    },
    {
      "epoch": 0.011365306422377897,
      "grad_norm": 9.374879837036133,
      "learning_rate": 0.00019991976823752653,
      "loss": 2.2647,
      "step": 29
    },
    {
      "epoch": 0.011757213540390927,
      "grad_norm": 9.310467720031738,
      "learning_rate": 0.00019991110182465032,
      "loss": 2.5877,
      "step": 30
    },
    {
      "epoch": 0.012149120658403957,
      "grad_norm": 7.976422309875488,
      "learning_rate": 0.00019990199125039174,
      "loss": 2.4156,
      "step": 31
    },
    {
      "epoch": 0.01254102777641699,
      "grad_norm": 12.730584144592285,
      "learning_rate": 0.00019989243655525247,
      "loss": 2.4063,
      "step": 32
    },
    {
      "epoch": 0.01293293489443002,
      "grad_norm": 9.110078811645508,
      "learning_rate": 0.00019988243778170853,
      "loss": 2.8182,
      "step": 33
    },
    {
      "epoch": 0.013324842012443052,
      "grad_norm": 7.658226490020752,
      "learning_rate": 0.0001998719949742101,
      "loss": 2.3453,
      "step": 34
    },
    {
      "epoch": 0.013716749130456082,
      "grad_norm": 7.961971759796143,
      "learning_rate": 0.0001998611081791814,
      "loss": 2.3757,
      "step": 35
    },
    {
      "epoch": 0.014108656248469112,
      "grad_norm": 7.835509777069092,
      "learning_rate": 0.00019984977744502038,
      "loss": 2.3206,
      "step": 36
    },
    {
      "epoch": 0.014500563366482144,
      "grad_norm": 7.709591388702393,
      "learning_rate": 0.00019983800282209857,
      "loss": 2.2347,
      "step": 37
    },
    {
      "epoch": 0.014892470484495175,
      "grad_norm": 9.329100608825684,
      "learning_rate": 0.00019982578436276082,
      "loss": 2.5129,
      "step": 38
    },
    {
      "epoch": 0.015284377602508205,
      "grad_norm": 8.68140983581543,
      "learning_rate": 0.00019981312212132512,
      "loss": 2.634,
      "step": 39
    },
    {
      "epoch": 0.015676284720521235,
      "grad_norm": 7.602488994598389,
      "learning_rate": 0.00019980001615408228,
      "loss": 1.8818,
      "step": 40
    },
    {
      "epoch": 0.01606819183853427,
      "grad_norm": 7.863831996917725,
      "learning_rate": 0.00019978646651929572,
      "loss": 2.1863,
      "step": 41
    },
    {
      "epoch": 0.0164600989565473,
      "grad_norm": 7.6244025230407715,
      "learning_rate": 0.00019977247327720128,
      "loss": 2.2169,
      "step": 42
    },
    {
      "epoch": 0.01685200607456033,
      "grad_norm": 8.031611442565918,
      "learning_rate": 0.0001997580364900068,
      "loss": 2.2365,
      "step": 43
    },
    {
      "epoch": 0.01724391319257336,
      "grad_norm": 8.743836402893066,
      "learning_rate": 0.000199743156221892,
      "loss": 2.1857,
      "step": 44
    },
    {
      "epoch": 0.01763582031058639,
      "grad_norm": 8.32661247253418,
      "learning_rate": 0.00019972783253900808,
      "loss": 2.3725,
      "step": 45
    },
    {
      "epoch": 0.018027727428599424,
      "grad_norm": 10.382034301757812,
      "learning_rate": 0.00019971206550947748,
      "loss": 2.5629,
      "step": 46
    },
    {
      "epoch": 0.018419634546612454,
      "grad_norm": 9.828290939331055,
      "learning_rate": 0.00019969585520339354,
      "loss": 2.297,
      "step": 47
    },
    {
      "epoch": 0.018811541664625484,
      "grad_norm": 10.920280456542969,
      "learning_rate": 0.0001996792016928203,
      "loss": 2.6634,
      "step": 48
    },
    {
      "epoch": 0.019203448782638514,
      "grad_norm": 11.918550491333008,
      "learning_rate": 0.00019966210505179197,
      "loss": 2.8942,
      "step": 49
    },
    {
      "epoch": 0.019595355900651545,
      "grad_norm": 16.75621795654297,
      "learning_rate": 0.00019964456535631286,
      "loss": 3.1331,
      "step": 50
    },
    {
      "epoch": 0.019987263018664575,
      "grad_norm": 6.9211578369140625,
      "learning_rate": 0.0001996265826843568,
      "loss": 2.1134,
      "step": 51
    },
    {
      "epoch": 0.02037917013667761,
      "grad_norm": 5.556185245513916,
      "learning_rate": 0.00019960815711586696,
      "loss": 1.8269,
      "step": 52
    },
    {
      "epoch": 0.02077107725469064,
      "grad_norm": 4.672778129577637,
      "learning_rate": 0.00019958928873275539,
      "loss": 2.0583,
      "step": 53
    },
    {
      "epoch": 0.02116298437270367,
      "grad_norm": 4.661110877990723,
      "learning_rate": 0.00019956997761890277,
      "loss": 2.253,
      "step": 54
    },
    {
      "epoch": 0.0215548914907167,
      "grad_norm": 3.444990396499634,
      "learning_rate": 0.00019955022386015792,
      "loss": 2.0609,
      "step": 55
    },
    {
      "epoch": 0.02194679860872973,
      "grad_norm": 3.829105854034424,
      "learning_rate": 0.00019953002754433743,
      "loss": 2.0191,
      "step": 56
    },
    {
      "epoch": 0.022338705726742764,
      "grad_norm": 3.4666848182678223,
      "learning_rate": 0.00019950938876122542,
      "loss": 1.7476,
      "step": 57
    },
    {
      "epoch": 0.022730612844755794,
      "grad_norm": 3.9293630123138428,
      "learning_rate": 0.00019948830760257291,
      "loss": 2.1581,
      "step": 58
    },
    {
      "epoch": 0.023122519962768824,
      "grad_norm": 4.393327713012695,
      "learning_rate": 0.0001994667841620976,
      "loss": 2.1992,
      "step": 59
    },
    {
      "epoch": 0.023514427080781854,
      "grad_norm": 3.8993756771087646,
      "learning_rate": 0.00019944481853548335,
      "loss": 1.7487,
      "step": 60
    },
    {
      "epoch": 0.023906334198794885,
      "grad_norm": 3.933692693710327,
      "learning_rate": 0.00019942241082037982,
      "loss": 2.1577,
      "step": 61
    },
    {
      "epoch": 0.024298241316807915,
      "grad_norm": 4.808890342712402,
      "learning_rate": 0.00019939956111640197,
      "loss": 2.4188,
      "step": 62
    },
    {
      "epoch": 0.02469014843482095,
      "grad_norm": 4.369194507598877,
      "learning_rate": 0.00019937626952512964,
      "loss": 2.2937,
      "step": 63
    },
    {
      "epoch": 0.02508205555283398,
      "grad_norm": 4.327810764312744,
      "learning_rate": 0.0001993525361501072,
      "loss": 2.5809,
      "step": 64
    },
    {
      "epoch": 0.02547396267084701,
      "grad_norm": 4.78913688659668,
      "learning_rate": 0.00019932836109684286,
      "loss": 2.4685,
      "step": 65
    },
    {
      "epoch": 0.02586586978886004,
      "grad_norm": 4.4713053703308105,
      "learning_rate": 0.00019930374447280845,
      "loss": 2.2876,
      "step": 66
    },
    {
      "epoch": 0.02625777690687307,
      "grad_norm": 5.143622398376465,
      "learning_rate": 0.00019927868638743875,
      "loss": 2.3328,
      "step": 67
    },
    {
      "epoch": 0.026649684024886103,
      "grad_norm": 4.836973667144775,
      "learning_rate": 0.0001992531869521312,
      "loss": 1.9804,
      "step": 68
    },
    {
      "epoch": 0.027041591142899134,
      "grad_norm": 4.850903511047363,
      "learning_rate": 0.00019922724628024515,
      "loss": 2.515,
      "step": 69
    },
    {
      "epoch": 0.027433498260912164,
      "grad_norm": 4.306402683258057,
      "learning_rate": 0.0001992008644871016,
      "loss": 1.9078,
      "step": 70
    },
    {
      "epoch": 0.027825405378925194,
      "grad_norm": 4.903380870819092,
      "learning_rate": 0.00019917404168998256,
      "loss": 2.188,
      "step": 71
    },
    {
      "epoch": 0.028217312496938225,
      "grad_norm": 5.258169651031494,
      "learning_rate": 0.0001991467780081305,
      "loss": 2.6452,
      "step": 72
    },
    {
      "epoch": 0.028609219614951255,
      "grad_norm": 5.272015571594238,
      "learning_rate": 0.00019911907356274795,
      "loss": 2.1523,
      "step": 73
    },
    {
      "epoch": 0.02900112673296429,
      "grad_norm": 5.139151573181152,
      "learning_rate": 0.00019909092847699683,
      "loss": 2.3782,
      "step": 74
    },
    {
      "epoch": 0.02939303385097732,
      "grad_norm": 5.182451248168945,
      "learning_rate": 0.00019906234287599798,
      "loss": 2.537,
      "step": 75
    },
    {
      "epoch": 0.02978494096899035,
      "grad_norm": 5.881833076477051,
      "learning_rate": 0.00019903331688683057,
      "loss": 2.2717,
      "step": 76
    },
    {
      "epoch": 0.03017684808700338,
      "grad_norm": 5.213593482971191,
      "learning_rate": 0.00019900385063853154,
      "loss": 2.2767,
      "step": 77
    },
    {
      "epoch": 0.03056875520501641,
      "grad_norm": 5.310039043426514,
      "learning_rate": 0.00019897394426209505,
      "loss": 1.9019,
      "step": 78
    },
    {
      "epoch": 0.030960662323029443,
      "grad_norm": 5.669753551483154,
      "learning_rate": 0.00019894359789047187,
      "loss": 2.3209,
      "step": 79
    },
    {
      "epoch": 0.03135256944104247,
      "grad_norm": 5.6143035888671875,
      "learning_rate": 0.00019891281165856873,
      "loss": 2.2974,
      "step": 80
    },
    {
      "epoch": 0.031744476559055504,
      "grad_norm": 5.407499313354492,
      "learning_rate": 0.00019888158570324795,
      "loss": 2.6135,
      "step": 81
    },
    {
      "epoch": 0.03213638367706854,
      "grad_norm": 5.769077777862549,
      "learning_rate": 0.0001988499201633265,
      "loss": 2.6562,
      "step": 82
    },
    {
      "epoch": 0.032528290795081564,
      "grad_norm": 5.897034645080566,
      "learning_rate": 0.00019881781517957562,
      "loss": 2.4627,
      "step": 83
    },
    {
      "epoch": 0.0329201979130946,
      "grad_norm": 6.268710613250732,
      "learning_rate": 0.0001987852708947202,
      "loss": 2.4825,
      "step": 84
    },
    {
      "epoch": 0.033312105031107625,
      "grad_norm": 5.6505446434021,
      "learning_rate": 0.00019875228745343794,
      "loss": 2.4253,
      "step": 85
    },
    {
      "epoch": 0.03370401214912066,
      "grad_norm": 5.68675422668457,
      "learning_rate": 0.0001987188650023589,
      "loss": 2.4389,
      "step": 86
    },
    {
      "epoch": 0.03409591926713369,
      "grad_norm": 5.489636421203613,
      "learning_rate": 0.0001986850036900648,
      "loss": 2.1438,
      "step": 87
    },
    {
      "epoch": 0.03448782638514672,
      "grad_norm": 6.088352203369141,
      "learning_rate": 0.00019865070366708836,
      "loss": 2.2766,
      "step": 88
    },
    {
      "epoch": 0.03487973350315975,
      "grad_norm": 6.177804946899414,
      "learning_rate": 0.00019861596508591255,
      "loss": 2.5071,
      "step": 89
    },
    {
      "epoch": 0.03527164062117278,
      "grad_norm": 7.232726573944092,
      "learning_rate": 0.00019858078810097002,
      "loss": 2.5596,
      "step": 90
    },
    {
      "epoch": 0.035663547739185814,
      "grad_norm": 6.583539009094238,
      "learning_rate": 0.00019854517286864245,
      "loss": 2.1408,
      "step": 91
    },
    {
      "epoch": 0.03605545485719885,
      "grad_norm": 6.5620222091674805,
      "learning_rate": 0.0001985091195472596,
      "loss": 2.2073,
      "step": 92
    },
    {
      "epoch": 0.036447361975211874,
      "grad_norm": 7.709228515625,
      "learning_rate": 0.0001984726282970989,
      "loss": 2.4447,
      "step": 93
    },
    {
      "epoch": 0.03683926909322491,
      "grad_norm": 7.582402229309082,
      "learning_rate": 0.0001984356992803847,
      "loss": 2.4791,
      "step": 94
    },
    {
      "epoch": 0.037231176211237935,
      "grad_norm": 7.506585121154785,
      "learning_rate": 0.00019839833266128724,
      "loss": 2.4639,
      "step": 95
    },
    {
      "epoch": 0.03762308332925097,
      "grad_norm": 8.241847038269043,
      "learning_rate": 0.00019836052860592237,
      "loss": 2.6568,
      "step": 96
    },
    {
      "epoch": 0.038014990447263995,
      "grad_norm": 8.446549415588379,
      "learning_rate": 0.0001983222872823505,
      "loss": 2.6246,
      "step": 97
    },
    {
      "epoch": 0.03840689756527703,
      "grad_norm": 9.503031730651855,
      "learning_rate": 0.00019828360886057594,
      "loss": 2.5282,
      "step": 98
    },
    {
      "epoch": 0.03879880468329006,
      "grad_norm": 10.875327110290527,
      "learning_rate": 0.00019824449351254616,
      "loss": 2.6898,
      "step": 99
    },
    {
      "epoch": 0.03919071180130309,
      "grad_norm": 15.940088272094727,
      "learning_rate": 0.00019820494141215104,
      "loss": 2.2733,
      "step": 100
    },
    {
      "epoch": 0.03958261891931612,
      "grad_norm": 2.890528440475464,
      "learning_rate": 0.000198164952735222,
      "loss": 1.5781,
      "step": 101
    },
    {
      "epoch": 0.03997452603732915,
      "grad_norm": 3.628880500793457,
      "learning_rate": 0.00019812452765953135,
      "loss": 2.0458,
      "step": 102
    },
    {
      "epoch": 0.040366433155342184,
      "grad_norm": 3.4456260204315186,
      "learning_rate": 0.00019808366636479147,
      "loss": 1.5245,
      "step": 103
    },
    {
      "epoch": 0.04075834027335522,
      "grad_norm": 2.958019495010376,
      "learning_rate": 0.00019804236903265388,
      "loss": 1.6941,
      "step": 104
    },
    {
      "epoch": 0.041150247391368244,
      "grad_norm": 3.015265464782715,
      "learning_rate": 0.00019800063584670863,
      "loss": 1.6669,
      "step": 105
    },
    {
      "epoch": 0.04154215450938128,
      "grad_norm": 3.04085373878479,
      "learning_rate": 0.00019795846699248332,
      "loss": 1.8397,
      "step": 106
    },
    {
      "epoch": 0.041934061627394305,
      "grad_norm": 3.519737720489502,
      "learning_rate": 0.00019791586265744237,
      "loss": 2.2419,
      "step": 107
    },
    {
      "epoch": 0.04232596874540734,
      "grad_norm": 3.7447965145111084,
      "learning_rate": 0.00019787282303098617,
      "loss": 2.15,
      "step": 108
    },
    {
      "epoch": 0.04271787586342037,
      "grad_norm": 3.9315185546875,
      "learning_rate": 0.0001978293483044502,
      "loss": 2.5411,
      "step": 109
    },
    {
      "epoch": 0.0431097829814334,
      "grad_norm": 4.080818176269531,
      "learning_rate": 0.00019778543867110426,
      "loss": 2.372,
      "step": 110
    },
    {
      "epoch": 0.04350169009944643,
      "grad_norm": 3.9099278450012207,
      "learning_rate": 0.00019774109432615147,
      "loss": 2.2772,
      "step": 111
    },
    {
      "epoch": 0.04389359721745946,
      "grad_norm": 4.135699272155762,
      "learning_rate": 0.00019769631546672756,
      "loss": 2.5242,
      "step": 112
    },
    {
      "epoch": 0.04428550433547249,
      "grad_norm": 3.988497495651245,
      "learning_rate": 0.00019765110229189988,
      "loss": 2.6587,
      "step": 113
    },
    {
      "epoch": 0.04467741145348553,
      "grad_norm": 3.955430030822754,
      "learning_rate": 0.00019760545500266657,
      "loss": 2.3386,
      "step": 114
    },
    {
      "epoch": 0.045069318571498554,
      "grad_norm": 3.933865547180176,
      "learning_rate": 0.00019755937380195568,
      "loss": 2.2738,
      "step": 115
    },
    {
      "epoch": 0.04546122568951159,
      "grad_norm": 4.202532768249512,
      "learning_rate": 0.00019751285889462423,
      "loss": 2.4007,
      "step": 116
    },
    {
      "epoch": 0.045853132807524614,
      "grad_norm": 4.660050392150879,
      "learning_rate": 0.0001974659104874573,
      "loss": 2.4451,
      "step": 117
    },
    {
      "epoch": 0.04624503992553765,
      "grad_norm": 4.146893501281738,
      "learning_rate": 0.0001974185287891671,
      "loss": 2.4197,
      "step": 118
    },
    {
      "epoch": 0.04663694704355068,
      "grad_norm": 4.476595878601074,
      "learning_rate": 0.0001973707140103921,
      "loss": 2.2659,
      "step": 119
    },
    {
      "epoch": 0.04702885416156371,
      "grad_norm": 4.406892776489258,
      "learning_rate": 0.00019732246636369605,
      "loss": 2.1719,
      "step": 120
    },
    {
      "epoch": 0.04742076127957674,
      "grad_norm": 4.415012836456299,
      "learning_rate": 0.00019727378606356703,
      "loss": 2.297,
      "step": 121
    },
    {
      "epoch": 0.04781266839758977,
      "grad_norm": 4.748732566833496,
      "learning_rate": 0.00019722467332641656,
      "loss": 2.1139,
      "step": 122
    },
    {
      "epoch": 0.0482045755156028,
      "grad_norm": 5.826638221740723,
      "learning_rate": 0.00019717512837057855,
      "loss": 2.6594,
      "step": 123
    },
    {
      "epoch": 0.04859648263361583,
      "grad_norm": 4.8779401779174805,
      "learning_rate": 0.0001971251514163083,
      "loss": 2.3262,
      "step": 124
    },
    {
      "epoch": 0.048988389751628864,
      "grad_norm": 5.031620025634766,
      "learning_rate": 0.0001970747426857817,
      "loss": 2.3497,
      "step": 125
    },
    {
      "epoch": 0.0493802968696419,
      "grad_norm": 4.675917625427246,
      "learning_rate": 0.00019702390240309404,
      "loss": 2.0591,
      "step": 126
    },
    {
      "epoch": 0.049772203987654924,
      "grad_norm": 5.943875312805176,
      "learning_rate": 0.0001969726307942592,
      "loss": 2.4269,
      "step": 127
    },
    {
      "epoch": 0.05016411110566796,
      "grad_norm": 4.981433391571045,
      "learning_rate": 0.00019692092808720846,
      "loss": 2.4568,
      "step": 128
    },
    {
      "epoch": 0.050556018223680985,
      "grad_norm": 5.801111221313477,
      "learning_rate": 0.0001968687945117896,
      "loss": 2.3288,
      "step": 129
    },
    {
      "epoch": 0.05094792534169402,
      "grad_norm": 4.998170852661133,
      "learning_rate": 0.00019681623029976588,
      "loss": 2.4064,
      "step": 130
    },
    {
      "epoch": 0.05133983245970705,
      "grad_norm": 5.1848297119140625,
      "learning_rate": 0.00019676323568481498,
      "loss": 2.2295,
      "step": 131
    },
    {
      "epoch": 0.05173173957772008,
      "grad_norm": 5.462472438812256,
      "learning_rate": 0.00019670981090252792,
      "loss": 2.617,
      "step": 132
    },
    {
      "epoch": 0.05212364669573311,
      "grad_norm": 5.522066593170166,
      "learning_rate": 0.00019665595619040808,
      "loss": 2.4642,
      "step": 133
    },
    {
      "epoch": 0.05251555381374614,
      "grad_norm": 5.3126540184021,
      "learning_rate": 0.0001966016717878702,
      "loss": 2.2181,
      "step": 134
    },
    {
      "epoch": 0.05290746093175917,
      "grad_norm": 5.776416778564453,
      "learning_rate": 0.00019654695793623907,
      "loss": 2.3574,
      "step": 135
    },
    {
      "epoch": 0.05329936804977221,
      "grad_norm": 6.367827892303467,
      "learning_rate": 0.0001964918148787488,
      "loss": 2.4075,
      "step": 136
    },
    {
      "epoch": 0.053691275167785234,
      "grad_norm": 5.58833646774292,
      "learning_rate": 0.00019643624286054144,
      "loss": 2.3319,
      "step": 137
    },
    {
      "epoch": 0.05408318228579827,
      "grad_norm": 5.91164493560791,
      "learning_rate": 0.00019638024212866606,
      "loss": 2.382,
      "step": 138
    },
    {
      "epoch": 0.054475089403811294,
      "grad_norm": 6.908247470855713,
      "learning_rate": 0.0001963238129320776,
      "loss": 2.418,
      "step": 139
    },
    {
      "epoch": 0.05486699652182433,
      "grad_norm": 5.939513683319092,
      "learning_rate": 0.00019626695552163578,
      "loss": 2.1877,
      "step": 140
    },
    {
      "epoch": 0.05525890363983736,
      "grad_norm": 5.971606731414795,
      "learning_rate": 0.00019620967015010395,
      "loss": 2.1398,
      "step": 141
    },
    {
      "epoch": 0.05565081075785039,
      "grad_norm": 6.077500820159912,
      "learning_rate": 0.00019615195707214803,
      "loss": 2.2221,
      "step": 142
    },
    {
      "epoch": 0.05604271787586342,
      "grad_norm": 6.928093433380127,
      "learning_rate": 0.0001960938165443353,
      "loss": 2.317,
      "step": 143
    },
    {
      "epoch": 0.05643462499387645,
      "grad_norm": 7.14592170715332,
      "learning_rate": 0.00019603524882513327,
      "loss": 2.6168,
      "step": 144
    },
    {
      "epoch": 0.05682653211188948,
      "grad_norm": 6.889165878295898,
      "learning_rate": 0.0001959762541749086,
      "loss": 2.6895,
      "step": 145
    },
    {
      "epoch": 0.05721843922990251,
      "grad_norm": 8.114351272583008,
      "learning_rate": 0.00019591683285592593,
      "loss": 3.2964,
      "step": 146
    },
    {
      "epoch": 0.05761034634791554,
      "grad_norm": 7.2522993087768555,
      "learning_rate": 0.00019585698513234663,
      "loss": 2.307,
      "step": 147
    },
    {
      "epoch": 0.05800225346592858,
      "grad_norm": 8.145492553710938,
      "learning_rate": 0.0001957967112702277,
      "loss": 2.8325,
      "step": 148
    },
    {
      "epoch": 0.058394160583941604,
      "grad_norm": 8.395540237426758,
      "learning_rate": 0.00019573601153752052,
      "loss": 2.5268,
      "step": 149
    },
    {
      "epoch": 0.05878606770195464,
      "grad_norm": 11.565598487854004,
      "learning_rate": 0.00019567488620406983,
      "loss": 2.8473,
      "step": 150
    },
    {
      "epoch": 0.059177974819967664,
      "grad_norm": 2.2443935871124268,
      "learning_rate": 0.00019561333554161224,
      "loss": 1.7915,
      "step": 151
    },
    {
      "epoch": 0.0595698819379807,
      "grad_norm": 2.2534091472625732,
      "learning_rate": 0.0001955513598237753,
      "loss": 1.312,
      "step": 152
    },
    {
      "epoch": 0.05996178905599373,
      "grad_norm": 2.237297534942627,
      "learning_rate": 0.00019548895932607621,
      "loss": 1.5415,
      "step": 153
    },
    {
      "epoch": 0.06035369617400676,
      "grad_norm": 2.9409773349761963,
      "learning_rate": 0.00019542613432592038,
      "loss": 1.6718,
      "step": 154
    },
    {
      "epoch": 0.06074560329201979,
      "grad_norm": 3.303316116333008,
      "learning_rate": 0.00019536288510260056,
      "loss": 1.823,
      "step": 155
    },
    {
      "epoch": 0.06113751041003282,
      "grad_norm": 4.067633152008057,
      "learning_rate": 0.00019529921193729534,
      "loss": 2.3863,
      "step": 156
    },
    {
      "epoch": 0.06152941752804585,
      "grad_norm": 3.5242106914520264,
      "learning_rate": 0.00019523511511306793,
      "loss": 1.8536,
      "step": 157
    },
    {
      "epoch": 0.06192132464605889,
      "grad_norm": 3.330686330795288,
      "learning_rate": 0.000195170594914865,
      "loss": 1.9198,
      "step": 158
    },
    {
      "epoch": 0.062313231764071914,
      "grad_norm": 3.9775593280792236,
      "learning_rate": 0.00019510565162951537,
      "loss": 2.1291,
      "step": 159
    },
    {
      "epoch": 0.06270513888208494,
      "grad_norm": 4.036441326141357,
      "learning_rate": 0.00019504028554572864,
      "loss": 2.3018,
      "step": 160
    },
    {
      "epoch": 0.06309704600009798,
      "grad_norm": 3.449341058731079,
      "learning_rate": 0.00019497449695409408,
      "loss": 1.9706,
      "step": 161
    },
    {
      "epoch": 0.06348895311811101,
      "grad_norm": 4.01742696762085,
      "learning_rate": 0.00019490828614707916,
      "loss": 2.4563,
      "step": 162
    },
    {
      "epoch": 0.06388086023612403,
      "grad_norm": 3.7384417057037354,
      "learning_rate": 0.00019484165341902845,
      "loss": 2.405,
      "step": 163
    },
    {
      "epoch": 0.06427276735413708,
      "grad_norm": 3.8216614723205566,
      "learning_rate": 0.00019477459906616206,
      "loss": 2.3349,
      "step": 164
    },
    {
      "epoch": 0.0646646744721501,
      "grad_norm": 3.733248710632324,
      "learning_rate": 0.00019470712338657458,
      "loss": 2.172,
      "step": 165
    },
    {
      "epoch": 0.06505658159016313,
      "grad_norm": 4.645929336547852,
      "learning_rate": 0.0001946392266802336,
      "loss": 2.2428,
      "step": 166
    },
    {
      "epoch": 0.06544848870817616,
      "grad_norm": 4.384825229644775,
      "learning_rate": 0.0001945709092489783,
      "loss": 2.2313,
      "step": 167
    },
    {
      "epoch": 0.0658403958261892,
      "grad_norm": 3.8709452152252197,
      "learning_rate": 0.00019450217139651844,
      "loss": 2.207,
      "step": 168
    },
    {
      "epoch": 0.06623230294420222,
      "grad_norm": 4.236559867858887,
      "learning_rate": 0.0001944330134284326,
      "loss": 2.1996,
      "step": 169
    },
    {
      "epoch": 0.06662421006221525,
      "grad_norm": 4.030088901519775,
      "learning_rate": 0.00019436343565216711,
      "loss": 2.3297,
      "step": 170
    },
    {
      "epoch": 0.06701611718022829,
      "grad_norm": 4.6742143630981445,
      "learning_rate": 0.00019429343837703455,
      "loss": 2.4162,
      "step": 171
    },
    {
      "epoch": 0.06740802429824132,
      "grad_norm": 4.425167560577393,
      "learning_rate": 0.0001942230219142124,
      "loss": 2.2061,
      "step": 172
    },
    {
      "epoch": 0.06779993141625434,
      "grad_norm": 4.481382846832275,
      "learning_rate": 0.0001941521865767417,
      "loss": 1.9266,
      "step": 173
    },
    {
      "epoch": 0.06819183853426738,
      "grad_norm": 4.531622886657715,
      "learning_rate": 0.0001940809326795256,
      "loss": 2.2317,
      "step": 174
    },
    {
      "epoch": 0.06858374565228041,
      "grad_norm": 5.041552543640137,
      "learning_rate": 0.000194009260539328,
      "loss": 2.4007,
      "step": 175
    },
    {
      "epoch": 0.06897565277029344,
      "grad_norm": 5.10752534866333,
      "learning_rate": 0.0001939371704747721,
      "loss": 2.5272,
      "step": 176
    },
    {
      "epoch": 0.06936755988830647,
      "grad_norm": 5.61458683013916,
      "learning_rate": 0.00019386466280633906,
      "loss": 2.2779,
      "step": 177
    },
    {
      "epoch": 0.0697594670063195,
      "grad_norm": 5.012031555175781,
      "learning_rate": 0.00019379173785636646,
      "loss": 2.3196,
      "step": 178
    },
    {
      "epoch": 0.07015137412433253,
      "grad_norm": 5.032773017883301,
      "learning_rate": 0.000193718395949047,
      "loss": 2.1713,
      "step": 179
    },
    {
      "epoch": 0.07054328124234556,
      "grad_norm": 5.115957260131836,
      "learning_rate": 0.00019364463741042694,
      "loss": 2.3927,
      "step": 180
    },
    {
      "epoch": 0.0709351883603586,
      "grad_norm": 5.048713207244873,
      "learning_rate": 0.00019357046256840473,
      "loss": 2.5807,
      "step": 181
    },
    {
      "epoch": 0.07132709547837163,
      "grad_norm": 6.194578647613525,
      "learning_rate": 0.00019349587175272948,
      "loss": 2.8996,
      "step": 182
    },
    {
      "epoch": 0.07171900259638465,
      "grad_norm": 5.2247233390808105,
      "learning_rate": 0.0001934208652949996,
      "loss": 2.6243,
      "step": 183
    },
    {
      "epoch": 0.0721109097143977,
      "grad_norm": 5.3520731925964355,
      "learning_rate": 0.00019334544352866127,
      "loss": 2.4337,
      "step": 184
    },
    {
      "epoch": 0.07250281683241072,
      "grad_norm": 5.82279634475708,
      "learning_rate": 0.00019326960678900688,
      "loss": 2.5451,
      "step": 185
    },
    {
      "epoch": 0.07289472395042375,
      "grad_norm": 5.09345817565918,
      "learning_rate": 0.00019319335541317361,
      "loss": 2.2975,
      "step": 186
    },
    {
      "epoch": 0.07328663106843677,
      "grad_norm": 5.493734359741211,
      "learning_rate": 0.00019311668974014208,
      "loss": 2.1813,
      "step": 187
    },
    {
      "epoch": 0.07367853818644982,
      "grad_norm": 5.5384721755981445,
      "learning_rate": 0.00019303961011073447,
      "loss": 2.1508,
      "step": 188
    },
    {
      "epoch": 0.07407044530446284,
      "grad_norm": 5.593307018280029,
      "learning_rate": 0.00019296211686761346,
      "loss": 2.3944,
      "step": 189
    },
    {
      "epoch": 0.07446235242247587,
      "grad_norm": 5.579277038574219,
      "learning_rate": 0.00019288421035528028,
      "loss": 2.1628,
      "step": 190
    },
    {
      "epoch": 0.07485425954048891,
      "grad_norm": 5.322564601898193,
      "learning_rate": 0.00019280589092007352,
      "loss": 2.266,
      "step": 191
    },
    {
      "epoch": 0.07524616665850194,
      "grad_norm": 5.771650791168213,
      "learning_rate": 0.00019272715891016735,
      "loss": 2.4567,
      "step": 192
    },
    {
      "epoch": 0.07563807377651496,
      "grad_norm": 5.810284614562988,
      "learning_rate": 0.00019264801467557007,
      "loss": 2.3087,
      "step": 193
    },
    {
      "epoch": 0.07602998089452799,
      "grad_norm": 6.122025489807129,
      "learning_rate": 0.00019256845856812266,
      "loss": 2.0887,
      "step": 194
    },
    {
      "epoch": 0.07642188801254103,
      "grad_norm": 6.334949493408203,
      "learning_rate": 0.000192488490941497,
      "loss": 2.4949,
      "step": 195
    },
    {
      "epoch": 0.07681379513055406,
      "grad_norm": 6.712754726409912,
      "learning_rate": 0.00019240811215119448,
      "loss": 2.4567,
      "step": 196
    },
    {
      "epoch": 0.07720570224856708,
      "grad_norm": 7.652520656585693,
      "learning_rate": 0.00019232732255454422,
      "loss": 2.3585,
      "step": 197
    },
    {
      "epoch": 0.07759760936658013,
      "grad_norm": 8.646698951721191,
      "learning_rate": 0.00019224612251070175,
      "loss": 2.7979,
      "step": 198
    },
    {
      "epoch": 0.07798951648459315,
      "grad_norm": 8.767402648925781,
      "learning_rate": 0.0001921645123806472,
      "loss": 2.4685,
      "step": 199
    },
    {
      "epoch": 0.07838142360260618,
      "grad_norm": 13.505264282226562,
      "learning_rate": 0.0001920824925271838,
      "loss": 2.6335,
      "step": 200
    },
    {
      "epoch": 0.07877333072061922,
      "grad_norm": 2.2149970531463623,
      "learning_rate": 0.0001920000633149362,
      "loss": 1.4194,
      "step": 201
    },
    {
      "epoch": 0.07916523783863225,
      "grad_norm": 2.8648226261138916,
      "learning_rate": 0.00019191722511034884,
      "loss": 2.0748,
      "step": 202
    },
    {
      "epoch": 0.07955714495664527,
      "grad_norm": 2.7229807376861572,
      "learning_rate": 0.00019183397828168448,
      "loss": 1.7257,
      "step": 203
    },
    {
      "epoch": 0.0799490520746583,
      "grad_norm": 2.3575518131256104,
      "learning_rate": 0.00019175032319902234,
      "loss": 1.6773,
      "step": 204
    },
    {
      "epoch": 0.08034095919267134,
      "grad_norm": 3.080052614212036,
      "learning_rate": 0.00019166626023425662,
      "loss": 1.9105,
      "step": 205
    },
    {
      "epoch": 0.08073286631068437,
      "grad_norm": 2.6623058319091797,
      "learning_rate": 0.00019158178976109476,
      "loss": 1.8251,
      "step": 206
    },
    {
      "epoch": 0.0811247734286974,
      "grad_norm": 2.9029884338378906,
      "learning_rate": 0.0001914969121550558,
      "loss": 1.8388,
      "step": 207
    },
    {
      "epoch": 0.08151668054671043,
      "grad_norm": 3.3467230796813965,
      "learning_rate": 0.00019141162779346874,
      "loss": 2.0548,
      "step": 208
    },
    {
      "epoch": 0.08190858766472346,
      "grad_norm": 3.099581003189087,
      "learning_rate": 0.00019132593705547082,
      "loss": 2.1584,
      "step": 209
    },
    {
      "epoch": 0.08230049478273649,
      "grad_norm": 3.642759084701538,
      "learning_rate": 0.00019123984032200586,
      "loss": 2.2956,
      "step": 210
    },
    {
      "epoch": 0.08269240190074953,
      "grad_norm": 3.8225338459014893,
      "learning_rate": 0.00019115333797582254,
      "loss": 2.2096,
      "step": 211
    },
    {
      "epoch": 0.08308430901876256,
      "grad_norm": 3.777524471282959,
      "learning_rate": 0.00019106643040147278,
      "loss": 2.2618,
      "step": 212
    },
    {
      "epoch": 0.08347621613677558,
      "grad_norm": 3.4649555683135986,
      "learning_rate": 0.00019097911798530987,
      "loss": 2.1903,
      "step": 213
    },
    {
      "epoch": 0.08386812325478861,
      "grad_norm": 3.6316721439361572,
      "learning_rate": 0.00019089140111548696,
      "loss": 2.3619,
      "step": 214
    },
    {
      "epoch": 0.08426003037280165,
      "grad_norm": 4.004083633422852,
      "learning_rate": 0.00019080328018195513,
      "loss": 2.0776,
      "step": 215
    },
    {
      "epoch": 0.08465193749081468,
      "grad_norm": 4.068933486938477,
      "learning_rate": 0.0001907147555764618,
      "loss": 2.1852,
      "step": 216
    },
    {
      "epoch": 0.0850438446088277,
      "grad_norm": 3.9678616523742676,
      "learning_rate": 0.00019062582769254895,
      "loss": 2.3382,
      "step": 217
    },
    {
      "epoch": 0.08543575172684074,
      "grad_norm": 4.06026554107666,
      "learning_rate": 0.00019053649692555135,
      "loss": 2.305,
      "step": 218
    },
    {
      "epoch": 0.08582765884485377,
      "grad_norm": 4.151678085327148,
      "learning_rate": 0.00019044676367259476,
      "loss": 2.3729,
      "step": 219
    },
    {
      "epoch": 0.0862195659628668,
      "grad_norm": 3.966586112976074,
      "learning_rate": 0.00019035662833259432,
      "loss": 1.9456,
      "step": 220
    },
    {
      "epoch": 0.08661147308087982,
      "grad_norm": 4.350526332855225,
      "learning_rate": 0.00019026609130625257,
      "loss": 2.1743,
      "step": 221
    },
    {
      "epoch": 0.08700338019889287,
      "grad_norm": 3.9810585975646973,
      "learning_rate": 0.00019017515299605788,
      "loss": 1.9138,
      "step": 222
    },
    {
      "epoch": 0.08739528731690589,
      "grad_norm": 3.9274182319641113,
      "learning_rate": 0.00019008381380628247,
      "loss": 2.0237,
      "step": 223
    },
    {
      "epoch": 0.08778719443491892,
      "grad_norm": 4.107150554656982,
      "learning_rate": 0.00018999207414298067,
      "loss": 1.9284,
      "step": 224
    },
    {
      "epoch": 0.08817910155293196,
      "grad_norm": 4.72376012802124,
      "learning_rate": 0.00018989993441398726,
      "loss": 2.334,
      "step": 225
    },
    {
      "epoch": 0.08857100867094499,
      "grad_norm": 4.973194122314453,
      "learning_rate": 0.00018980739502891546,
      "loss": 2.2631,
      "step": 226
    },
    {
      "epoch": 0.08896291578895801,
      "grad_norm": 5.1790452003479,
      "learning_rate": 0.0001897144563991552,
      "loss": 2.4052,
      "step": 227
    },
    {
      "epoch": 0.08935482290697105,
      "grad_norm": 4.695955276489258,
      "learning_rate": 0.00018962111893787128,
      "loss": 2.1749,
      "step": 228
    },
    {
      "epoch": 0.08974673002498408,
      "grad_norm": 4.675246715545654,
      "learning_rate": 0.00018952738306000151,
      "loss": 1.7668,
      "step": 229
    },
    {
      "epoch": 0.09013863714299711,
      "grad_norm": 4.9759955406188965,
      "learning_rate": 0.00018943324918225494,
      "loss": 2.1394,
      "step": 230
    },
    {
      "epoch": 0.09053054426101013,
      "grad_norm": 5.0605669021606445,
      "learning_rate": 0.0001893387177231099,
      "loss": 2.5426,
      "step": 231
    },
    {
      "epoch": 0.09092245137902318,
      "grad_norm": 5.194053649902344,
      "learning_rate": 0.0001892437891028122,
      "loss": 2.376,
      "step": 232
    },
    {
      "epoch": 0.0913143584970362,
      "grad_norm": 5.404090881347656,
      "learning_rate": 0.0001891484637433733,
      "loss": 2.3723,
      "step": 233
    },
    {
      "epoch": 0.09170626561504923,
      "grad_norm": 5.550242900848389,
      "learning_rate": 0.00018905274206856837,
      "loss": 2.7379,
      "step": 234
    },
    {
      "epoch": 0.09209817273306227,
      "grad_norm": 5.352462291717529,
      "learning_rate": 0.00018895662450393438,
      "loss": 2.4677,
      "step": 235
    },
    {
      "epoch": 0.0924900798510753,
      "grad_norm": 5.692246913909912,
      "learning_rate": 0.00018886011147676833,
      "loss": 2.5777,
      "step": 236
    },
    {
      "epoch": 0.09288198696908832,
      "grad_norm": 5.4188385009765625,
      "learning_rate": 0.00018876320341612522,
      "loss": 2.3903,
      "step": 237
    },
    {
      "epoch": 0.09327389408710136,
      "grad_norm": 5.1750335693359375,
      "learning_rate": 0.00018866590075281624,
      "loss": 2.1595,
      "step": 238
    },
    {
      "epoch": 0.09366580120511439,
      "grad_norm": 5.4043450355529785,
      "learning_rate": 0.00018856820391940674,
      "loss": 2.3534,
      "step": 239
    },
    {
      "epoch": 0.09405770832312742,
      "grad_norm": 5.143820285797119,
      "learning_rate": 0.00018847011335021449,
      "loss": 2.2766,
      "step": 240
    },
    {
      "epoch": 0.09444961544114044,
      "grad_norm": 5.421477317810059,
      "learning_rate": 0.00018837162948130752,
      "loss": 2.0494,
      "step": 241
    },
    {
      "epoch": 0.09484152255915348,
      "grad_norm": 5.6673994064331055,
      "learning_rate": 0.00018827275275050233,
      "loss": 2.1331,
      "step": 242
    },
    {
      "epoch": 0.09523342967716651,
      "grad_norm": 5.99592924118042,
      "learning_rate": 0.00018817348359736203,
      "loss": 2.2909,
      "step": 243
    },
    {
      "epoch": 0.09562533679517954,
      "grad_norm": 6.570507526397705,
      "learning_rate": 0.00018807382246319412,
      "loss": 2.6113,
      "step": 244
    },
    {
      "epoch": 0.09601724391319258,
      "grad_norm": 6.969163417816162,
      "learning_rate": 0.00018797376979104872,
      "loss": 2.4251,
      "step": 245
    },
    {
      "epoch": 0.0964091510312056,
      "grad_norm": 6.864210605621338,
      "learning_rate": 0.00018787332602571662,
      "loss": 2.5026,
      "step": 246
    },
    {
      "epoch": 0.09680105814921863,
      "grad_norm": 6.941864967346191,
      "learning_rate": 0.00018777249161372713,
      "loss": 1.9526,
      "step": 247
    },
    {
      "epoch": 0.09719296526723166,
      "grad_norm": 9.50263500213623,
      "learning_rate": 0.00018767126700334634,
      "loss": 3.1596,
      "step": 248
    },
    {
      "epoch": 0.0975848723852447,
      "grad_norm": 9.44256591796875,
      "learning_rate": 0.0001875696526445749,
      "loss": 2.2232,
      "step": 249
    },
    {
      "epoch": 0.09797677950325773,
      "grad_norm": 14.88141918182373,
      "learning_rate": 0.0001874676489891461,
      "loss": 2.3402,
      "step": 250
    },
    {
      "epoch": 0.09836868662127075,
      "grad_norm": 1.9347553253173828,
      "learning_rate": 0.00018736525649052394,
      "loss": 1.1573,
      "step": 251
    },
    {
      "epoch": 0.0987605937392838,
      "grad_norm": 2.0383098125457764,
      "learning_rate": 0.00018726247560390099,
      "loss": 1.1994,
      "step": 252
    },
    {
      "epoch": 0.09915250085729682,
      "grad_norm": 2.4099888801574707,
      "learning_rate": 0.00018715930678619644,
      "loss": 1.8045,
      "step": 253
    },
    {
      "epoch": 0.09954440797530985,
      "grad_norm": 2.6598567962646484,
      "learning_rate": 0.00018705575049605413,
      "loss": 1.5569,
      "step": 254
    },
    {
      "epoch": 0.09993631509332289,
      "grad_norm": 2.696537971496582,
      "learning_rate": 0.00018695180719384029,
      "loss": 1.5028,
      "step": 255
    },
    {
      "epoch": 0.10032822221133592,
      "grad_norm": 2.9044137001037598,
      "learning_rate": 0.00018684747734164177,
      "loss": 1.7132,
      "step": 256
    },
    {
      "epoch": 0.10072012932934894,
      "grad_norm": 3.2916829586029053,
      "learning_rate": 0.00018674276140326376,
      "loss": 2.2433,
      "step": 257
    },
    {
      "epoch": 0.10111203644736197,
      "grad_norm": 3.2555952072143555,
      "learning_rate": 0.00018663765984422786,
      "loss": 1.9122,
      "step": 258
    },
    {
      "epoch": 0.10150394356537501,
      "grad_norm": 3.4895172119140625,
      "learning_rate": 0.00018653217313177004,
      "loss": 2.3405,
      "step": 259
    },
    {
      "epoch": 0.10189585068338804,
      "grad_norm": 2.9184229373931885,
      "learning_rate": 0.00018642630173483832,
      "loss": 1.941,
      "step": 260
    },
    {
      "epoch": 0.10228775780140106,
      "grad_norm": 3.770773410797119,
      "learning_rate": 0.00018632004612409103,
      "loss": 2.122,
      "step": 261
    },
    {
      "epoch": 0.1026796649194141,
      "grad_norm": 3.84694242477417,
      "learning_rate": 0.00018621340677189453,
      "loss": 2.2583,
      "step": 262
    },
    {
      "epoch": 0.10307157203742713,
      "grad_norm": 3.803971290588379,
      "learning_rate": 0.00018610638415232097,
      "loss": 2.2408,
      "step": 263
    },
    {
      "epoch": 0.10346347915544016,
      "grad_norm": 3.8336987495422363,
      "learning_rate": 0.00018599897874114652,
      "loss": 2.3862,
      "step": 264
    },
    {
      "epoch": 0.1038553862734532,
      "grad_norm": 3.8328397274017334,
      "learning_rate": 0.00018589119101584898,
      "loss": 2.5022,
      "step": 265
    },
    {
      "epoch": 0.10424729339146623,
      "grad_norm": 3.664729118347168,
      "learning_rate": 0.00018578302145560584,
      "loss": 2.2202,
      "step": 266
    },
    {
      "epoch": 0.10463920050947925,
      "grad_norm": 3.5646121501922607,
      "learning_rate": 0.00018567447054129195,
      "loss": 2.1967,
      "step": 267
    },
    {
      "epoch": 0.10503110762749228,
      "grad_norm": 3.505354642868042,
      "learning_rate": 0.00018556553875547754,
      "loss": 2.0468,
      "step": 268
    },
    {
      "epoch": 0.10542301474550532,
      "grad_norm": 4.06969690322876,
      "learning_rate": 0.00018545622658242607,
      "loss": 2.3006,
      "step": 269
    },
    {
      "epoch": 0.10581492186351835,
      "grad_norm": 4.113366603851318,
      "learning_rate": 0.00018534653450809197,
      "loss": 2.2625,
      "step": 270
    },
    {
      "epoch": 0.10620682898153137,
      "grad_norm": 3.9531657695770264,
      "learning_rate": 0.00018523646302011867,
      "loss": 2.2709,
      "step": 271
    },
    {
      "epoch": 0.10659873609954441,
      "grad_norm": 3.4070498943328857,
      "learning_rate": 0.00018512601260783606,
      "loss": 1.8429,
      "step": 272
    },
    {
      "epoch": 0.10699064321755744,
      "grad_norm": 4.637960433959961,
      "learning_rate": 0.00018501518376225887,
      "loss": 2.2168,
      "step": 273
    },
    {
      "epoch": 0.10738255033557047,
      "grad_norm": 4.301741123199463,
      "learning_rate": 0.00018490397697608395,
      "loss": 2.1897,
      "step": 274
    },
    {
      "epoch": 0.1077744574535835,
      "grad_norm": 4.210464954376221,
      "learning_rate": 0.0001847923927436884,
      "loss": 2.265,
      "step": 275
    },
    {
      "epoch": 0.10816636457159653,
      "grad_norm": 4.377344608306885,
      "learning_rate": 0.00018468043156112728,
      "loss": 2.1169,
      "step": 276
    },
    {
      "epoch": 0.10855827168960956,
      "grad_norm": 4.746586799621582,
      "learning_rate": 0.0001845680939261314,
      "loss": 2.2903,
      "step": 277
    },
    {
      "epoch": 0.10895017880762259,
      "grad_norm": 4.96962308883667,
      "learning_rate": 0.00018445538033810515,
      "loss": 2.1894,
      "step": 278
    },
    {
      "epoch": 0.10934208592563563,
      "grad_norm": 4.564748764038086,
      "learning_rate": 0.00018434229129812418,
      "loss": 2.1928,
      "step": 279
    },
    {
      "epoch": 0.10973399304364866,
      "grad_norm": 4.818445205688477,
      "learning_rate": 0.0001842288273089332,
      "loss": 2.3273,
      "step": 280
    },
    {
      "epoch": 0.11012590016166168,
      "grad_norm": 4.9433512687683105,
      "learning_rate": 0.00018411498887494396,
      "loss": 2.2565,
      "step": 281
    },
    {
      "epoch": 0.11051780727967472,
      "grad_norm": 4.769103050231934,
      "learning_rate": 0.00018400077650223263,
      "loss": 2.3077,
      "step": 282
    },
    {
      "epoch": 0.11090971439768775,
      "grad_norm": 4.960122108459473,
      "learning_rate": 0.0001838861906985379,
      "loss": 2.7258,
      "step": 283
    },
    {
      "epoch": 0.11130162151570078,
      "grad_norm": 5.20425271987915,
      "learning_rate": 0.00018377123197325842,
      "loss": 2.45,
      "step": 284
    },
    {
      "epoch": 0.1116935286337138,
      "grad_norm": 4.894191741943359,
      "learning_rate": 0.00018365590083745085,
      "loss": 2.3947,
      "step": 285
    },
    {
      "epoch": 0.11208543575172684,
      "grad_norm": 4.970159530639648,
      "learning_rate": 0.00018354019780382735,
      "loss": 2.3255,
      "step": 286
    },
    {
      "epoch": 0.11247734286973987,
      "grad_norm": 5.984259128570557,
      "learning_rate": 0.0001834241233867533,
      "loss": 2.3562,
      "step": 287
    },
    {
      "epoch": 0.1128692499877529,
      "grad_norm": 4.8289794921875,
      "learning_rate": 0.00018330767810224524,
      "loss": 2.2809,
      "step": 288
    },
    {
      "epoch": 0.11326115710576594,
      "grad_norm": 5.076712608337402,
      "learning_rate": 0.0001831908624679683,
      "loss": 2.2486,
      "step": 289
    },
    {
      "epoch": 0.11365306422377897,
      "grad_norm": 5.225539207458496,
      "learning_rate": 0.0001830736770032341,
      "loss": 2.5909,
      "step": 290
    },
    {
      "epoch": 0.11404497134179199,
      "grad_norm": 5.344139099121094,
      "learning_rate": 0.0001829561222289984,
      "loss": 2.2674,
      "step": 291
    },
    {
      "epoch": 0.11443687845980502,
      "grad_norm": 5.655645847320557,
      "learning_rate": 0.00018283819866785853,
      "loss": 1.732,
      "step": 292
    },
    {
      "epoch": 0.11482878557781806,
      "grad_norm": 5.749541759490967,
      "learning_rate": 0.0001827199068440516,
      "loss": 1.988,
      "step": 293
    },
    {
      "epoch": 0.11522069269583109,
      "grad_norm": 6.434301853179932,
      "learning_rate": 0.00018260124728345162,
      "loss": 2.2762,
      "step": 294
    },
    {
      "epoch": 0.11561259981384411,
      "grad_norm": 6.319149971008301,
      "learning_rate": 0.00018248222051356754,
      "loss": 2.2588,
      "step": 295
    },
    {
      "epoch": 0.11600450693185715,
      "grad_norm": 7.47788143157959,
      "learning_rate": 0.00018236282706354063,
      "loss": 2.4081,
      "step": 296
    },
    {
      "epoch": 0.11639641404987018,
      "grad_norm": 7.883748531341553,
      "learning_rate": 0.00018224306746414238,
      "loss": 2.3595,
      "step": 297
    },
    {
      "epoch": 0.11678832116788321,
      "grad_norm": 9.36921215057373,
      "learning_rate": 0.00018212294224777197,
      "loss": 2.9917,
      "step": 298
    },
    {
      "epoch": 0.11718022828589625,
      "grad_norm": 10.421967506408691,
      "learning_rate": 0.00018200245194845399,
      "loss": 2.6343,
      "step": 299
    },
    {
      "epoch": 0.11757213540390928,
      "grad_norm": 13.730263710021973,
      "learning_rate": 0.00018188159710183594,
      "loss": 3.2354,
      "step": 300
    },
    {
      "epoch": 0.1179640425219223,
      "grad_norm": 1.781528353691101,
      "learning_rate": 0.000181760378245186,
      "loss": 1.0837,
      "step": 301
    },
    {
      "epoch": 0.11835594963993533,
      "grad_norm": 2.0308916568756104,
      "learning_rate": 0.00018163879591739067,
      "loss": 1.1909,
      "step": 302
    },
    {
      "epoch": 0.11874785675794837,
      "grad_norm": 1.9647469520568848,
      "learning_rate": 0.0001815168506589521,
      "loss": 0.9793,
      "step": 303
    },
    {
      "epoch": 0.1191397638759614,
      "grad_norm": 2.2539548873901367,
      "learning_rate": 0.000181394543011986,
      "loss": 1.3216,
      "step": 304
    },
    {
      "epoch": 0.11953167099397442,
      "grad_norm": 2.8216898441314697,
      "learning_rate": 0.00018127187352021907,
      "loss": 1.9377,
      "step": 305
    },
    {
      "epoch": 0.11992357811198746,
      "grad_norm": 3.2739992141723633,
      "learning_rate": 0.0001811488427289866,
      "loss": 1.7594,
      "step": 306
    },
    {
      "epoch": 0.12031548523000049,
      "grad_norm": 3.9117140769958496,
      "learning_rate": 0.00018102545118523007,
      "loss": 2.3575,
      "step": 307
    },
    {
      "epoch": 0.12070739234801352,
      "grad_norm": 2.9577624797821045,
      "learning_rate": 0.00018090169943749476,
      "loss": 2.0479,
      "step": 308
    },
    {
      "epoch": 0.12109929946602656,
      "grad_norm": 3.4503133296966553,
      "learning_rate": 0.00018077758803592718,
      "loss": 2.0082,
      "step": 309
    },
    {
      "epoch": 0.12149120658403958,
      "grad_norm": 3.760124683380127,
      "learning_rate": 0.00018065311753227273,
      "loss": 2.1893,
      "step": 310
    },
    {
      "epoch": 0.12188311370205261,
      "grad_norm": 3.3614439964294434,
      "learning_rate": 0.0001805282884798732,
      "loss": 1.7611,
      "step": 311
    },
    {
      "epoch": 0.12227502082006564,
      "grad_norm": 2.9867730140686035,
      "learning_rate": 0.00018040310143366446,
      "loss": 1.8959,
      "step": 312
    },
    {
      "epoch": 0.12266692793807868,
      "grad_norm": 3.6297998428344727,
      "learning_rate": 0.00018027755695017368,
      "loss": 1.934,
      "step": 313
    },
    {
      "epoch": 0.1230588350560917,
      "grad_norm": 4.229430198669434,
      "learning_rate": 0.00018015165558751717,
      "loss": 2.2022,
      "step": 314
    },
    {
      "epoch": 0.12345074217410473,
      "grad_norm": 3.8777761459350586,
      "learning_rate": 0.00018002539790539773,
      "loss": 2.0718,
      "step": 315
    },
    {
      "epoch": 0.12384264929211777,
      "grad_norm": 3.9300601482391357,
      "learning_rate": 0.00017989878446510215,
      "loss": 2.1771,
      "step": 316
    },
    {
      "epoch": 0.1242345564101308,
      "grad_norm": 3.7910778522491455,
      "learning_rate": 0.00017977181582949888,
      "loss": 2.0851,
      "step": 317
    },
    {
      "epoch": 0.12462646352814383,
      "grad_norm": 3.9136838912963867,
      "learning_rate": 0.0001796444925630353,
      "loss": 2.9307,
      "step": 318
    },
    {
      "epoch": 0.12501837064615687,
      "grad_norm": 3.8064234256744385,
      "learning_rate": 0.00017951681523173542,
      "loss": 2.1792,
      "step": 319
    },
    {
      "epoch": 0.12541027776416988,
      "grad_norm": 4.254158020019531,
      "learning_rate": 0.0001793887844031972,
      "loss": 2.4866,
      "step": 320
    },
    {
      "epoch": 0.12580218488218292,
      "grad_norm": 3.957192897796631,
      "learning_rate": 0.00017926040064659014,
      "loss": 2.538,
      "step": 321
    },
    {
      "epoch": 0.12619409200019596,
      "grad_norm": 4.485204219818115,
      "learning_rate": 0.0001791316645326526,
      "loss": 2.4038,
      "step": 322
    },
    {
      "epoch": 0.12658599911820897,
      "grad_norm": 4.317760944366455,
      "learning_rate": 0.00017900257663368963,
      "loss": 2.2812,
      "step": 323
    },
    {
      "epoch": 0.12697790623622202,
      "grad_norm": 4.180060386657715,
      "learning_rate": 0.0001788731375235698,
      "loss": 2.3711,
      "step": 324
    },
    {
      "epoch": 0.12736981335423506,
      "grad_norm": 4.722744941711426,
      "learning_rate": 0.00017874334777772327,
      "loss": 2.1355,
      "step": 325
    },
    {
      "epoch": 0.12776172047224807,
      "grad_norm": 4.653100490570068,
      "learning_rate": 0.00017861320797313892,
      "loss": 2.3015,
      "step": 326
    },
    {
      "epoch": 0.1281536275902611,
      "grad_norm": 4.426767349243164,
      "learning_rate": 0.0001784827186883618,
      "loss": 2.2799,
      "step": 327
    },
    {
      "epoch": 0.12854553470827415,
      "grad_norm": 4.597682476043701,
      "learning_rate": 0.00017835188050349064,
      "loss": 2.1414,
      "step": 328
    },
    {
      "epoch": 0.12893744182628716,
      "grad_norm": 4.900806903839111,
      "learning_rate": 0.00017822069400017516,
      "loss": 2.3065,
      "step": 329
    },
    {
      "epoch": 0.1293293489443002,
      "grad_norm": 5.382739543914795,
      "learning_rate": 0.00017808915976161362,
      "loss": 2.4255,
      "step": 330
    },
    {
      "epoch": 0.12972125606231324,
      "grad_norm": 5.385190010070801,
      "learning_rate": 0.00017795727837255015,
      "loss": 2.6805,
      "step": 331
    },
    {
      "epoch": 0.13011316318032626,
      "grad_norm": 5.173645973205566,
      "learning_rate": 0.00017782505041927216,
      "loss": 2.5031,
      "step": 332
    },
    {
      "epoch": 0.1305050702983393,
      "grad_norm": 4.92276668548584,
      "learning_rate": 0.00017769247648960774,
      "loss": 2.5355,
      "step": 333
    },
    {
      "epoch": 0.1308969774163523,
      "grad_norm": 4.867099285125732,
      "learning_rate": 0.00017755955717292296,
      "loss": 2.535,
      "step": 334
    },
    {
      "epoch": 0.13128888453436535,
      "grad_norm": 4.533221244812012,
      "learning_rate": 0.00017742629306011944,
      "loss": 2.3636,
      "step": 335
    },
    {
      "epoch": 0.1316807916523784,
      "grad_norm": 5.3090667724609375,
      "learning_rate": 0.00017729268474363154,
      "loss": 2.4942,
      "step": 336
    },
    {
      "epoch": 0.1320726987703914,
      "grad_norm": 4.582738399505615,
      "learning_rate": 0.0001771587328174239,
      "loss": 2.371,
      "step": 337
    },
    {
      "epoch": 0.13246460588840445,
      "grad_norm": 5.376871585845947,
      "learning_rate": 0.0001770244378769885,
      "loss": 2.482,
      "step": 338
    },
    {
      "epoch": 0.1328565130064175,
      "grad_norm": 5.913669586181641,
      "learning_rate": 0.0001768898005193425,
      "loss": 2.4587,
      "step": 339
    },
    {
      "epoch": 0.1332484201244305,
      "grad_norm": 5.702977657318115,
      "learning_rate": 0.000176754821343025,
      "loss": 2.4858,
      "step": 340
    },
    {
      "epoch": 0.13364032724244354,
      "grad_norm": 5.84987211227417,
      "learning_rate": 0.0001766195009480949,
      "loss": 1.9115,
      "step": 341
    },
    {
      "epoch": 0.13403223436045658,
      "grad_norm": 5.594895362854004,
      "learning_rate": 0.0001764838399361279,
      "loss": 2.4662,
      "step": 342
    },
    {
      "epoch": 0.1344241414784696,
      "grad_norm": 5.69754695892334,
      "learning_rate": 0.00017634783891021393,
      "loss": 2.0795,
      "step": 343
    },
    {
      "epoch": 0.13481604859648263,
      "grad_norm": 6.249038219451904,
      "learning_rate": 0.00017621149847495458,
      "loss": 1.9406,
      "step": 344
    },
    {
      "epoch": 0.13520795571449568,
      "grad_norm": 6.313895225524902,
      "learning_rate": 0.00017607481923646016,
      "loss": 2.1859,
      "step": 345
    },
    {
      "epoch": 0.1355998628325087,
      "grad_norm": 7.00896692276001,
      "learning_rate": 0.0001759378018023473,
      "loss": 2.5898,
      "step": 346
    },
    {
      "epoch": 0.13599176995052173,
      "grad_norm": 7.788276195526123,
      "learning_rate": 0.00017580044678173592,
      "loss": 2.0872,
      "step": 347
    },
    {
      "epoch": 0.13638367706853477,
      "grad_norm": 8.23760986328125,
      "learning_rate": 0.00017566275478524693,
      "loss": 2.767,
      "step": 348
    },
    {
      "epoch": 0.13677558418654778,
      "grad_norm": 9.444133758544922,
      "learning_rate": 0.0001755247264249991,
      "loss": 2.311,
      "step": 349
    },
    {
      "epoch": 0.13716749130456082,
      "grad_norm": 10.550444602966309,
      "learning_rate": 0.0001753863623146066,
      "loss": 2.186,
      "step": 350
    },
    {
      "epoch": 0.13755939842257384,
      "grad_norm": 2.2415695190429688,
      "learning_rate": 0.00017524766306917618,
      "loss": 1.2191,
      "step": 351
    },
    {
      "epoch": 0.13795130554058688,
      "grad_norm": 2.398531913757324,
      "learning_rate": 0.0001751086293053045,
      "loss": 1.5496,
      "step": 352
    },
    {
      "epoch": 0.13834321265859992,
      "grad_norm": 2.9486899375915527,
      "learning_rate": 0.0001749692616410753,
      "loss": 1.6313,
      "step": 353
    },
    {
      "epoch": 0.13873511977661293,
      "grad_norm": 2.7810165882110596,
      "learning_rate": 0.00017482956069605668,
      "loss": 1.4301,
      "step": 354
    },
    {
      "epoch": 0.13912702689462597,
      "grad_norm": 3.2558860778808594,
      "learning_rate": 0.00017468952709129846,
      "loss": 2.3144,
      "step": 355
    },
    {
      "epoch": 0.139518934012639,
      "grad_norm": 3.2652111053466797,
      "learning_rate": 0.00017454916144932922,
      "loss": 2.0155,
      "step": 356
    },
    {
      "epoch": 0.13991084113065202,
      "grad_norm": 3.123067617416382,
      "learning_rate": 0.0001744084643941536,
      "loss": 1.7605,
      "step": 357
    },
    {
      "epoch": 0.14030274824866507,
      "grad_norm": 3.028930187225342,
      "learning_rate": 0.00017426743655124974,
      "loss": 1.8005,
      "step": 358
    },
    {
      "epoch": 0.1406946553666781,
      "grad_norm": 2.9984774589538574,
      "learning_rate": 0.0001741260785475661,
      "loss": 1.8025,
      "step": 359
    },
    {
      "epoch": 0.14108656248469112,
      "grad_norm": 3.357543706893921,
      "learning_rate": 0.00017398439101151905,
      "loss": 2.1195,
      "step": 360
    },
    {
      "epoch": 0.14147846960270416,
      "grad_norm": 3.2198123931884766,
      "learning_rate": 0.00017384237457298987,
      "loss": 2.3509,
      "step": 361
    },
    {
      "epoch": 0.1418703767207172,
      "grad_norm": 2.8786911964416504,
      "learning_rate": 0.00017370002986332193,
      "loss": 1.9398,
      "step": 362
    },
    {
      "epoch": 0.1422622838387302,
      "grad_norm": 3.4074885845184326,
      "learning_rate": 0.00017355735751531807,
      "loss": 2.1364,
      "step": 363
    },
    {
      "epoch": 0.14265419095674325,
      "grad_norm": 3.5557861328125,
      "learning_rate": 0.00017341435816323756,
      "loss": 1.7583,
      "step": 364
    },
    {
      "epoch": 0.1430460980747563,
      "grad_norm": 4.004567623138428,
      "learning_rate": 0.00017327103244279348,
      "loss": 2.1772,
      "step": 365
    },
    {
      "epoch": 0.1434380051927693,
      "grad_norm": 4.297161102294922,
      "learning_rate": 0.00017312738099114973,
      "loss": 2.2777,
      "step": 366
    },
    {
      "epoch": 0.14382991231078235,
      "grad_norm": 3.917724609375,
      "learning_rate": 0.00017298340444691835,
      "loss": 2.061,
      "step": 367
    },
    {
      "epoch": 0.1442218194287954,
      "grad_norm": 4.060068130493164,
      "learning_rate": 0.00017283910345015647,
      "loss": 2.2657,
      "step": 368
    },
    {
      "epoch": 0.1446137265468084,
      "grad_norm": 4.18174409866333,
      "learning_rate": 0.0001726944786423637,
      "loss": 2.4496,
      "step": 369
    },
    {
      "epoch": 0.14500563366482144,
      "grad_norm": 4.178645610809326,
      "learning_rate": 0.00017254953066647913,
      "loss": 2.257,
      "step": 370
    },
    {
      "epoch": 0.14539754078283446,
      "grad_norm": 3.9574522972106934,
      "learning_rate": 0.00017240426016687863,
      "loss": 2.1858,
      "step": 371
    },
    {
      "epoch": 0.1457894479008475,
      "grad_norm": 4.0800251960754395,
      "learning_rate": 0.00017225866778937165,
      "loss": 2.5598,
      "step": 372
    },
    {
      "epoch": 0.14618135501886054,
      "grad_norm": 3.9451498985290527,
      "learning_rate": 0.00017211275418119876,
      "loss": 2.1192,
      "step": 373
    },
    {
      "epoch": 0.14657326213687355,
      "grad_norm": 4.095687389373779,
      "learning_rate": 0.0001719665199910285,
      "loss": 2.0935,
      "step": 374
    },
    {
      "epoch": 0.1469651692548866,
      "grad_norm": 4.13465690612793,
      "learning_rate": 0.00017181996586895454,
      "loss": 1.8691,
      "step": 375
    },
    {
      "epoch": 0.1469651692548866,
      "eval_loss": 2.1597940921783447,
      "eval_runtime": 37.8261,
      "eval_samples_per_second": 56.813,
      "eval_steps_per_second": 28.42,
      "step": 375
    },
    {
      "epoch": 0.14735707637289963,
      "grad_norm": 4.48436164855957,
      "learning_rate": 0.00017167309246649297,
      "loss": 2.2666,
      "step": 376
    },
    {
      "epoch": 0.14774898349091264,
      "grad_norm": 4.76546049118042,
      "learning_rate": 0.0001715259004365791,
      "loss": 2.1245,
      "step": 377
    },
    {
      "epoch": 0.14814089060892568,
      "grad_norm": 4.740818023681641,
      "learning_rate": 0.00017137839043356484,
      "loss": 2.5205,
      "step": 378
    },
    {
      "epoch": 0.14853279772693873,
      "grad_norm": 4.898871898651123,
      "learning_rate": 0.00017123056311321562,
      "loss": 2.2961,
      "step": 379
    },
    {
      "epoch": 0.14892470484495174,
      "grad_norm": 4.882875919342041,
      "learning_rate": 0.0001710824191327075,
      "loss": 2.3712,
      "step": 380
    },
    {
      "epoch": 0.14931661196296478,
      "grad_norm": 4.777733325958252,
      "learning_rate": 0.00017093395915062428,
      "loss": 2.2939,
      "step": 381
    },
    {
      "epoch": 0.14970851908097782,
      "grad_norm": 5.037533283233643,
      "learning_rate": 0.00017078518382695465,
      "loss": 2.5068,
      "step": 382
    },
    {
      "epoch": 0.15010042619899083,
      "grad_norm": 5.1660590171813965,
      "learning_rate": 0.00017063609382308908,
      "loss": 2.4199,
      "step": 383
    },
    {
      "epoch": 0.15049233331700387,
      "grad_norm": 5.036225318908691,
      "learning_rate": 0.00017048668980181698,
      "loss": 2.0712,
      "step": 384
    },
    {
      "epoch": 0.15088424043501691,
      "grad_norm": 5.3519134521484375,
      "learning_rate": 0.00017033697242732377,
      "loss": 2.2018,
      "step": 385
    },
    {
      "epoch": 0.15127614755302993,
      "grad_norm": 5.288197040557861,
      "learning_rate": 0.0001701869423651879,
      "loss": 2.5597,
      "step": 386
    },
    {
      "epoch": 0.15166805467104297,
      "grad_norm": 5.077462673187256,
      "learning_rate": 0.00017003660028237793,
      "loss": 2.2897,
      "step": 387
    },
    {
      "epoch": 0.15205996178905598,
      "grad_norm": 5.021242141723633,
      "learning_rate": 0.00016988594684724947,
      "loss": 2.4151,
      "step": 388
    },
    {
      "epoch": 0.15245186890706902,
      "grad_norm": 5.304116725921631,
      "learning_rate": 0.00016973498272954222,
      "loss": 2.334,
      "step": 389
    },
    {
      "epoch": 0.15284377602508206,
      "grad_norm": 5.193856239318848,
      "learning_rate": 0.00016958370860037717,
      "loss": 2.2287,
      "step": 390
    },
    {
      "epoch": 0.15323568314309507,
      "grad_norm": 5.028045654296875,
      "learning_rate": 0.00016943212513225345,
      "loss": 1.9385,
      "step": 391
    },
    {
      "epoch": 0.15362759026110812,
      "grad_norm": 5.609121322631836,
      "learning_rate": 0.00016928023299904533,
      "loss": 2.0316,
      "step": 392
    },
    {
      "epoch": 0.15401949737912116,
      "grad_norm": 4.961055755615234,
      "learning_rate": 0.0001691280328759992,
      "loss": 1.8508,
      "step": 393
    },
    {
      "epoch": 0.15441140449713417,
      "grad_norm": 5.5503740310668945,
      "learning_rate": 0.00016897552543973084,
      "loss": 2.2215,
      "step": 394
    },
    {
      "epoch": 0.1548033116151472,
      "grad_norm": 6.502624988555908,
      "learning_rate": 0.00016882271136822206,
      "loss": 2.3217,
      "step": 395
    },
    {
      "epoch": 0.15519521873316025,
      "grad_norm": 5.8435163497924805,
      "learning_rate": 0.0001686695913408179,
      "loss": 2.0812,
      "step": 396
    },
    {
      "epoch": 0.15558712585117326,
      "grad_norm": 7.10982608795166,
      "learning_rate": 0.0001685161660382235,
      "loss": 2.0555,
      "step": 397
    },
    {
      "epoch": 0.1559790329691863,
      "grad_norm": 8.9778470993042,
      "learning_rate": 0.00016836243614250113,
      "loss": 2.5491,
      "step": 398
    },
    {
      "epoch": 0.15637094008719934,
      "grad_norm": 9.061491966247559,
      "learning_rate": 0.00016820840233706719,
      "loss": 2.421,
      "step": 399
    },
    {
      "epoch": 0.15676284720521236,
      "grad_norm": 16.60405731201172,
      "learning_rate": 0.0001680540653066891,
      "loss": 2.8275,
      "step": 400
    },
    {
      "epoch": 0.1571547543232254,
      "grad_norm": 1.9591330289840698,
      "learning_rate": 0.00016789942573748232,
      "loss": 0.6914,
      "step": 401
    },
    {
      "epoch": 0.15754666144123844,
      "grad_norm": 2.1030266284942627,
      "learning_rate": 0.0001677444843169072,
      "loss": 1.2701,
      "step": 402
    },
    {
      "epoch": 0.15793856855925145,
      "grad_norm": 2.0107295513153076,
      "learning_rate": 0.00016758924173376603,
      "loss": 1.2645,
      "step": 403
    },
    {
      "epoch": 0.1583304756772645,
      "grad_norm": 2.576596736907959,
      "learning_rate": 0.0001674336986781999,
      "loss": 2.002,
      "step": 404
    },
    {
      "epoch": 0.1587223827952775,
      "grad_norm": 2.4120073318481445,
      "learning_rate": 0.00016727785584168581,
      "loss": 1.5059,
      "step": 405
    },
    {
      "epoch": 0.15911428991329055,
      "grad_norm": 2.9901487827301025,
      "learning_rate": 0.0001671217139170333,
      "loss": 1.8947,
      "step": 406
    },
    {
      "epoch": 0.1595061970313036,
      "grad_norm": 2.6051316261291504,
      "learning_rate": 0.00016696527359838154,
      "loss": 1.4584,
      "step": 407
    },
    {
      "epoch": 0.1598981041493166,
      "grad_norm": 3.1282503604888916,
      "learning_rate": 0.00016680853558119632,
      "loss": 2.042,
      "step": 408
    },
    {
      "epoch": 0.16029001126732964,
      "grad_norm": 3.4535140991210938,
      "learning_rate": 0.0001666515005622668,
      "loss": 2.4349,
      "step": 409
    },
    {
      "epoch": 0.16068191838534268,
      "grad_norm": 3.724137306213379,
      "learning_rate": 0.0001664941692397025,
      "loss": 2.685,
      "step": 410
    },
    {
      "epoch": 0.1610738255033557,
      "grad_norm": 3.480076313018799,
      "learning_rate": 0.00016633654231293013,
      "loss": 2.2549,
      "step": 411
    },
    {
      "epoch": 0.16146573262136873,
      "grad_norm": 3.014763832092285,
      "learning_rate": 0.00016617862048269065,
      "loss": 1.9814,
      "step": 412
    },
    {
      "epoch": 0.16185763973938178,
      "grad_norm": 3.1888105869293213,
      "learning_rate": 0.00016602040445103588,
      "loss": 2.0021,
      "step": 413
    },
    {
      "epoch": 0.1622495468573948,
      "grad_norm": 3.272312641143799,
      "learning_rate": 0.00016586189492132566,
      "loss": 2.6034,
      "step": 414
    },
    {
      "epoch": 0.16264145397540783,
      "grad_norm": 3.5542216300964355,
      "learning_rate": 0.00016570309259822453,
      "loss": 2.2921,
      "step": 415
    },
    {
      "epoch": 0.16303336109342087,
      "grad_norm": 3.416416645050049,
      "learning_rate": 0.0001655439981876987,
      "loss": 2.1046,
      "step": 416
    },
    {
      "epoch": 0.16342526821143388,
      "grad_norm": 3.698093891143799,
      "learning_rate": 0.00016538461239701277,
      "loss": 2.2895,
      "step": 417
    },
    {
      "epoch": 0.16381717532944692,
      "grad_norm": 3.490540027618408,
      "learning_rate": 0.00016522493593472683,
      "loss": 2.3106,
      "step": 418
    },
    {
      "epoch": 0.16420908244745996,
      "grad_norm": 4.165564060211182,
      "learning_rate": 0.0001650649695106931,
      "loss": 2.0274,
      "step": 419
    },
    {
      "epoch": 0.16460098956547298,
      "grad_norm": 4.105908393859863,
      "learning_rate": 0.00016490471383605288,
      "loss": 2.2011,
      "step": 420
    },
    {
      "epoch": 0.16499289668348602,
      "grad_norm": 3.973095178604126,
      "learning_rate": 0.00016474416962323325,
      "loss": 2.052,
      "step": 421
    },
    {
      "epoch": 0.16538480380149906,
      "grad_norm": 4.401207447052002,
      "learning_rate": 0.00016458333758594414,
      "loss": 2.4918,
      "step": 422
    },
    {
      "epoch": 0.16577671091951207,
      "grad_norm": 5.076011657714844,
      "learning_rate": 0.00016442221843917496,
      "loss": 2.1481,
      "step": 423
    },
    {
      "epoch": 0.1661686180375251,
      "grad_norm": 4.560831546783447,
      "learning_rate": 0.00016426081289919143,
      "loss": 2.3135,
      "step": 424
    },
    {
      "epoch": 0.16656052515553812,
      "grad_norm": 4.631618022918701,
      "learning_rate": 0.0001640991216835326,
      "loss": 2.3501,
      "step": 425
    },
    {
      "epoch": 0.16695243227355117,
      "grad_norm": 4.037489414215088,
      "learning_rate": 0.00016393714551100734,
      "loss": 1.971,
      "step": 426
    },
    {
      "epoch": 0.1673443393915642,
      "grad_norm": 4.578770637512207,
      "learning_rate": 0.0001637748851016914,
      "loss": 2.4199,
      "step": 427
    },
    {
      "epoch": 0.16773624650957722,
      "grad_norm": 4.557089328765869,
      "learning_rate": 0.00016361234117692413,
      "loss": 2.1958,
      "step": 428
    },
    {
      "epoch": 0.16812815362759026,
      "grad_norm": 4.220493316650391,
      "learning_rate": 0.00016344951445930526,
      "loss": 2.1845,
      "step": 429
    },
    {
      "epoch": 0.1685200607456033,
      "grad_norm": 4.7399210929870605,
      "learning_rate": 0.0001632864056726917,
      "loss": 2.3616,
      "step": 430
    },
    {
      "epoch": 0.1689119678636163,
      "grad_norm": 4.861191272735596,
      "learning_rate": 0.00016312301554219426,
      "loss": 2.0798,
      "step": 431
    },
    {
      "epoch": 0.16930387498162935,
      "grad_norm": 4.483139514923096,
      "learning_rate": 0.00016295934479417453,
      "loss": 2.366,
      "step": 432
    },
    {
      "epoch": 0.1696957820996424,
      "grad_norm": 4.41169548034668,
      "learning_rate": 0.00016279539415624164,
      "loss": 2.3118,
      "step": 433
    },
    {
      "epoch": 0.1700876892176554,
      "grad_norm": 4.68226957321167,
      "learning_rate": 0.0001626311643572489,
      "loss": 2.2067,
      "step": 434
    },
    {
      "epoch": 0.17047959633566845,
      "grad_norm": 4.946905612945557,
      "learning_rate": 0.00016246665612729074,
      "loss": 2.515,
      "step": 435
    },
    {
      "epoch": 0.1708715034536815,
      "grad_norm": 5.144153594970703,
      "learning_rate": 0.00016230187019769928,
      "loss": 2.7226,
      "step": 436
    },
    {
      "epoch": 0.1712634105716945,
      "grad_norm": 4.604719638824463,
      "learning_rate": 0.00016213680730104124,
      "loss": 2.222,
      "step": 437
    },
    {
      "epoch": 0.17165531768970754,
      "grad_norm": 4.556640625,
      "learning_rate": 0.0001619714681711146,
      "loss": 1.8309,
      "step": 438
    },
    {
      "epoch": 0.17204722480772058,
      "grad_norm": 5.362233638763428,
      "learning_rate": 0.00016180585354294536,
      "loss": 2.2543,
      "step": 439
    },
    {
      "epoch": 0.1724391319257336,
      "grad_norm": 5.174863338470459,
      "learning_rate": 0.00016163996415278424,
      "loss": 2.113,
      "step": 440
    },
    {
      "epoch": 0.17283103904374664,
      "grad_norm": 5.174488067626953,
      "learning_rate": 0.00016147380073810346,
      "loss": 2.3167,
      "step": 441
    },
    {
      "epoch": 0.17322294616175965,
      "grad_norm": 5.406106948852539,
      "learning_rate": 0.0001613073640375934,
      "loss": 1.9535,
      "step": 442
    },
    {
      "epoch": 0.1736148532797727,
      "grad_norm": 5.434831142425537,
      "learning_rate": 0.00016114065479115946,
      "loss": 2.4388,
      "step": 443
    },
    {
      "epoch": 0.17400676039778573,
      "grad_norm": 5.338566780090332,
      "learning_rate": 0.00016097367373991842,
      "loss": 2.1174,
      "step": 444
    },
    {
      "epoch": 0.17439866751579874,
      "grad_norm": 5.678027153015137,
      "learning_rate": 0.00016080642162619565,
      "loss": 2.3497,
      "step": 445
    },
    {
      "epoch": 0.17479057463381178,
      "grad_norm": 6.345289707183838,
      "learning_rate": 0.0001606388991935214,
      "loss": 2.1913,
      "step": 446
    },
    {
      "epoch": 0.17518248175182483,
      "grad_norm": 6.457273483276367,
      "learning_rate": 0.0001604711071866277,
      "loss": 2.357,
      "step": 447
    },
    {
      "epoch": 0.17557438886983784,
      "grad_norm": 7.04536771774292,
      "learning_rate": 0.00016030304635144494,
      "loss": 2.0677,
      "step": 448
    },
    {
      "epoch": 0.17596629598785088,
      "grad_norm": 7.80436372756958,
      "learning_rate": 0.00016013471743509862,
      "loss": 2.0042,
      "step": 449
    },
    {
      "epoch": 0.17635820310586392,
      "grad_norm": 10.383035659790039,
      "learning_rate": 0.00015996612118590603,
      "loss": 2.2413,
      "step": 450
    },
    {
      "epoch": 0.17675011022387693,
      "grad_norm": 1.8237245082855225,
      "learning_rate": 0.00015979725835337294,
      "loss": 1.4466,
      "step": 451
    },
    {
      "epoch": 0.17714201734188997,
      "grad_norm": 1.836155891418457,
      "learning_rate": 0.00015962812968819016,
      "loss": 1.1336,
      "step": 452
    },
    {
      "epoch": 0.17753392445990301,
      "grad_norm": 2.188556432723999,
      "learning_rate": 0.0001594587359422303,
      "loss": 1.0752,
      "step": 453
    },
    {
      "epoch": 0.17792583157791603,
      "grad_norm": 1.9821218252182007,
      "learning_rate": 0.0001592890778685444,
      "loss": 1.1553,
      "step": 454
    },
    {
      "epoch": 0.17831773869592907,
      "grad_norm": 2.5855178833007812,
      "learning_rate": 0.00015911915622135862,
      "loss": 1.8215,
      "step": 455
    },
    {
      "epoch": 0.1787096458139421,
      "grad_norm": 3.0647382736206055,
      "learning_rate": 0.00015894897175607086,
      "loss": 1.8789,
      "step": 456
    },
    {
      "epoch": 0.17910155293195512,
      "grad_norm": 2.7413856983184814,
      "learning_rate": 0.00015877852522924732,
      "loss": 1.8168,
      "step": 457
    },
    {
      "epoch": 0.17949346004996816,
      "grad_norm": 3.7904868125915527,
      "learning_rate": 0.00015860781739861928,
      "loss": 2.3411,
      "step": 458
    },
    {
      "epoch": 0.17988536716798117,
      "grad_norm": 3.5881943702697754,
      "learning_rate": 0.00015843684902307962,
      "loss": 2.1839,
      "step": 459
    },
    {
      "epoch": 0.18027727428599422,
      "grad_norm": 3.0449819564819336,
      "learning_rate": 0.00015826562086267956,
      "loss": 1.4773,
      "step": 460
    },
    {
      "epoch": 0.18066918140400726,
      "grad_norm": 4.0939717292785645,
      "learning_rate": 0.00015809413367862512,
      "loss": 2.4854,
      "step": 461
    },
    {
      "epoch": 0.18106108852202027,
      "grad_norm": 3.642366409301758,
      "learning_rate": 0.00015792238823327388,
      "loss": 2.2104,
      "step": 462
    },
    {
      "epoch": 0.1814529956400333,
      "grad_norm": 3.401036500930786,
      "learning_rate": 0.00015775038529013152,
      "loss": 2.2412,
      "step": 463
    },
    {
      "epoch": 0.18184490275804635,
      "grad_norm": 3.307555913925171,
      "learning_rate": 0.0001575781256138485,
      "loss": 2.0512,
      "step": 464
    },
    {
      "epoch": 0.18223680987605936,
      "grad_norm": 3.5533156394958496,
      "learning_rate": 0.00015740560997021648,
      "loss": 2.0149,
      "step": 465
    },
    {
      "epoch": 0.1826287169940724,
      "grad_norm": 3.1206984519958496,
      "learning_rate": 0.00015723283912616513,
      "loss": 1.8049,
      "step": 466
    },
    {
      "epoch": 0.18302062411208544,
      "grad_norm": 3.612071990966797,
      "learning_rate": 0.00015705981384975866,
      "loss": 2.1186,
      "step": 467
    },
    {
      "epoch": 0.18341253123009846,
      "grad_norm": 3.69101619720459,
      "learning_rate": 0.0001568865349101923,
      "loss": 2.2134,
      "step": 468
    },
    {
      "epoch": 0.1838044383481115,
      "grad_norm": 4.195282936096191,
      "learning_rate": 0.00015671300307778898,
      "loss": 2.1176,
      "step": 469
    },
    {
      "epoch": 0.18419634546612454,
      "grad_norm": 3.7252259254455566,
      "learning_rate": 0.00015653921912399589,
      "loss": 1.9158,
      "step": 470
    },
    {
      "epoch": 0.18458825258413755,
      "grad_norm": 3.7424659729003906,
      "learning_rate": 0.00015636518382138107,
      "loss": 2.1337,
      "step": 471
    },
    {
      "epoch": 0.1849801597021506,
      "grad_norm": 4.340978145599365,
      "learning_rate": 0.0001561908979436299,
      "loss": 2.5192,
      "step": 472
    },
    {
      "epoch": 0.18537206682016363,
      "grad_norm": 3.913496494293213,
      "learning_rate": 0.00015601636226554168,
      "loss": 1.8156,
      "step": 473
    },
    {
      "epoch": 0.18576397393817665,
      "grad_norm": 4.26788854598999,
      "learning_rate": 0.00015584157756302634,
      "loss": 2.0489,
      "step": 474
    },
    {
      "epoch": 0.1861558810561897,
      "grad_norm": 4.134047985076904,
      "learning_rate": 0.0001556665446131007,
      "loss": 2.1049,
      "step": 475
    },
    {
      "epoch": 0.18654778817420273,
      "grad_norm": 4.316871166229248,
      "learning_rate": 0.00015549126419388536,
      "loss": 2.1545,
      "step": 476
    },
    {
      "epoch": 0.18693969529221574,
      "grad_norm": 3.9756827354431152,
      "learning_rate": 0.0001553157370846009,
      "loss": 2.144,
      "step": 477
    },
    {
      "epoch": 0.18733160241022878,
      "grad_norm": 4.604708671569824,
      "learning_rate": 0.00015513996406556465,
      "loss": 1.931,
      "step": 478
    },
    {
      "epoch": 0.1877235095282418,
      "grad_norm": 4.540436744689941,
      "learning_rate": 0.00015496394591818716,
      "loss": 1.9981,
      "step": 479
    },
    {
      "epoch": 0.18811541664625483,
      "grad_norm": 4.5301194190979,
      "learning_rate": 0.0001547876834249687,
      "loss": 2.1515,
      "step": 480
    },
    {
      "epoch": 0.18850732376426788,
      "grad_norm": 4.311919212341309,
      "learning_rate": 0.00015461117736949577,
      "loss": 1.879,
      "step": 481
    },
    {
      "epoch": 0.1888992308822809,
      "grad_norm": 4.7697319984436035,
      "learning_rate": 0.00015443442853643762,
      "loss": 2.4863,
      "step": 482
    },
    {
      "epoch": 0.18929113800029393,
      "grad_norm": 4.516111850738525,
      "learning_rate": 0.00015425743771154294,
      "loss": 2.2706,
      "step": 483
    },
    {
      "epoch": 0.18968304511830697,
      "grad_norm": 5.043210029602051,
      "learning_rate": 0.00015408020568163602,
      "loss": 2.4226,
      "step": 484
    },
    {
      "epoch": 0.19007495223631998,
      "grad_norm": 4.873130798339844,
      "learning_rate": 0.00015390273323461352,
      "loss": 2.2661,
      "step": 485
    },
    {
      "epoch": 0.19046685935433302,
      "grad_norm": 5.221062660217285,
      "learning_rate": 0.0001537250211594409,
      "loss": 2.1652,
      "step": 486
    },
    {
      "epoch": 0.19085876647234606,
      "grad_norm": 4.338252544403076,
      "learning_rate": 0.0001535470702461489,
      "loss": 1.8144,
      "step": 487
    },
    {
      "epoch": 0.19125067359035908,
      "grad_norm": 4.886384963989258,
      "learning_rate": 0.00015336888128583,
      "loss": 2.5762,
      "step": 488
    },
    {
      "epoch": 0.19164258070837212,
      "grad_norm": 4.927480697631836,
      "learning_rate": 0.000153190455070635,
      "loss": 2.4133,
      "step": 489
    },
    {
      "epoch": 0.19203448782638516,
      "grad_norm": 5.423128604888916,
      "learning_rate": 0.00015301179239376938,
      "loss": 2.2332,
      "step": 490
    },
    {
      "epoch": 0.19242639494439817,
      "grad_norm": 4.800963878631592,
      "learning_rate": 0.00015283289404948976,
      "loss": 1.8301,
      "step": 491
    },
    {
      "epoch": 0.1928183020624112,
      "grad_norm": 5.158536911010742,
      "learning_rate": 0.0001526537608331006,
      "loss": 1.8981,
      "step": 492
    },
    {
      "epoch": 0.19321020918042425,
      "grad_norm": 6.057976245880127,
      "learning_rate": 0.00015247439354095041,
      "loss": 2.3559,
      "step": 493
    },
    {
      "epoch": 0.19360211629843727,
      "grad_norm": 5.408287048339844,
      "learning_rate": 0.00015229479297042823,
      "loss": 1.8217,
      "step": 494
    },
    {
      "epoch": 0.1939940234164503,
      "grad_norm": 6.81193733215332,
      "learning_rate": 0.00015211495991996027,
      "loss": 2.2438,
      "step": 495
    },
    {
      "epoch": 0.19438593053446332,
      "grad_norm": 6.9760422706604,
      "learning_rate": 0.0001519348951890062,
      "loss": 2.2358,
      "step": 496
    },
    {
      "epoch": 0.19477783765247636,
      "grad_norm": 7.879918098449707,
      "learning_rate": 0.0001517545995780556,
      "loss": 1.9936,
      "step": 497
    },
    {
      "epoch": 0.1951697447704894,
      "grad_norm": 8.005513191223145,
      "learning_rate": 0.00015157407388862452,
      "loss": 1.8717,
      "step": 498
    },
    {
      "epoch": 0.1955616518885024,
      "grad_norm": 8.993185997009277,
      "learning_rate": 0.00015139331892325179,
      "loss": 2.4458,
      "step": 499
    },
    {
      "epoch": 0.19595355900651545,
      "grad_norm": 10.837003707885742,
      "learning_rate": 0.0001512123354854955,
      "loss": 2.5082,
      "step": 500
    },
    {
      "epoch": 0.1963454661245285,
      "grad_norm": 2.084827423095703,
      "learning_rate": 0.0001510311243799295,
      "loss": 1.9394,
      "step": 501
    },
    {
      "epoch": 0.1967373732425415,
      "grad_norm": 1.8055182695388794,
      "learning_rate": 0.00015084968641213958,
      "loss": 1.6824,
      "step": 502
    },
    {
      "epoch": 0.19712928036055455,
      "grad_norm": 2.165750026702881,
      "learning_rate": 0.00015066802238872023,
      "loss": 1.458,
      "step": 503
    },
    {
      "epoch": 0.1975211874785676,
      "grad_norm": 2.118333339691162,
      "learning_rate": 0.0001504861331172709,
      "loss": 1.3127,
      "step": 504
    },
    {
      "epoch": 0.1979130945965806,
      "grad_norm": 2.1731691360473633,
      "learning_rate": 0.0001503040194063922,
      "loss": 1.2644,
      "step": 505
    },
    {
      "epoch": 0.19830500171459364,
      "grad_norm": 2.6870672702789307,
      "learning_rate": 0.00015012168206568268,
      "loss": 1.3931,
      "step": 506
    },
    {
      "epoch": 0.19869690883260668,
      "grad_norm": 2.8041768074035645,
      "learning_rate": 0.00014993912190573505,
      "loss": 1.9654,
      "step": 507
    },
    {
      "epoch": 0.1990888159506197,
      "grad_norm": 3.3024322986602783,
      "learning_rate": 0.00014975633973813242,
      "loss": 2.1867,
      "step": 508
    },
    {
      "epoch": 0.19948072306863274,
      "grad_norm": 2.9356207847595215,
      "learning_rate": 0.00014957333637544503,
      "loss": 1.8808,
      "step": 509
    },
    {
      "epoch": 0.19987263018664578,
      "grad_norm": 3.1021625995635986,
      "learning_rate": 0.00014939011263122634,
      "loss": 1.732,
      "step": 510
    },
    {
      "epoch": 0.2002645373046588,
      "grad_norm": 3.367109537124634,
      "learning_rate": 0.0001492066693200096,
      "loss": 2.0601,
      "step": 511
    },
    {
      "epoch": 0.20065644442267183,
      "grad_norm": 3.0847725868225098,
      "learning_rate": 0.00014902300725730413,
      "loss": 1.9873,
      "step": 512
    },
    {
      "epoch": 0.20104835154068484,
      "grad_norm": 3.4438552856445312,
      "learning_rate": 0.00014883912725959167,
      "loss": 1.8075,
      "step": 513
    },
    {
      "epoch": 0.20144025865869788,
      "grad_norm": 3.7888762950897217,
      "learning_rate": 0.00014865503014432292,
      "loss": 2.2993,
      "step": 514
    },
    {
      "epoch": 0.20183216577671093,
      "grad_norm": 3.3986544609069824,
      "learning_rate": 0.00014847071672991367,
      "loss": 1.9194,
      "step": 515
    },
    {
      "epoch": 0.20222407289472394,
      "grad_norm": 3.4824161529541016,
      "learning_rate": 0.0001482861878357414,
      "loss": 1.7933,
      "step": 516
    },
    {
      "epoch": 0.20261598001273698,
      "grad_norm": 3.770272970199585,
      "learning_rate": 0.00014810144428214144,
      "loss": 2.1901,
      "step": 517
    },
    {
      "epoch": 0.20300788713075002,
      "grad_norm": 3.4453983306884766,
      "learning_rate": 0.0001479164868904034,
      "loss": 2.0661,
      "step": 518
    },
    {
      "epoch": 0.20339979424876303,
      "grad_norm": 3.7636916637420654,
      "learning_rate": 0.00014773131648276758,
      "loss": 1.9643,
      "step": 519
    },
    {
      "epoch": 0.20379170136677607,
      "grad_norm": 4.10495138168335,
      "learning_rate": 0.00014754593388242117,
      "loss": 2.2404,
      "step": 520
    },
    {
      "epoch": 0.20418360848478911,
      "grad_norm": 3.7105236053466797,
      "learning_rate": 0.0001473603399134948,
      "loss": 1.7959,
      "step": 521
    },
    {
      "epoch": 0.20457551560280213,
      "grad_norm": 4.353282928466797,
      "learning_rate": 0.0001471745354010586,
      "loss": 2.6949,
      "step": 522
    },
    {
      "epoch": 0.20496742272081517,
      "grad_norm": 4.381839752197266,
      "learning_rate": 0.00014698852117111884,
      "loss": 2.1864,
      "step": 523
    },
    {
      "epoch": 0.2053593298388282,
      "grad_norm": 4.699688911437988,
      "learning_rate": 0.000146802298050614,
      "loss": 2.0759,
      "step": 524
    },
    {
      "epoch": 0.20575123695684122,
      "grad_norm": 4.765788555145264,
      "learning_rate": 0.0001466158668674112,
      "loss": 2.2164,
      "step": 525
    },
    {
      "epoch": 0.20614314407485426,
      "grad_norm": 4.909315586090088,
      "learning_rate": 0.00014642922845030257,
      "loss": 2.4036,
      "step": 526
    },
    {
      "epoch": 0.2065350511928673,
      "grad_norm": 4.43619441986084,
      "learning_rate": 0.0001462423836290015,
      "loss": 1.8574,
      "step": 527
    },
    {
      "epoch": 0.20692695831088032,
      "grad_norm": 4.671886444091797,
      "learning_rate": 0.00014605533323413887,
      "loss": 2.2032,
      "step": 528
    },
    {
      "epoch": 0.20731886542889336,
      "grad_norm": 4.383705139160156,
      "learning_rate": 0.00014586807809725962,
      "loss": 2.2437,
      "step": 529
    },
    {
      "epoch": 0.2077107725469064,
      "grad_norm": 4.422569274902344,
      "learning_rate": 0.00014568061905081875,
      "loss": 2.2945,
      "step": 530
    },
    {
      "epoch": 0.2081026796649194,
      "grad_norm": 4.5209059715271,
      "learning_rate": 0.00014549295692817778,
      "loss": 2.3492,
      "step": 531
    },
    {
      "epoch": 0.20849458678293245,
      "grad_norm": 4.5657572746276855,
      "learning_rate": 0.00014530509256360102,
      "loss": 2.2819,
      "step": 532
    },
    {
      "epoch": 0.20888649390094546,
      "grad_norm": 4.804976463317871,
      "learning_rate": 0.00014511702679225193,
      "loss": 2.4465,
      "step": 533
    },
    {
      "epoch": 0.2092784010189585,
      "grad_norm": 5.026522159576416,
      "learning_rate": 0.0001449287604501893,
      "loss": 2.1862,
      "step": 534
    },
    {
      "epoch": 0.20967030813697154,
      "grad_norm": 4.093308448791504,
      "learning_rate": 0.00014474029437436348,
      "loss": 1.9233,
      "step": 535
    },
    {
      "epoch": 0.21006221525498456,
      "grad_norm": 5.002185821533203,
      "learning_rate": 0.00014455162940261285,
      "loss": 2.3678,
      "step": 536
    },
    {
      "epoch": 0.2104541223729976,
      "grad_norm": 4.857903003692627,
      "learning_rate": 0.0001443627663736599,
      "loss": 1.9952,
      "step": 537
    },
    {
      "epoch": 0.21084602949101064,
      "grad_norm": 4.661373615264893,
      "learning_rate": 0.00014417370612710778,
      "loss": 2.3169,
      "step": 538
    },
    {
      "epoch": 0.21123793660902365,
      "grad_norm": 5.031216144561768,
      "learning_rate": 0.00014398444950343623,
      "loss": 2.2273,
      "step": 539
    },
    {
      "epoch": 0.2116298437270367,
      "grad_norm": 5.006038188934326,
      "learning_rate": 0.00014379499734399798,
      "loss": 2.544,
      "step": 540
    },
    {
      "epoch": 0.21202175084504973,
      "grad_norm": 5.0471367835998535,
      "learning_rate": 0.0001436053504910151,
      "loss": 2.4543,
      "step": 541
    },
    {
      "epoch": 0.21241365796306275,
      "grad_norm": 5.20703649520874,
      "learning_rate": 0.0001434155097875752,
      "loss": 2.1759,
      "step": 542
    },
    {
      "epoch": 0.2128055650810758,
      "grad_norm": 6.197621822357178,
      "learning_rate": 0.00014322547607762762,
      "loss": 2.3323,
      "step": 543
    },
    {
      "epoch": 0.21319747219908883,
      "grad_norm": 5.347182273864746,
      "learning_rate": 0.0001430352502059797,
      "loss": 2.312,
      "step": 544
    },
    {
      "epoch": 0.21358937931710184,
      "grad_norm": 5.9377522468566895,
      "learning_rate": 0.0001428448330182931,
      "loss": 2.0284,
      "step": 545
    },
    {
      "epoch": 0.21398128643511488,
      "grad_norm": 5.400750160217285,
      "learning_rate": 0.00014265422536107993,
      "loss": 1.9658,
      "step": 546
    },
    {
      "epoch": 0.21437319355312792,
      "grad_norm": 6.990032196044922,
      "learning_rate": 0.00014246342808169914,
      "loss": 2.1277,
      "step": 547
    },
    {
      "epoch": 0.21476510067114093,
      "grad_norm": 6.949008464813232,
      "learning_rate": 0.00014227244202835257,
      "loss": 1.8988,
      "step": 548
    },
    {
      "epoch": 0.21515700778915398,
      "grad_norm": 7.984277725219727,
      "learning_rate": 0.0001420812680500813,
      "loss": 2.3868,
      "step": 549
    },
    {
      "epoch": 0.215548914907167,
      "grad_norm": 9.413142204284668,
      "learning_rate": 0.00014188990699676184,
      "loss": 2.115,
      "step": 550
    },
    {
      "epoch": 0.21594082202518003,
      "grad_norm": 1.9424299001693726,
      "learning_rate": 0.00014169835971910238,
      "loss": 1.0177,
      "step": 551
    },
    {
      "epoch": 0.21633272914319307,
      "grad_norm": 2.0713860988616943,
      "learning_rate": 0.0001415066270686389,
      "loss": 1.2306,
      "step": 552
    },
    {
      "epoch": 0.21672463626120608,
      "grad_norm": 2.247049331665039,
      "learning_rate": 0.00014131470989773158,
      "loss": 1.2081,
      "step": 553
    },
    {
      "epoch": 0.21711654337921912,
      "grad_norm": 2.8240339756011963,
      "learning_rate": 0.0001411226090595608,
      "loss": 1.9308,
      "step": 554
    },
    {
      "epoch": 0.21750845049723216,
      "grad_norm": 1.9344254732131958,
      "learning_rate": 0.00014093032540812348,
      "loss": 1.1561,
      "step": 555
    },
    {
      "epoch": 0.21790035761524518,
      "grad_norm": 2.516598701477051,
      "learning_rate": 0.0001407378597982293,
      "loss": 1.4645,
      "step": 556
    },
    {
      "epoch": 0.21829226473325822,
      "grad_norm": 3.075981616973877,
      "learning_rate": 0.00014054521308549673,
      "loss": 1.6393,
      "step": 557
    },
    {
      "epoch": 0.21868417185127126,
      "grad_norm": 3.4341344833374023,
      "learning_rate": 0.0001403523861263495,
      "loss": 2.3237,
      "step": 558
    },
    {
      "epoch": 0.21907607896928427,
      "grad_norm": 3.016479253768921,
      "learning_rate": 0.00014015937977801256,
      "loss": 1.6781,
      "step": 559
    },
    {
      "epoch": 0.2194679860872973,
      "grad_norm": 3.3220484256744385,
      "learning_rate": 0.00013996619489850822,
      "loss": 2.3032,
      "step": 560
    },
    {
      "epoch": 0.21985989320531035,
      "grad_norm": 3.534581422805786,
      "learning_rate": 0.00013977283234665273,
      "loss": 2.1142,
      "step": 561
    },
    {
      "epoch": 0.22025180032332337,
      "grad_norm": 3.3365044593811035,
      "learning_rate": 0.00013957929298205195,
      "loss": 2.0351,
      "step": 562
    },
    {
      "epoch": 0.2206437074413364,
      "grad_norm": 3.2380876541137695,
      "learning_rate": 0.00013938557766509792,
      "loss": 1.9141,
      "step": 563
    },
    {
      "epoch": 0.22103561455934945,
      "grad_norm": 3.426805257797241,
      "learning_rate": 0.0001391916872569648,
      "loss": 2.0084,
      "step": 564
    },
    {
      "epoch": 0.22142752167736246,
      "grad_norm": 3.502251625061035,
      "learning_rate": 0.00013899762261960518,
      "loss": 1.7679,
      "step": 565
    },
    {
      "epoch": 0.2218194287953755,
      "grad_norm": 3.6389667987823486,
      "learning_rate": 0.0001388033846157462,
      "loss": 2.1981,
      "step": 566
    },
    {
      "epoch": 0.2222113359133885,
      "grad_norm": 3.6977298259735107,
      "learning_rate": 0.0001386089741088857,
      "loss": 2.0532,
      "step": 567
    },
    {
      "epoch": 0.22260324303140155,
      "grad_norm": 3.8215689659118652,
      "learning_rate": 0.00013841439196328836,
      "loss": 2.2478,
      "step": 568
    },
    {
      "epoch": 0.2229951501494146,
      "grad_norm": 3.5352799892425537,
      "learning_rate": 0.00013821963904398193,
      "loss": 2.0626,
      "step": 569
    },
    {
      "epoch": 0.2233870572674276,
      "grad_norm": 4.282069683074951,
      "learning_rate": 0.00013802471621675338,
      "loss": 2.313,
      "step": 570
    },
    {
      "epoch": 0.22377896438544065,
      "grad_norm": 4.219744682312012,
      "learning_rate": 0.00013782962434814492,
      "loss": 2.3029,
      "step": 571
    },
    {
      "epoch": 0.2241708715034537,
      "grad_norm": 4.317134380340576,
      "learning_rate": 0.00013763436430545034,
      "loss": 2.0636,
      "step": 572
    },
    {
      "epoch": 0.2245627786214667,
      "grad_norm": 4.42091178894043,
      "learning_rate": 0.00013743893695671096,
      "loss": 2.2136,
      "step": 573
    },
    {
      "epoch": 0.22495468573947974,
      "grad_norm": 4.037647724151611,
      "learning_rate": 0.00013724334317071198,
      "loss": 1.8086,
      "step": 574
    },
    {
      "epoch": 0.22534659285749278,
      "grad_norm": 4.512540340423584,
      "learning_rate": 0.00013704758381697844,
      "loss": 2.1628,
      "step": 575
    },
    {
      "epoch": 0.2257384999755058,
      "grad_norm": 4.394838333129883,
      "learning_rate": 0.00013685165976577146,
      "loss": 2.0534,
      "step": 576
    },
    {
      "epoch": 0.22613040709351884,
      "grad_norm": 3.961338520050049,
      "learning_rate": 0.0001366555718880843,
      "loss": 1.7181,
      "step": 577
    },
    {
      "epoch": 0.22652231421153188,
      "grad_norm": 4.635353088378906,
      "learning_rate": 0.00013645932105563844,
      "loss": 2.4672,
      "step": 578
    },
    {
      "epoch": 0.2269142213295449,
      "grad_norm": 4.992366313934326,
      "learning_rate": 0.00013626290814088005,
      "loss": 2.4059,
      "step": 579
    },
    {
      "epoch": 0.22730612844755793,
      "grad_norm": 4.703888416290283,
      "learning_rate": 0.00013606633401697557,
      "loss": 2.1682,
      "step": 580
    },
    {
      "epoch": 0.22769803556557097,
      "grad_norm": 5.251219272613525,
      "learning_rate": 0.00013586959955780824,
      "loss": 2.7481,
      "step": 581
    },
    {
      "epoch": 0.22808994268358398,
      "grad_norm": 4.759291648864746,
      "learning_rate": 0.00013567270563797398,
      "loss": 2.3373,
      "step": 582
    },
    {
      "epoch": 0.22848184980159703,
      "grad_norm": 5.617950916290283,
      "learning_rate": 0.00013547565313277776,
      "loss": 2.3882,
      "step": 583
    },
    {
      "epoch": 0.22887375691961004,
      "grad_norm": 4.730344772338867,
      "learning_rate": 0.00013527844291822948,
      "loss": 2.0457,
      "step": 584
    },
    {
      "epoch": 0.22926566403762308,
      "grad_norm": 4.950216770172119,
      "learning_rate": 0.0001350810758710401,
      "loss": 2.5274,
      "step": 585
    },
    {
      "epoch": 0.22965757115563612,
      "grad_norm": 4.839338302612305,
      "learning_rate": 0.00013488355286861783,
      "loss": 2.1933,
      "step": 586
    },
    {
      "epoch": 0.23004947827364913,
      "grad_norm": 4.941376686096191,
      "learning_rate": 0.0001346858747890642,
      "loss": 2.3662,
      "step": 587
    },
    {
      "epoch": 0.23044138539166217,
      "grad_norm": 5.314912796020508,
      "learning_rate": 0.00013448804251117003,
      "loss": 2.292,
      "step": 588
    },
    {
      "epoch": 0.23083329250967521,
      "grad_norm": 5.097457408905029,
      "learning_rate": 0.0001342900569144119,
      "loss": 2.4955,
      "step": 589
    },
    {
      "epoch": 0.23122519962768823,
      "grad_norm": 5.079673767089844,
      "learning_rate": 0.0001340919188789477,
      "loss": 2.0859,
      "step": 590
    },
    {
      "epoch": 0.23161710674570127,
      "grad_norm": 5.03080415725708,
      "learning_rate": 0.00013389362928561317,
      "loss": 1.8944,
      "step": 591
    },
    {
      "epoch": 0.2320090138637143,
      "grad_norm": 5.92555570602417,
      "learning_rate": 0.00013369518901591772,
      "loss": 2.5309,
      "step": 592
    },
    {
      "epoch": 0.23240092098172732,
      "grad_norm": 5.738211154937744,
      "learning_rate": 0.00013349659895204067,
      "loss": 2.3107,
      "step": 593
    },
    {
      "epoch": 0.23279282809974036,
      "grad_norm": 5.753419399261475,
      "learning_rate": 0.0001332978599768272,
      "loss": 2.4748,
      "step": 594
    },
    {
      "epoch": 0.2331847352177534,
      "grad_norm": 5.056491851806641,
      "learning_rate": 0.00013309897297378455,
      "loss": 1.6244,
      "step": 595
    },
    {
      "epoch": 0.23357664233576642,
      "grad_norm": 6.7019195556640625,
      "learning_rate": 0.00013289993882707797,
      "loss": 2.0327,
      "step": 596
    },
    {
      "epoch": 0.23396854945377946,
      "grad_norm": 6.616027355194092,
      "learning_rate": 0.00013270075842152678,
      "loss": 2.1368,
      "step": 597
    },
    {
      "epoch": 0.2343604565717925,
      "grad_norm": 6.35405158996582,
      "learning_rate": 0.00013250143264260074,
      "loss": 1.9113,
      "step": 598
    },
    {
      "epoch": 0.2347523636898055,
      "grad_norm": 8.543137550354004,
      "learning_rate": 0.0001323019623764156,
      "loss": 2.0973,
      "step": 599
    },
    {
      "epoch": 0.23514427080781855,
      "grad_norm": 12.823476791381836,
      "learning_rate": 0.00013210234850972964,
      "loss": 2.1368,
      "step": 600
    },
    {
      "epoch": 0.2355361779258316,
      "grad_norm": 1.6406444311141968,
      "learning_rate": 0.0001319025919299394,
      "loss": 1.363,
      "step": 601
    },
    {
      "epoch": 0.2359280850438446,
      "grad_norm": 1.6655279397964478,
      "learning_rate": 0.00013170269352507597,
      "loss": 0.8635,
      "step": 602
    },
    {
      "epoch": 0.23631999216185764,
      "grad_norm": 1.832244873046875,
      "learning_rate": 0.0001315026541838008,
      "loss": 0.9792,
      "step": 603
    },
    {
      "epoch": 0.23671189927987066,
      "grad_norm": 2.127171516418457,
      "learning_rate": 0.00013130247479540202,
      "loss": 1.4686,
      "step": 604
    },
    {
      "epoch": 0.2371038063978837,
      "grad_norm": 2.5363073348999023,
      "learning_rate": 0.00013110215624979025,
      "loss": 1.7612,
      "step": 605
    },
    {
      "epoch": 0.23749571351589674,
      "grad_norm": 2.310441732406616,
      "learning_rate": 0.00013090169943749476,
      "loss": 1.4604,
      "step": 606
    },
    {
      "epoch": 0.23788762063390975,
      "grad_norm": 2.558234691619873,
      "learning_rate": 0.00013070110524965954,
      "loss": 1.3568,
      "step": 607
    },
    {
      "epoch": 0.2382795277519228,
      "grad_norm": 3.641190528869629,
      "learning_rate": 0.00013050037457803924,
      "loss": 2.4134,
      "step": 608
    },
    {
      "epoch": 0.23867143486993583,
      "grad_norm": 3.0264925956726074,
      "learning_rate": 0.0001302995083149953,
      "loss": 2.1581,
      "step": 609
    },
    {
      "epoch": 0.23906334198794885,
      "grad_norm": 3.2105143070220947,
      "learning_rate": 0.0001300985073534919,
      "loss": 2.0458,
      "step": 610
    },
    {
      "epoch": 0.2394552491059619,
      "grad_norm": 4.151467800140381,
      "learning_rate": 0.00012989737258709203,
      "loss": 2.4478,
      "step": 611
    },
    {
      "epoch": 0.23984715622397493,
      "grad_norm": 3.7684624195098877,
      "learning_rate": 0.00012969610490995358,
      "loss": 2.3586,
      "step": 612
    },
    {
      "epoch": 0.24023906334198794,
      "grad_norm": 3.685589075088501,
      "learning_rate": 0.00012949470521682528,
      "loss": 2.0556,
      "step": 613
    },
    {
      "epoch": 0.24063097046000098,
      "grad_norm": 3.374575614929199,
      "learning_rate": 0.0001292931744030427,
      "loss": 2.2973,
      "step": 614
    },
    {
      "epoch": 0.24102287757801402,
      "grad_norm": 3.7616665363311768,
      "learning_rate": 0.0001290915133645243,
      "loss": 2.1567,
      "step": 615
    },
    {
      "epoch": 0.24141478469602703,
      "grad_norm": 4.679097652435303,
      "learning_rate": 0.00012888972299776754,
      "loss": 1.8012,
      "step": 616
    },
    {
      "epoch": 0.24180669181404008,
      "grad_norm": 3.4685792922973633,
      "learning_rate": 0.00012868780419984482,
      "loss": 1.9529,
      "step": 617
    },
    {
      "epoch": 0.24219859893205312,
      "grad_norm": 3.4313740730285645,
      "learning_rate": 0.00012848575786839943,
      "loss": 2.1742,
      "step": 618
    },
    {
      "epoch": 0.24259050605006613,
      "grad_norm": 4.039928436279297,
      "learning_rate": 0.0001282835849016416,
      "loss": 2.3025,
      "step": 619
    },
    {
      "epoch": 0.24298241316807917,
      "grad_norm": 3.6950714588165283,
      "learning_rate": 0.00012808128619834461,
      "loss": 2.3249,
      "step": 620
    },
    {
      "epoch": 0.24337432028609218,
      "grad_norm": 3.994861602783203,
      "learning_rate": 0.0001278788626578407,
      "loss": 2.521,
      "step": 621
    },
    {
      "epoch": 0.24376622740410522,
      "grad_norm": 3.6618876457214355,
      "learning_rate": 0.00012767631518001698,
      "loss": 1.9042,
      "step": 622
    },
    {
      "epoch": 0.24415813452211826,
      "grad_norm": 4.187183856964111,
      "learning_rate": 0.00012747364466531163,
      "loss": 2.3416,
      "step": 623
    },
    {
      "epoch": 0.24455004164013128,
      "grad_norm": 4.390460968017578,
      "learning_rate": 0.00012727085201470973,
      "loss": 1.8398,
      "step": 624
    },
    {
      "epoch": 0.24494194875814432,
      "grad_norm": 3.8790488243103027,
      "learning_rate": 0.00012706793812973941,
      "loss": 1.7984,
      "step": 625
    },
    {
      "epoch": 0.24533385587615736,
      "grad_norm": 3.9729421138763428,
      "learning_rate": 0.0001268649039124677,
      "loss": 1.8921,
      "step": 626
    },
    {
      "epoch": 0.24572576299417037,
      "grad_norm": 4.497593402862549,
      "learning_rate": 0.00012666175026549662,
      "loss": 2.1272,
      "step": 627
    },
    {
      "epoch": 0.2461176701121834,
      "grad_norm": 5.016843318939209,
      "learning_rate": 0.000126458478091959,
      "loss": 2.2075,
      "step": 628
    },
    {
      "epoch": 0.24650957723019645,
      "grad_norm": 4.8698201179504395,
      "learning_rate": 0.00012625508829551473,
      "loss": 2.2719,
      "step": 629
    },
    {
      "epoch": 0.24690148434820947,
      "grad_norm": 4.81376314163208,
      "learning_rate": 0.00012605158178034654,
      "loss": 2.1356,
      "step": 630
    },
    {
      "epoch": 0.2472933914662225,
      "grad_norm": 5.149053573608398,
      "learning_rate": 0.00012584795945115603,
      "loss": 1.8937,
      "step": 631
    },
    {
      "epoch": 0.24768529858423555,
      "grad_norm": 4.874909400939941,
      "learning_rate": 0.0001256442222131597,
      "loss": 2.0577,
      "step": 632
    },
    {
      "epoch": 0.24807720570224856,
      "grad_norm": 5.0145978927612305,
      "learning_rate": 0.0001254403709720848,
      "loss": 2.2016,
      "step": 633
    },
    {
      "epoch": 0.2484691128202616,
      "grad_norm": 4.988081455230713,
      "learning_rate": 0.0001252364066341655,
      "loss": 2.059,
      "step": 634
    },
    {
      "epoch": 0.24886101993827464,
      "grad_norm": 4.725140571594238,
      "learning_rate": 0.00012503233010613865,
      "loss": 2.1826,
      "step": 635
    },
    {
      "epoch": 0.24925292705628765,
      "grad_norm": 4.772844314575195,
      "learning_rate": 0.00012482814229523997,
      "loss": 1.9065,
      "step": 636
    },
    {
      "epoch": 0.2496448341743007,
      "grad_norm": 4.761850833892822,
      "learning_rate": 0.00012462384410919975,
      "loss": 1.934,
      "step": 637
    },
    {
      "epoch": 0.25003674129231374,
      "grad_norm": 4.64818811416626,
      "learning_rate": 0.00012441943645623903,
      "loss": 2.1843,
      "step": 638
    },
    {
      "epoch": 0.25042864841032675,
      "grad_norm": 5.017551898956299,
      "learning_rate": 0.00012421492024506555,
      "loss": 2.0085,
      "step": 639
    },
    {
      "epoch": 0.25082055552833976,
      "grad_norm": 5.214256286621094,
      "learning_rate": 0.00012401029638486953,
      "loss": 2.1769,
      "step": 640
    },
    {
      "epoch": 0.25121246264635283,
      "grad_norm": 5.537733554840088,
      "learning_rate": 0.0001238055657853198,
      "loss": 1.8173,
      "step": 641
    },
    {
      "epoch": 0.25160436976436584,
      "grad_norm": 5.596163749694824,
      "learning_rate": 0.00012360072935655982,
      "loss": 2.6256,
      "step": 642
    },
    {
      "epoch": 0.25199627688237886,
      "grad_norm": 5.426098823547363,
      "learning_rate": 0.00012339578800920332,
      "loss": 1.7246,
      "step": 643
    },
    {
      "epoch": 0.2523881840003919,
      "grad_norm": 5.337130069732666,
      "learning_rate": 0.00012319074265433063,
      "loss": 2.1601,
      "step": 644
    },
    {
      "epoch": 0.25278009111840494,
      "grad_norm": 4.870218753814697,
      "learning_rate": 0.00012298559420348437,
      "loss": 1.8481,
      "step": 645
    },
    {
      "epoch": 0.25317199823641795,
      "grad_norm": 6.6590471267700195,
      "learning_rate": 0.00012278034356866545,
      "loss": 2.1174,
      "step": 646
    },
    {
      "epoch": 0.253563905354431,
      "grad_norm": 7.224428653717041,
      "learning_rate": 0.00012257499166232907,
      "loss": 1.8234,
      "step": 647
    },
    {
      "epoch": 0.25395581247244403,
      "grad_norm": 7.146002292633057,
      "learning_rate": 0.0001223695393973807,
      "loss": 2.2665,
      "step": 648
    },
    {
      "epoch": 0.25434771959045704,
      "grad_norm": 6.773650169372559,
      "learning_rate": 0.0001221639876871719,
      "loss": 1.3724,
      "step": 649
    },
    {
      "epoch": 0.2547396267084701,
      "grad_norm": 10.051617622375488,
      "learning_rate": 0.0001219583374454963,
      "loss": 1.9674,
      "step": 650
    },
    {
      "epoch": 0.2551315338264831,
      "grad_norm": 1.8488813638687134,
      "learning_rate": 0.00012175258958658564,
      "loss": 1.3794,
      "step": 651
    },
    {
      "epoch": 0.25552344094449614,
      "grad_norm": 1.6268174648284912,
      "learning_rate": 0.00012154674502510555,
      "loss": 0.5106,
      "step": 652
    },
    {
      "epoch": 0.2559153480625092,
      "grad_norm": 2.014295816421509,
      "learning_rate": 0.00012134080467615159,
      "loss": 1.5028,
      "step": 653
    },
    {
      "epoch": 0.2563072551805222,
      "grad_norm": 2.389409065246582,
      "learning_rate": 0.00012113476945524513,
      "loss": 1.7337,
      "step": 654
    },
    {
      "epoch": 0.25669916229853523,
      "grad_norm": 1.951183795928955,
      "learning_rate": 0.00012092864027832933,
      "loss": 1.3969,
      "step": 655
    },
    {
      "epoch": 0.2570910694165483,
      "grad_norm": 2.6875839233398438,
      "learning_rate": 0.000120722418061765,
      "loss": 1.8038,
      "step": 656
    },
    {
      "epoch": 0.2574829765345613,
      "grad_norm": 2.546265125274658,
      "learning_rate": 0.0001205161037223266,
      "loss": 1.3306,
      "step": 657
    },
    {
      "epoch": 0.2578748836525743,
      "grad_norm": 3.043895959854126,
      "learning_rate": 0.00012030969817719808,
      "loss": 1.7659,
      "step": 658
    },
    {
      "epoch": 0.2582667907705874,
      "grad_norm": 3.4225120544433594,
      "learning_rate": 0.00012010320234396894,
      "loss": 1.891,
      "step": 659
    },
    {
      "epoch": 0.2586586978886004,
      "grad_norm": 3.3638079166412354,
      "learning_rate": 0.00011989661714062999,
      "loss": 1.9788,
      "step": 660
    },
    {
      "epoch": 0.2590506050066134,
      "grad_norm": 3.2426440715789795,
      "learning_rate": 0.0001196899434855693,
      "loss": 1.7649,
      "step": 661
    },
    {
      "epoch": 0.2594425121246265,
      "grad_norm": 3.241036891937256,
      "learning_rate": 0.00011948318229756827,
      "loss": 1.9507,
      "step": 662
    },
    {
      "epoch": 0.2598344192426395,
      "grad_norm": 3.3028314113616943,
      "learning_rate": 0.00011927633449579735,
      "loss": 2.1611,
      "step": 663
    },
    {
      "epoch": 0.2602263263606525,
      "grad_norm": 3.6756632328033447,
      "learning_rate": 0.0001190694009998121,
      "loss": 2.2348,
      "step": 664
    },
    {
      "epoch": 0.26061823347866553,
      "grad_norm": 3.3084332942962646,
      "learning_rate": 0.00011886238272954897,
      "loss": 1.9863,
      "step": 665
    },
    {
      "epoch": 0.2610101405966786,
      "grad_norm": 3.353722333908081,
      "learning_rate": 0.00011865528060532127,
      "loss": 1.8989,
      "step": 666
    },
    {
      "epoch": 0.2614020477146916,
      "grad_norm": 4.0705084800720215,
      "learning_rate": 0.0001184480955478152,
      "loss": 2.34,
      "step": 667
    },
    {
      "epoch": 0.2617939548327046,
      "grad_norm": 3.1370604038238525,
      "learning_rate": 0.00011824082847808558,
      "loss": 1.6035,
      "step": 668
    },
    {
      "epoch": 0.2621858619507177,
      "grad_norm": 4.098093509674072,
      "learning_rate": 0.00011803348031755179,
      "loss": 2.0394,
      "step": 669
    },
    {
      "epoch": 0.2625777690687307,
      "grad_norm": 3.729846239089966,
      "learning_rate": 0.0001178260519879937,
      "loss": 1.9398,
      "step": 670
    },
    {
      "epoch": 0.2629696761867437,
      "grad_norm": 3.684011220932007,
      "learning_rate": 0.00011761854441154767,
      "loss": 1.9818,
      "step": 671
    },
    {
      "epoch": 0.2633615833047568,
      "grad_norm": 4.12087869644165,
      "learning_rate": 0.00011741095851070228,
      "loss": 1.9224,
      "step": 672
    },
    {
      "epoch": 0.2637534904227698,
      "grad_norm": 4.0052409172058105,
      "learning_rate": 0.00011720329520829429,
      "loss": 1.9587,
      "step": 673
    },
    {
      "epoch": 0.2641453975407828,
      "grad_norm": 3.933027505874634,
      "learning_rate": 0.0001169955554275046,
      "loss": 2.0708,
      "step": 674
    },
    {
      "epoch": 0.2645373046587959,
      "grad_norm": 3.583723783493042,
      "learning_rate": 0.0001167877400918541,
      "loss": 1.6082,
      "step": 675
    },
    {
      "epoch": 0.2649292117768089,
      "grad_norm": 4.3401198387146,
      "learning_rate": 0.00011657985012519952,
      "loss": 1.6093,
      "step": 676
    },
    {
      "epoch": 0.2653211188948219,
      "grad_norm": 4.603343963623047,
      "learning_rate": 0.00011637188645172944,
      "loss": 1.9834,
      "step": 677
    },
    {
      "epoch": 0.265713026012835,
      "grad_norm": 4.323055267333984,
      "learning_rate": 0.00011616384999596006,
      "loss": 1.6506,
      "step": 678
    },
    {
      "epoch": 0.266104933130848,
      "grad_norm": 4.101743698120117,
      "learning_rate": 0.00011595574168273111,
      "loss": 1.8332,
      "step": 679
    },
    {
      "epoch": 0.266496840248861,
      "grad_norm": 4.907125473022461,
      "learning_rate": 0.0001157475624372018,
      "loss": 2.2441,
      "step": 680
    },
    {
      "epoch": 0.26688874736687407,
      "grad_norm": 4.450485706329346,
      "learning_rate": 0.0001155393131848467,
      "loss": 2.0209,
      "step": 681
    },
    {
      "epoch": 0.2672806544848871,
      "grad_norm": 4.408572196960449,
      "learning_rate": 0.00011533099485145155,
      "loss": 2.3167,
      "step": 682
    },
    {
      "epoch": 0.2676725616029001,
      "grad_norm": 4.719447135925293,
      "learning_rate": 0.00011512260836310924,
      "loss": 2.1785,
      "step": 683
    },
    {
      "epoch": 0.26806446872091316,
      "grad_norm": 4.557635307312012,
      "learning_rate": 0.00011491415464621562,
      "loss": 2.1555,
      "step": 684
    },
    {
      "epoch": 0.2684563758389262,
      "grad_norm": 4.538199424743652,
      "learning_rate": 0.00011470563462746541,
      "loss": 2.1471,
      "step": 685
    },
    {
      "epoch": 0.2688482829569392,
      "grad_norm": 4.565958499908447,
      "learning_rate": 0.00011449704923384812,
      "loss": 1.8063,
      "step": 686
    },
    {
      "epoch": 0.26924019007495226,
      "grad_norm": 4.572544097900391,
      "learning_rate": 0.00011428839939264382,
      "loss": 2.2692,
      "step": 687
    },
    {
      "epoch": 0.26963209719296527,
      "grad_norm": 4.675000190734863,
      "learning_rate": 0.0001140796860314191,
      "loss": 2.2891,
      "step": 688
    },
    {
      "epoch": 0.2700240043109783,
      "grad_norm": 4.8797688484191895,
      "learning_rate": 0.00011387091007802297,
      "loss": 2.5848,
      "step": 689
    },
    {
      "epoch": 0.27041591142899135,
      "grad_norm": 4.958676815032959,
      "learning_rate": 0.0001136620724605827,
      "loss": 2.4756,
      "step": 690
    },
    {
      "epoch": 0.27080781854700436,
      "grad_norm": 4.943301200866699,
      "learning_rate": 0.00011345317410749964,
      "loss": 2.0039,
      "step": 691
    },
    {
      "epoch": 0.2711997256650174,
      "grad_norm": 5.0932817459106445,
      "learning_rate": 0.00011324421594744516,
      "loss": 2.1676,
      "step": 692
    },
    {
      "epoch": 0.27159163278303045,
      "grad_norm": 5.181539058685303,
      "learning_rate": 0.00011303519890935656,
      "loss": 2.0991,
      "step": 693
    },
    {
      "epoch": 0.27198353990104346,
      "grad_norm": 6.030118942260742,
      "learning_rate": 0.00011282612392243286,
      "loss": 2.5055,
      "step": 694
    },
    {
      "epoch": 0.27237544701905647,
      "grad_norm": 6.721047878265381,
      "learning_rate": 0.00011261699191613066,
      "loss": 2.2963,
      "step": 695
    },
    {
      "epoch": 0.27276735413706954,
      "grad_norm": 6.683830261230469,
      "learning_rate": 0.00011240780382016005,
      "loss": 2.3368,
      "step": 696
    },
    {
      "epoch": 0.27315926125508255,
      "grad_norm": 7.51063346862793,
      "learning_rate": 0.00011219856056448051,
      "loss": 2.0302,
      "step": 697
    },
    {
      "epoch": 0.27355116837309557,
      "grad_norm": 7.9241251945495605,
      "learning_rate": 0.00011198926307929664,
      "loss": 2.2444,
      "step": 698
    },
    {
      "epoch": 0.27394307549110863,
      "grad_norm": 9.606241226196289,
      "learning_rate": 0.00011177991229505431,
      "loss": 1.9064,
      "step": 699
    },
    {
      "epoch": 0.27433498260912165,
      "grad_norm": 11.184551239013672,
      "learning_rate": 0.00011157050914243614,
      "loss": 1.6894,
      "step": 700
    },
    {
      "epoch": 0.27472688972713466,
      "grad_norm": 1.7691681385040283,
      "learning_rate": 0.00011136105455235766,
      "loss": 1.0697,
      "step": 701
    },
    {
      "epoch": 0.2751187968451477,
      "grad_norm": 1.953180193901062,
      "learning_rate": 0.00011115154945596305,
      "loss": 1.4046,
      "step": 702
    },
    {
      "epoch": 0.27551070396316074,
      "grad_norm": 2.012139320373535,
      "learning_rate": 0.00011094199478462095,
      "loss": 0.925,
      "step": 703
    },
    {
      "epoch": 0.27590261108117375,
      "grad_norm": 2.0681586265563965,
      "learning_rate": 0.00011073239146992054,
      "loss": 1.1502,
      "step": 704
    },
    {
      "epoch": 0.27629451819918677,
      "grad_norm": 2.858717441558838,
      "learning_rate": 0.00011052274044366711,
      "loss": 1.2743,
      "step": 705
    },
    {
      "epoch": 0.27668642531719984,
      "grad_norm": 2.757593870162964,
      "learning_rate": 0.00011031304263787812,
      "loss": 1.7307,
      "step": 706
    },
    {
      "epoch": 0.27707833243521285,
      "grad_norm": 2.695608377456665,
      "learning_rate": 0.00011010329898477891,
      "loss": 1.7161,
      "step": 707
    },
    {
      "epoch": 0.27747023955322586,
      "grad_norm": 3.355782985687256,
      "learning_rate": 0.0001098935104167988,
      "loss": 2.3015,
      "step": 708
    },
    {
      "epoch": 0.27786214667123893,
      "grad_norm": 3.0121407508850098,
      "learning_rate": 0.00010968367786656663,
      "loss": 2.1074,
      "step": 709
    },
    {
      "epoch": 0.27825405378925194,
      "grad_norm": 3.3730697631835938,
      "learning_rate": 0.00010947380226690684,
      "loss": 2.3066,
      "step": 710
    },
    {
      "epoch": 0.27864596090726496,
      "grad_norm": 3.3932297229766846,
      "learning_rate": 0.00010926388455083522,
      "loss": 2.0767,
      "step": 711
    },
    {
      "epoch": 0.279037868025278,
      "grad_norm": 3.357414722442627,
      "learning_rate": 0.00010905392565155477,
      "loss": 2.1983,
      "step": 712
    },
    {
      "epoch": 0.27942977514329104,
      "grad_norm": 3.569929361343384,
      "learning_rate": 0.00010884392650245165,
      "loss": 2.1021,
      "step": 713
    },
    {
      "epoch": 0.27982168226130405,
      "grad_norm": 3.4855105876922607,
      "learning_rate": 0.00010863388803709089,
      "loss": 2.3997,
      "step": 714
    },
    {
      "epoch": 0.2802135893793171,
      "grad_norm": 3.621185541152954,
      "learning_rate": 0.00010842381118921232,
      "loss": 1.7347,
      "step": 715
    },
    {
      "epoch": 0.28060549649733013,
      "grad_norm": 3.18060040473938,
      "learning_rate": 0.00010821369689272638,
      "loss": 1.8315,
      "step": 716
    },
    {
      "epoch": 0.28099740361534314,
      "grad_norm": 3.2936997413635254,
      "learning_rate": 0.00010800354608171003,
      "loss": 1.8763,
      "step": 717
    },
    {
      "epoch": 0.2813893107333562,
      "grad_norm": 3.8024206161499023,
      "learning_rate": 0.00010779335969040252,
      "loss": 2.1204,
      "step": 718
    },
    {
      "epoch": 0.2817812178513692,
      "grad_norm": 4.0876922607421875,
      "learning_rate": 0.00010758313865320134,
      "loss": 2.5391,
      "step": 719
    },
    {
      "epoch": 0.28217312496938224,
      "grad_norm": 3.5979557037353516,
      "learning_rate": 0.00010737288390465792,
      "loss": 1.7866,
      "step": 720
    },
    {
      "epoch": 0.2825650320873953,
      "grad_norm": 3.624145984649658,
      "learning_rate": 0.00010716259637947357,
      "loss": 2.0231,
      "step": 721
    },
    {
      "epoch": 0.2829569392054083,
      "grad_norm": 3.5677337646484375,
      "learning_rate": 0.00010695227701249537,
      "loss": 1.8784,
      "step": 722
    },
    {
      "epoch": 0.28334884632342133,
      "grad_norm": 3.925076961517334,
      "learning_rate": 0.00010674192673871191,
      "loss": 1.8387,
      "step": 723
    },
    {
      "epoch": 0.2837407534414344,
      "grad_norm": 4.278164386749268,
      "learning_rate": 0.00010653154649324917,
      "loss": 1.8828,
      "step": 724
    },
    {
      "epoch": 0.2841326605594474,
      "grad_norm": 4.132238864898682,
      "learning_rate": 0.00010632113721136636,
      "loss": 2.0964,
      "step": 725
    },
    {
      "epoch": 0.2845245676774604,
      "grad_norm": 4.509267330169678,
      "learning_rate": 0.00010611069982845183,
      "loss": 2.1214,
      "step": 726
    },
    {
      "epoch": 0.2849164747954735,
      "grad_norm": 4.787820339202881,
      "learning_rate": 0.00010590023528001884,
      "loss": 1.906,
      "step": 727
    },
    {
      "epoch": 0.2853083819134865,
      "grad_norm": 4.6233086585998535,
      "learning_rate": 0.00010568974450170139,
      "loss": 2.1821,
      "step": 728
    },
    {
      "epoch": 0.2857002890314995,
      "grad_norm": 4.608364582061768,
      "learning_rate": 0.00010547922842925008,
      "loss": 2.1558,
      "step": 729
    },
    {
      "epoch": 0.2860921961495126,
      "grad_norm": 4.798117160797119,
      "learning_rate": 0.00010526868799852796,
      "loss": 2.3414,
      "step": 730
    },
    {
      "epoch": 0.2864841032675256,
      "grad_norm": 4.810368537902832,
      "learning_rate": 0.0001050581241455064,
      "loss": 2.28,
      "step": 731
    },
    {
      "epoch": 0.2868760103855386,
      "grad_norm": 4.538679599761963,
      "learning_rate": 0.00010484753780626089,
      "loss": 2.3225,
      "step": 732
    },
    {
      "epoch": 0.2872679175035517,
      "grad_norm": 4.918915748596191,
      "learning_rate": 0.00010463692991696685,
      "loss": 2.3115,
      "step": 733
    },
    {
      "epoch": 0.2876598246215647,
      "grad_norm": 4.871672630310059,
      "learning_rate": 0.00010442630141389549,
      "loss": 2.2778,
      "step": 734
    },
    {
      "epoch": 0.2880517317395777,
      "grad_norm": 4.887431621551514,
      "learning_rate": 0.00010421565323340971,
      "loss": 2.0452,
      "step": 735
    },
    {
      "epoch": 0.2884436388575908,
      "grad_norm": 4.556756973266602,
      "learning_rate": 0.00010400498631195992,
      "loss": 2.03,
      "step": 736
    },
    {
      "epoch": 0.2888355459756038,
      "grad_norm": 4.994616508483887,
      "learning_rate": 0.00010379430158607975,
      "loss": 2.1591,
      "step": 737
    },
    {
      "epoch": 0.2892274530936168,
      "grad_norm": 4.89581823348999,
      "learning_rate": 0.000103583599992382,
      "loss": 2.0955,
      "step": 738
    },
    {
      "epoch": 0.2896193602116298,
      "grad_norm": 4.764329433441162,
      "learning_rate": 0.0001033728824675545,
      "loss": 2.0741,
      "step": 739
    },
    {
      "epoch": 0.2900112673296429,
      "grad_norm": 5.454378604888916,
      "learning_rate": 0.0001031621499483559,
      "loss": 2.3522,
      "step": 740
    },
    {
      "epoch": 0.2904031744476559,
      "grad_norm": 4.7729082107543945,
      "learning_rate": 0.00010295140337161146,
      "loss": 1.5703,
      "step": 741
    },
    {
      "epoch": 0.2907950815656689,
      "grad_norm": 4.945082187652588,
      "learning_rate": 0.00010274064367420897,
      "loss": 1.9774,
      "step": 742
    },
    {
      "epoch": 0.291186988683682,
      "grad_norm": 5.7353925704956055,
      "learning_rate": 0.00010252987179309459,
      "loss": 1.8481,
      "step": 743
    },
    {
      "epoch": 0.291578895801695,
      "grad_norm": 5.375218868255615,
      "learning_rate": 0.00010231908866526851,
      "loss": 1.855,
      "step": 744
    },
    {
      "epoch": 0.291970802919708,
      "grad_norm": 6.668516159057617,
      "learning_rate": 0.00010210829522778111,
      "loss": 2.2298,
      "step": 745
    },
    {
      "epoch": 0.2923627100377211,
      "grad_norm": 7.301716327667236,
      "learning_rate": 0.00010189749241772844,
      "loss": 1.5179,
      "step": 746
    },
    {
      "epoch": 0.2927546171557341,
      "grad_norm": 6.345820426940918,
      "learning_rate": 0.00010168668117224825,
      "loss": 1.8064,
      "step": 747
    },
    {
      "epoch": 0.2931465242737471,
      "grad_norm": 7.448675155639648,
      "learning_rate": 0.00010147586242851585,
      "loss": 1.9337,
      "step": 748
    },
    {
      "epoch": 0.29353843139176017,
      "grad_norm": 9.384629249572754,
      "learning_rate": 0.00010126503712373982,
      "loss": 2.114,
      "step": 749
    },
    {
      "epoch": 0.2939303385097732,
      "grad_norm": 14.601536750793457,
      "learning_rate": 0.00010105420619515798,
      "loss": 2.0272,
      "step": 750
    },
    {
      "epoch": 0.2939303385097732,
      "eval_loss": 1.9826831817626953,
      "eval_runtime": 38.0676,
      "eval_samples_per_second": 56.452,
      "eval_steps_per_second": 28.239,
      "step": 750
    },
    {
      "epoch": 0.2943222456277862,
      "grad_norm": 1.8205848932266235,
      "learning_rate": 0.00010084337058003303,
      "loss": 1.1298,
      "step": 751
    },
    {
      "epoch": 0.29471415274579926,
      "grad_norm": 1.7140920162200928,
      "learning_rate": 0.00010063253121564868,
      "loss": 0.8539,
      "step": 752
    },
    {
      "epoch": 0.2951060598638123,
      "grad_norm": 1.9727723598480225,
      "learning_rate": 0.00010042168903930514,
      "loss": 1.2751,
      "step": 753
    },
    {
      "epoch": 0.2954979669818253,
      "grad_norm": 1.6973793506622314,
      "learning_rate": 0.00010021084498831522,
      "loss": 0.7254,
      "step": 754
    },
    {
      "epoch": 0.29588987409983836,
      "grad_norm": 1.8619134426116943,
      "learning_rate": 0.0001,
      "loss": 1.2396,
      "step": 755
    },
    {
      "epoch": 0.29628178121785137,
      "grad_norm": 2.6488993167877197,
      "learning_rate": 9.97891550116848e-05,
      "loss": 1.9244,
      "step": 756
    },
    {
      "epoch": 0.2966736883358644,
      "grad_norm": 3.2024765014648438,
      "learning_rate": 9.957831096069488e-05,
      "loss": 2.3968,
      "step": 757
    },
    {
      "epoch": 0.29706559545387745,
      "grad_norm": 3.0408380031585693,
      "learning_rate": 9.936746878435136e-05,
      "loss": 1.8445,
      "step": 758
    },
    {
      "epoch": 0.29745750257189046,
      "grad_norm": 3.1097819805145264,
      "learning_rate": 9.915662941996699e-05,
      "loss": 2.0624,
      "step": 759
    },
    {
      "epoch": 0.2978494096899035,
      "grad_norm": 3.2179934978485107,
      "learning_rate": 9.894579380484204e-05,
      "loss": 1.838,
      "step": 760
    },
    {
      "epoch": 0.29824131680791655,
      "grad_norm": 3.476102590560913,
      "learning_rate": 9.873496287626019e-05,
      "loss": 2.1114,
      "step": 761
    },
    {
      "epoch": 0.29863322392592956,
      "grad_norm": 3.8200204372406006,
      "learning_rate": 9.852413757148417e-05,
      "loss": 2.1493,
      "step": 762
    },
    {
      "epoch": 0.29902513104394257,
      "grad_norm": 3.368173599243164,
      "learning_rate": 9.831331882775178e-05,
      "loss": 2.3542,
      "step": 763
    },
    {
      "epoch": 0.29941703816195564,
      "grad_norm": 3.4443352222442627,
      "learning_rate": 9.81025075822716e-05,
      "loss": 2.2044,
      "step": 764
    },
    {
      "epoch": 0.29980894527996865,
      "grad_norm": 3.226194381713867,
      "learning_rate": 9.789170477221891e-05,
      "loss": 1.6747,
      "step": 765
    },
    {
      "epoch": 0.30020085239798167,
      "grad_norm": 3.806288480758667,
      "learning_rate": 9.76809113347315e-05,
      "loss": 2.1696,
      "step": 766
    },
    {
      "epoch": 0.30059275951599473,
      "grad_norm": 3.4827165603637695,
      "learning_rate": 9.747012820690543e-05,
      "loss": 1.8458,
      "step": 767
    },
    {
      "epoch": 0.30098466663400775,
      "grad_norm": 3.2971951961517334,
      "learning_rate": 9.725935632579104e-05,
      "loss": 2.1725,
      "step": 768
    },
    {
      "epoch": 0.30137657375202076,
      "grad_norm": 3.6433308124542236,
      "learning_rate": 9.704859662838855e-05,
      "loss": 1.7483,
      "step": 769
    },
    {
      "epoch": 0.30176848087003383,
      "grad_norm": 3.6657750606536865,
      "learning_rate": 9.683785005164411e-05,
      "loss": 1.9419,
      "step": 770
    },
    {
      "epoch": 0.30216038798804684,
      "grad_norm": 3.5587973594665527,
      "learning_rate": 9.662711753244551e-05,
      "loss": 1.7249,
      "step": 771
    },
    {
      "epoch": 0.30255229510605985,
      "grad_norm": 4.286134719848633,
      "learning_rate": 9.641640000761802e-05,
      "loss": 1.9738,
      "step": 772
    },
    {
      "epoch": 0.30294420222407287,
      "grad_norm": 4.182912826538086,
      "learning_rate": 9.620569841392029e-05,
      "loss": 1.8668,
      "step": 773
    },
    {
      "epoch": 0.30333610934208594,
      "grad_norm": 4.520855903625488,
      "learning_rate": 9.59950136880401e-05,
      "loss": 2.0032,
      "step": 774
    },
    {
      "epoch": 0.30372801646009895,
      "grad_norm": 3.9210479259490967,
      "learning_rate": 9.57843467665903e-05,
      "loss": 1.7421,
      "step": 775
    },
    {
      "epoch": 0.30411992357811196,
      "grad_norm": 4.59743595123291,
      "learning_rate": 9.557369858610453e-05,
      "loss": 2.1979,
      "step": 776
    },
    {
      "epoch": 0.30451183069612503,
      "grad_norm": 4.364266395568848,
      "learning_rate": 9.53630700830332e-05,
      "loss": 2.1655,
      "step": 777
    },
    {
      "epoch": 0.30490373781413804,
      "grad_norm": 5.17692756652832,
      "learning_rate": 9.51524621937391e-05,
      "loss": 2.1983,
      "step": 778
    },
    {
      "epoch": 0.30529564493215106,
      "grad_norm": 4.385379791259766,
      "learning_rate": 9.494187585449358e-05,
      "loss": 1.8081,
      "step": 779
    },
    {
      "epoch": 0.3056875520501641,
      "grad_norm": 4.785798072814941,
      "learning_rate": 9.473131200147205e-05,
      "loss": 2.1047,
      "step": 780
    },
    {
      "epoch": 0.30607945916817714,
      "grad_norm": 4.512739658355713,
      "learning_rate": 9.452077157074994e-05,
      "loss": 2.0648,
      "step": 781
    },
    {
      "epoch": 0.30647136628619015,
      "grad_norm": 4.7011027336120605,
      "learning_rate": 9.431025549829862e-05,
      "loss": 2.1965,
      "step": 782
    },
    {
      "epoch": 0.3068632734042032,
      "grad_norm": 4.651373863220215,
      "learning_rate": 9.409976471998118e-05,
      "loss": 2.2154,
      "step": 783
    },
    {
      "epoch": 0.30725518052221623,
      "grad_norm": 5.008145332336426,
      "learning_rate": 9.388930017154819e-05,
      "loss": 2.5322,
      "step": 784
    },
    {
      "epoch": 0.30764708764022924,
      "grad_norm": 4.943636417388916,
      "learning_rate": 9.367886278863366e-05,
      "loss": 2.1245,
      "step": 785
    },
    {
      "epoch": 0.3080389947582423,
      "grad_norm": 5.034031867980957,
      "learning_rate": 9.346845350675088e-05,
      "loss": 2.5965,
      "step": 786
    },
    {
      "epoch": 0.3084309018762553,
      "grad_norm": 4.50614070892334,
      "learning_rate": 9.325807326128814e-05,
      "loss": 2.0138,
      "step": 787
    },
    {
      "epoch": 0.30882280899426834,
      "grad_norm": 4.857085704803467,
      "learning_rate": 9.304772298750463e-05,
      "loss": 2.177,
      "step": 788
    },
    {
      "epoch": 0.3092147161122814,
      "grad_norm": 4.98145866394043,
      "learning_rate": 9.283740362052642e-05,
      "loss": 2.081,
      "step": 789
    },
    {
      "epoch": 0.3096066232302944,
      "grad_norm": 4.522789478302002,
      "learning_rate": 9.26271160953421e-05,
      "loss": 1.8516,
      "step": 790
    },
    {
      "epoch": 0.30999853034830743,
      "grad_norm": 4.803216457366943,
      "learning_rate": 9.241686134679867e-05,
      "loss": 1.9302,
      "step": 791
    },
    {
      "epoch": 0.3103904374663205,
      "grad_norm": 5.030069828033447,
      "learning_rate": 9.220664030959749e-05,
      "loss": 1.8447,
      "step": 792
    },
    {
      "epoch": 0.3107823445843335,
      "grad_norm": 5.31541633605957,
      "learning_rate": 9.199645391828999e-05,
      "loss": 2.0349,
      "step": 793
    },
    {
      "epoch": 0.3111742517023465,
      "grad_norm": 6.141101837158203,
      "learning_rate": 9.178630310727365e-05,
      "loss": 1.8507,
      "step": 794
    },
    {
      "epoch": 0.3115661588203596,
      "grad_norm": 5.377287864685059,
      "learning_rate": 9.157618881078772e-05,
      "loss": 1.5943,
      "step": 795
    },
    {
      "epoch": 0.3119580659383726,
      "grad_norm": 6.283132076263428,
      "learning_rate": 9.136611196290915e-05,
      "loss": 1.921,
      "step": 796
    },
    {
      "epoch": 0.3123499730563856,
      "grad_norm": 6.990156650543213,
      "learning_rate": 9.115607349754834e-05,
      "loss": 1.6183,
      "step": 797
    },
    {
      "epoch": 0.3127418801743987,
      "grad_norm": 8.291764259338379,
      "learning_rate": 9.094607434844523e-05,
      "loss": 2.2806,
      "step": 798
    },
    {
      "epoch": 0.3131337872924117,
      "grad_norm": 10.434477806091309,
      "learning_rate": 9.07361154491648e-05,
      "loss": 2.0486,
      "step": 799
    },
    {
      "epoch": 0.3135256944104247,
      "grad_norm": 13.843595504760742,
      "learning_rate": 9.052619773309317e-05,
      "loss": 2.2136,
      "step": 800
    },
    {
      "epoch": 0.3139176015284378,
      "grad_norm": 1.5639240741729736,
      "learning_rate": 9.031632213343339e-05,
      "loss": 1.3076,
      "step": 801
    },
    {
      "epoch": 0.3143095086464508,
      "grad_norm": 1.711394190788269,
      "learning_rate": 9.01064895832012e-05,
      "loss": 1.4351,
      "step": 802
    },
    {
      "epoch": 0.3147014157644638,
      "grad_norm": 2.63399600982666,
      "learning_rate": 8.98967010152211e-05,
      "loss": 1.6325,
      "step": 803
    },
    {
      "epoch": 0.3150933228824769,
      "grad_norm": 2.4343230724334717,
      "learning_rate": 8.968695736212193e-05,
      "loss": 1.6442,
      "step": 804
    },
    {
      "epoch": 0.3154852300004899,
      "grad_norm": 2.337390899658203,
      "learning_rate": 8.947725955633294e-05,
      "loss": 1.9179,
      "step": 805
    },
    {
      "epoch": 0.3158771371185029,
      "grad_norm": 2.651967763900757,
      "learning_rate": 8.926760853007946e-05,
      "loss": 2.1174,
      "step": 806
    },
    {
      "epoch": 0.316269044236516,
      "grad_norm": 2.955799102783203,
      "learning_rate": 8.905800521537905e-05,
      "loss": 2.096,
      "step": 807
    },
    {
      "epoch": 0.316660951354529,
      "grad_norm": 2.910703420639038,
      "learning_rate": 8.884845054403699e-05,
      "loss": 1.851,
      "step": 808
    },
    {
      "epoch": 0.317052858472542,
      "grad_norm": 2.885549545288086,
      "learning_rate": 8.863894544764236e-05,
      "loss": 1.7677,
      "step": 809
    },
    {
      "epoch": 0.317444765590555,
      "grad_norm": 3.1844494342803955,
      "learning_rate": 8.84294908575639e-05,
      "loss": 2.0917,
      "step": 810
    },
    {
      "epoch": 0.3178366727085681,
      "grad_norm": 3.2735683917999268,
      "learning_rate": 8.822008770494572e-05,
      "loss": 2.0782,
      "step": 811
    },
    {
      "epoch": 0.3182285798265811,
      "grad_norm": 3.346047878265381,
      "learning_rate": 8.801073692070337e-05,
      "loss": 2.1218,
      "step": 812
    },
    {
      "epoch": 0.3186204869445941,
      "grad_norm": 3.1497979164123535,
      "learning_rate": 8.780143943551954e-05,
      "loss": 1.7055,
      "step": 813
    },
    {
      "epoch": 0.3190123940626072,
      "grad_norm": 3.5353710651397705,
      "learning_rate": 8.759219617983999e-05,
      "loss": 1.7831,
      "step": 814
    },
    {
      "epoch": 0.3194043011806202,
      "grad_norm": 3.6720974445343018,
      "learning_rate": 8.738300808386935e-05,
      "loss": 2.2325,
      "step": 815
    },
    {
      "epoch": 0.3197962082986332,
      "grad_norm": 3.518143653869629,
      "learning_rate": 8.717387607756713e-05,
      "loss": 1.7664,
      "step": 816
    },
    {
      "epoch": 0.32018811541664627,
      "grad_norm": 3.6279921531677246,
      "learning_rate": 8.696480109064342e-05,
      "loss": 1.8673,
      "step": 817
    },
    {
      "epoch": 0.3205800225346593,
      "grad_norm": 3.6512207984924316,
      "learning_rate": 8.675578405255485e-05,
      "loss": 1.9554,
      "step": 818
    },
    {
      "epoch": 0.3209719296526723,
      "grad_norm": 3.595334529876709,
      "learning_rate": 8.654682589250038e-05,
      "loss": 1.8544,
      "step": 819
    },
    {
      "epoch": 0.32136383677068536,
      "grad_norm": 3.5271990299224854,
      "learning_rate": 8.633792753941733e-05,
      "loss": 1.7899,
      "step": 820
    },
    {
      "epoch": 0.3217557438886984,
      "grad_norm": 3.9318435192108154,
      "learning_rate": 8.612908992197705e-05,
      "loss": 1.7226,
      "step": 821
    },
    {
      "epoch": 0.3221476510067114,
      "grad_norm": 3.9089550971984863,
      "learning_rate": 8.592031396858093e-05,
      "loss": 2.0319,
      "step": 822
    },
    {
      "epoch": 0.32253955812472446,
      "grad_norm": 3.938750743865967,
      "learning_rate": 8.571160060735624e-05,
      "loss": 1.8931,
      "step": 823
    },
    {
      "epoch": 0.32293146524273747,
      "grad_norm": 4.519107341766357,
      "learning_rate": 8.550295076615188e-05,
      "loss": 2.0449,
      "step": 824
    },
    {
      "epoch": 0.3233233723607505,
      "grad_norm": 4.3217315673828125,
      "learning_rate": 8.529436537253458e-05,
      "loss": 1.6261,
      "step": 825
    },
    {
      "epoch": 0.32371527947876355,
      "grad_norm": 4.959157466888428,
      "learning_rate": 8.508584535378439e-05,
      "loss": 2.0614,
      "step": 826
    },
    {
      "epoch": 0.32410718659677656,
      "grad_norm": 4.592066287994385,
      "learning_rate": 8.487739163689079e-05,
      "loss": 1.9628,
      "step": 827
    },
    {
      "epoch": 0.3244990937147896,
      "grad_norm": 4.376495361328125,
      "learning_rate": 8.466900514854847e-05,
      "loss": 2.0595,
      "step": 828
    },
    {
      "epoch": 0.32489100083280265,
      "grad_norm": 4.515799045562744,
      "learning_rate": 8.446068681515334e-05,
      "loss": 2.0949,
      "step": 829
    },
    {
      "epoch": 0.32528290795081566,
      "grad_norm": 4.855846881866455,
      "learning_rate": 8.425243756279824e-05,
      "loss": 2.0947,
      "step": 830
    },
    {
      "epoch": 0.32567481506882867,
      "grad_norm": 5.449527740478516,
      "learning_rate": 8.404425831726894e-05,
      "loss": 2.4299,
      "step": 831
    },
    {
      "epoch": 0.32606672218684174,
      "grad_norm": 4.645267009735107,
      "learning_rate": 8.383615000404e-05,
      "loss": 1.9787,
      "step": 832
    },
    {
      "epoch": 0.32645862930485475,
      "grad_norm": 4.741397857666016,
      "learning_rate": 8.362811354827059e-05,
      "loss": 2.1863,
      "step": 833
    },
    {
      "epoch": 0.32685053642286777,
      "grad_norm": 4.713899612426758,
      "learning_rate": 8.342014987480047e-05,
      "loss": 2.1401,
      "step": 834
    },
    {
      "epoch": 0.32724244354088083,
      "grad_norm": 4.170632839202881,
      "learning_rate": 8.321225990814591e-05,
      "loss": 2.0202,
      "step": 835
    },
    {
      "epoch": 0.32763435065889385,
      "grad_norm": 4.701237201690674,
      "learning_rate": 8.300444457249543e-05,
      "loss": 2.1456,
      "step": 836
    },
    {
      "epoch": 0.32802625777690686,
      "grad_norm": 4.378878593444824,
      "learning_rate": 8.279670479170573e-05,
      "loss": 1.9442,
      "step": 837
    },
    {
      "epoch": 0.32841816489491993,
      "grad_norm": 4.744080543518066,
      "learning_rate": 8.258904148929775e-05,
      "loss": 2.2387,
      "step": 838
    },
    {
      "epoch": 0.32881007201293294,
      "grad_norm": 4.985557556152344,
      "learning_rate": 8.238145558845235e-05,
      "loss": 2.0518,
      "step": 839
    },
    {
      "epoch": 0.32920197913094595,
      "grad_norm": 5.339816093444824,
      "learning_rate": 8.217394801200631e-05,
      "loss": 2.0914,
      "step": 840
    },
    {
      "epoch": 0.329593886248959,
      "grad_norm": 4.7329254150390625,
      "learning_rate": 8.196651968244826e-05,
      "loss": 1.9135,
      "step": 841
    },
    {
      "epoch": 0.32998579336697204,
      "grad_norm": 5.212874889373779,
      "learning_rate": 8.175917152191447e-05,
      "loss": 1.5935,
      "step": 842
    },
    {
      "epoch": 0.33037770048498505,
      "grad_norm": 5.2939300537109375,
      "learning_rate": 8.15519044521848e-05,
      "loss": 2.1032,
      "step": 843
    },
    {
      "epoch": 0.3307696076029981,
      "grad_norm": 5.305154800415039,
      "learning_rate": 8.134471939467874e-05,
      "loss": 1.7443,
      "step": 844
    },
    {
      "epoch": 0.33116151472101113,
      "grad_norm": 6.5364227294921875,
      "learning_rate": 8.113761727045105e-05,
      "loss": 2.1122,
      "step": 845
    },
    {
      "epoch": 0.33155342183902414,
      "grad_norm": 6.236037731170654,
      "learning_rate": 8.093059900018792e-05,
      "loss": 1.7195,
      "step": 846
    },
    {
      "epoch": 0.33194532895703716,
      "grad_norm": 6.322172164916992,
      "learning_rate": 8.072366550420266e-05,
      "loss": 1.7197,
      "step": 847
    },
    {
      "epoch": 0.3323372360750502,
      "grad_norm": 6.858587265014648,
      "learning_rate": 8.051681770243175e-05,
      "loss": 1.7137,
      "step": 848
    },
    {
      "epoch": 0.33272914319306324,
      "grad_norm": 9.607586860656738,
      "learning_rate": 8.031005651443073e-05,
      "loss": 2.2746,
      "step": 849
    },
    {
      "epoch": 0.33312105031107625,
      "grad_norm": 12.31042194366455,
      "learning_rate": 8.010338285937006e-05,
      "loss": 1.645,
      "step": 850
    },
    {
      "epoch": 0.3335129574290893,
      "grad_norm": 1.3039801120758057,
      "learning_rate": 7.989679765603108e-05,
      "loss": 1.1509,
      "step": 851
    },
    {
      "epoch": 0.33390486454710233,
      "grad_norm": 1.5482182502746582,
      "learning_rate": 7.969030182280192e-05,
      "loss": 0.8718,
      "step": 852
    },
    {
      "epoch": 0.33429677166511534,
      "grad_norm": 1.7824316024780273,
      "learning_rate": 7.948389627767343e-05,
      "loss": 0.8929,
      "step": 853
    },
    {
      "epoch": 0.3346886787831284,
      "grad_norm": 2.098639965057373,
      "learning_rate": 7.927758193823501e-05,
      "loss": 1.6104,
      "step": 854
    },
    {
      "epoch": 0.3350805859011414,
      "grad_norm": 2.3912746906280518,
      "learning_rate": 7.907135972167069e-05,
      "loss": 1.6464,
      "step": 855
    },
    {
      "epoch": 0.33547249301915444,
      "grad_norm": 2.458956003189087,
      "learning_rate": 7.88652305447549e-05,
      "loss": 1.7483,
      "step": 856
    },
    {
      "epoch": 0.3358644001371675,
      "grad_norm": 2.5473175048828125,
      "learning_rate": 7.865919532384844e-05,
      "loss": 1.5768,
      "step": 857
    },
    {
      "epoch": 0.3362563072551805,
      "grad_norm": 2.90496826171875,
      "learning_rate": 7.845325497489449e-05,
      "loss": 2.0058,
      "step": 858
    },
    {
      "epoch": 0.33664821437319353,
      "grad_norm": 2.830495834350586,
      "learning_rate": 7.82474104134144e-05,
      "loss": 1.5419,
      "step": 859
    },
    {
      "epoch": 0.3370401214912066,
      "grad_norm": 3.044736862182617,
      "learning_rate": 7.804166255450373e-05,
      "loss": 1.8302,
      "step": 860
    },
    {
      "epoch": 0.3374320286092196,
      "grad_norm": 3.152892589569092,
      "learning_rate": 7.783601231282812e-05,
      "loss": 1.8738,
      "step": 861
    },
    {
      "epoch": 0.3378239357272326,
      "grad_norm": 3.2449419498443604,
      "learning_rate": 7.763046060261932e-05,
      "loss": 1.8796,
      "step": 862
    },
    {
      "epoch": 0.3382158428452457,
      "grad_norm": 3.386857509613037,
      "learning_rate": 7.742500833767094e-05,
      "loss": 2.2094,
      "step": 863
    },
    {
      "epoch": 0.3386077499632587,
      "grad_norm": 3.4558780193328857,
      "learning_rate": 7.721965643133458e-05,
      "loss": 2.0639,
      "step": 864
    },
    {
      "epoch": 0.3389996570812717,
      "grad_norm": 3.317131996154785,
      "learning_rate": 7.701440579651566e-05,
      "loss": 1.7028,
      "step": 865
    },
    {
      "epoch": 0.3393915641992848,
      "grad_norm": 3.3992812633514404,
      "learning_rate": 7.680925734566937e-05,
      "loss": 2.125,
      "step": 866
    },
    {
      "epoch": 0.3397834713172978,
      "grad_norm": 3.851240396499634,
      "learning_rate": 7.660421199079669e-05,
      "loss": 2.1039,
      "step": 867
    },
    {
      "epoch": 0.3401753784353108,
      "grad_norm": 3.7830820083618164,
      "learning_rate": 7.639927064344022e-05,
      "loss": 2.2499,
      "step": 868
    },
    {
      "epoch": 0.3405672855533239,
      "grad_norm": 3.8149750232696533,
      "learning_rate": 7.619443421468021e-05,
      "loss": 2.0121,
      "step": 869
    },
    {
      "epoch": 0.3409591926713369,
      "grad_norm": 4.0605788230896,
      "learning_rate": 7.598970361513051e-05,
      "loss": 1.8563,
      "step": 870
    },
    {
      "epoch": 0.3413510997893499,
      "grad_norm": 4.0026726722717285,
      "learning_rate": 7.578507975493448e-05,
      "loss": 2.2058,
      "step": 871
    },
    {
      "epoch": 0.341743006907363,
      "grad_norm": 3.586193323135376,
      "learning_rate": 7.558056354376098e-05,
      "loss": 1.639,
      "step": 872
    },
    {
      "epoch": 0.342134914025376,
      "grad_norm": 4.302838325500488,
      "learning_rate": 7.537615589080027e-05,
      "loss": 2.0662,
      "step": 873
    },
    {
      "epoch": 0.342526821143389,
      "grad_norm": 3.8509387969970703,
      "learning_rate": 7.517185770476006e-05,
      "loss": 1.766,
      "step": 874
    },
    {
      "epoch": 0.3429187282614021,
      "grad_norm": 4.2101359367370605,
      "learning_rate": 7.496766989386136e-05,
      "loss": 2.0154,
      "step": 875
    },
    {
      "epoch": 0.3433106353794151,
      "grad_norm": 4.195278644561768,
      "learning_rate": 7.476359336583454e-05,
      "loss": 1.8871,
      "step": 876
    },
    {
      "epoch": 0.3437025424974281,
      "grad_norm": 4.238273620605469,
      "learning_rate": 7.455962902791522e-05,
      "loss": 1.747,
      "step": 877
    },
    {
      "epoch": 0.34409444961544117,
      "grad_norm": 4.351687908172607,
      "learning_rate": 7.435577778684033e-05,
      "loss": 2.1618,
      "step": 878
    },
    {
      "epoch": 0.3444863567334542,
      "grad_norm": 4.1494550704956055,
      "learning_rate": 7.415204054884399e-05,
      "loss": 2.1521,
      "step": 879
    },
    {
      "epoch": 0.3448782638514672,
      "grad_norm": 5.003871917724609,
      "learning_rate": 7.394841821965345e-05,
      "loss": 2.1101,
      "step": 880
    },
    {
      "epoch": 0.3452701709694802,
      "grad_norm": 4.884603023529053,
      "learning_rate": 7.374491170448525e-05,
      "loss": 2.2218,
      "step": 881
    },
    {
      "epoch": 0.3456620780874933,
      "grad_norm": 4.531919479370117,
      "learning_rate": 7.3541521908041e-05,
      "loss": 2.2419,
      "step": 882
    },
    {
      "epoch": 0.3460539852055063,
      "grad_norm": 5.185380458831787,
      "learning_rate": 7.33382497345034e-05,
      "loss": 2.3322,
      "step": 883
    },
    {
      "epoch": 0.3464458923235193,
      "grad_norm": 5.319368362426758,
      "learning_rate": 7.313509608753231e-05,
      "loss": 2.5454,
      "step": 884
    },
    {
      "epoch": 0.34683779944153237,
      "grad_norm": 4.65814208984375,
      "learning_rate": 7.293206187026061e-05,
      "loss": 2.2161,
      "step": 885
    },
    {
      "epoch": 0.3472297065595454,
      "grad_norm": 4.533596515655518,
      "learning_rate": 7.27291479852903e-05,
      "loss": 2.0049,
      "step": 886
    },
    {
      "epoch": 0.3476216136775584,
      "grad_norm": 5.1505889892578125,
      "learning_rate": 7.252635533468843e-05,
      "loss": 2.0545,
      "step": 887
    },
    {
      "epoch": 0.34801352079557146,
      "grad_norm": 4.4196457862854,
      "learning_rate": 7.232368481998309e-05,
      "loss": 1.7763,
      "step": 888
    },
    {
      "epoch": 0.3484054279135845,
      "grad_norm": 5.280087947845459,
      "learning_rate": 7.212113734215932e-05,
      "loss": 2.5429,
      "step": 889
    },
    {
      "epoch": 0.3487973350315975,
      "grad_norm": 4.985827445983887,
      "learning_rate": 7.191871380165538e-05,
      "loss": 2.1089,
      "step": 890
    },
    {
      "epoch": 0.34918924214961056,
      "grad_norm": 4.899125576019287,
      "learning_rate": 7.17164150983584e-05,
      "loss": 1.8911,
      "step": 891
    },
    {
      "epoch": 0.34958114926762357,
      "grad_norm": 4.845853805541992,
      "learning_rate": 7.151424213160061e-05,
      "loss": 2.0591,
      "step": 892
    },
    {
      "epoch": 0.3499730563856366,
      "grad_norm": 5.270240783691406,
      "learning_rate": 7.131219580015521e-05,
      "loss": 2.1822,
      "step": 893
    },
    {
      "epoch": 0.35036496350364965,
      "grad_norm": 5.15275764465332,
      "learning_rate": 7.11102770022325e-05,
      "loss": 2.2057,
      "step": 894
    },
    {
      "epoch": 0.35075687062166266,
      "grad_norm": 6.551304817199707,
      "learning_rate": 7.090848663547574e-05,
      "loss": 2.2339,
      "step": 895
    },
    {
      "epoch": 0.3511487777396757,
      "grad_norm": 5.904609680175781,
      "learning_rate": 7.070682559695736e-05,
      "loss": 2.2098,
      "step": 896
    },
    {
      "epoch": 0.35154068485768875,
      "grad_norm": 6.565858364105225,
      "learning_rate": 7.050529478317476e-05,
      "loss": 2.1291,
      "step": 897
    },
    {
      "epoch": 0.35193259197570176,
      "grad_norm": 6.813206195831299,
      "learning_rate": 7.03038950900464e-05,
      "loss": 1.9179,
      "step": 898
    },
    {
      "epoch": 0.35232449909371477,
      "grad_norm": 7.764680862426758,
      "learning_rate": 7.010262741290798e-05,
      "loss": 2.3574,
      "step": 899
    },
    {
      "epoch": 0.35271640621172784,
      "grad_norm": 12.371960639953613,
      "learning_rate": 6.990149264650814e-05,
      "loss": 2.7073,
      "step": 900
    },
    {
      "epoch": 0.35310831332974085,
      "grad_norm": 2.034923791885376,
      "learning_rate": 6.970049168500474e-05,
      "loss": 1.1205,
      "step": 901
    },
    {
      "epoch": 0.35350022044775387,
      "grad_norm": 1.8922945261001587,
      "learning_rate": 6.94996254219608e-05,
      "loss": 0.5483,
      "step": 902
    },
    {
      "epoch": 0.35389212756576693,
      "grad_norm": 2.138000726699829,
      "learning_rate": 6.929889475034048e-05,
      "loss": 0.9304,
      "step": 903
    },
    {
      "epoch": 0.35428403468377995,
      "grad_norm": 1.5044962167739868,
      "learning_rate": 6.909830056250527e-05,
      "loss": 0.3099,
      "step": 904
    },
    {
      "epoch": 0.35467594180179296,
      "grad_norm": 2.5971896648406982,
      "learning_rate": 6.889784375020978e-05,
      "loss": 1.9511,
      "step": 905
    },
    {
      "epoch": 0.35506784891980603,
      "grad_norm": 2.27099609375,
      "learning_rate": 6.869752520459803e-05,
      "loss": 1.4911,
      "step": 906
    },
    {
      "epoch": 0.35545975603781904,
      "grad_norm": 2.719440221786499,
      "learning_rate": 6.849734581619918e-05,
      "loss": 1.8989,
      "step": 907
    },
    {
      "epoch": 0.35585166315583205,
      "grad_norm": 2.9784562587738037,
      "learning_rate": 6.829730647492404e-05,
      "loss": 1.9251,
      "step": 908
    },
    {
      "epoch": 0.3562435702738451,
      "grad_norm": 3.241880416870117,
      "learning_rate": 6.80974080700606e-05,
      "loss": 1.8662,
      "step": 909
    },
    {
      "epoch": 0.35663547739185814,
      "grad_norm": 3.218980550765991,
      "learning_rate": 6.789765149027039e-05,
      "loss": 1.4134,
      "step": 910
    },
    {
      "epoch": 0.35702738450987115,
      "grad_norm": 4.105949878692627,
      "learning_rate": 6.769803762358443e-05,
      "loss": 2.5505,
      "step": 911
    },
    {
      "epoch": 0.3574192916278842,
      "grad_norm": 3.315664768218994,
      "learning_rate": 6.749856735739928e-05,
      "loss": 1.9885,
      "step": 912
    },
    {
      "epoch": 0.35781119874589723,
      "grad_norm": 3.5657966136932373,
      "learning_rate": 6.729924157847323e-05,
      "loss": 2.1779,
      "step": 913
    },
    {
      "epoch": 0.35820310586391024,
      "grad_norm": 2.9721014499664307,
      "learning_rate": 6.710006117292209e-05,
      "loss": 1.4424,
      "step": 914
    },
    {
      "epoch": 0.3585950129819233,
      "grad_norm": 3.0443904399871826,
      "learning_rate": 6.690102702621548e-05,
      "loss": 1.4581,
      "step": 915
    },
    {
      "epoch": 0.3589869200999363,
      "grad_norm": 3.878185749053955,
      "learning_rate": 6.670214002317278e-05,
      "loss": 2.0977,
      "step": 916
    },
    {
      "epoch": 0.35937882721794934,
      "grad_norm": 3.0507938861846924,
      "learning_rate": 6.650340104795932e-05,
      "loss": 1.4856,
      "step": 917
    },
    {
      "epoch": 0.35977073433596235,
      "grad_norm": 3.4639039039611816,
      "learning_rate": 6.630481098408228e-05,
      "loss": 2.008,
      "step": 918
    },
    {
      "epoch": 0.3601626414539754,
      "grad_norm": 3.338740110397339,
      "learning_rate": 6.610637071438686e-05,
      "loss": 1.8478,
      "step": 919
    },
    {
      "epoch": 0.36055454857198843,
      "grad_norm": 3.97996187210083,
      "learning_rate": 6.590808112105232e-05,
      "loss": 2.1923,
      "step": 920
    },
    {
      "epoch": 0.36094645569000144,
      "grad_norm": 3.694767951965332,
      "learning_rate": 6.570994308558812e-05,
      "loss": 2.1046,
      "step": 921
    },
    {
      "epoch": 0.3613383628080145,
      "grad_norm": 3.7716007232666016,
      "learning_rate": 6.551195748882997e-05,
      "loss": 1.8605,
      "step": 922
    },
    {
      "epoch": 0.3617302699260275,
      "grad_norm": 3.9174575805664062,
      "learning_rate": 6.531412521093586e-05,
      "loss": 1.5939,
      "step": 923
    },
    {
      "epoch": 0.36212217704404054,
      "grad_norm": 4.4075798988342285,
      "learning_rate": 6.51164471313822e-05,
      "loss": 2.1031,
      "step": 924
    },
    {
      "epoch": 0.3625140841620536,
      "grad_norm": 3.6604557037353516,
      "learning_rate": 6.491892412895995e-05,
      "loss": 1.6585,
      "step": 925
    },
    {
      "epoch": 0.3629059912800666,
      "grad_norm": 4.246099472045898,
      "learning_rate": 6.472155708177052e-05,
      "loss": 1.8409,
      "step": 926
    },
    {
      "epoch": 0.36329789839807963,
      "grad_norm": 4.356520652770996,
      "learning_rate": 6.452434686722224e-05,
      "loss": 2.0161,
      "step": 927
    },
    {
      "epoch": 0.3636898055160927,
      "grad_norm": 4.544940948486328,
      "learning_rate": 6.432729436202604e-05,
      "loss": 1.7202,
      "step": 928
    },
    {
      "epoch": 0.3640817126341057,
      "grad_norm": 4.730035305023193,
      "learning_rate": 6.41304004421918e-05,
      "loss": 2.1217,
      "step": 929
    },
    {
      "epoch": 0.3644736197521187,
      "grad_norm": 5.090596675872803,
      "learning_rate": 6.393366598302446e-05,
      "loss": 2.1715,
      "step": 930
    },
    {
      "epoch": 0.3648655268701318,
      "grad_norm": 4.58132266998291,
      "learning_rate": 6.373709185911998e-05,
      "loss": 2.0932,
      "step": 931
    },
    {
      "epoch": 0.3652574339881448,
      "grad_norm": 4.3024001121521,
      "learning_rate": 6.354067894436155e-05,
      "loss": 1.7598,
      "step": 932
    },
    {
      "epoch": 0.3656493411061578,
      "grad_norm": 4.19694185256958,
      "learning_rate": 6.334442811191576e-05,
      "loss": 1.7389,
      "step": 933
    },
    {
      "epoch": 0.3660412482241709,
      "grad_norm": 4.823878288269043,
      "learning_rate": 6.314834023422858e-05,
      "loss": 2.4373,
      "step": 934
    },
    {
      "epoch": 0.3664331553421839,
      "grad_norm": 5.063382148742676,
      "learning_rate": 6.295241618302156e-05,
      "loss": 2.3053,
      "step": 935
    },
    {
      "epoch": 0.3668250624601969,
      "grad_norm": 5.721461772918701,
      "learning_rate": 6.275665682928803e-05,
      "loss": 2.5877,
      "step": 936
    },
    {
      "epoch": 0.36721696957821,
      "grad_norm": 4.3630290031433105,
      "learning_rate": 6.256106304328905e-05,
      "loss": 1.7277,
      "step": 937
    },
    {
      "epoch": 0.367608876696223,
      "grad_norm": 4.9136176109313965,
      "learning_rate": 6.23656356945497e-05,
      "loss": 2.0431,
      "step": 938
    },
    {
      "epoch": 0.368000783814236,
      "grad_norm": 4.496310710906982,
      "learning_rate": 6.21703756518551e-05,
      "loss": 1.7988,
      "step": 939
    },
    {
      "epoch": 0.3683926909322491,
      "grad_norm": 5.020188808441162,
      "learning_rate": 6.197528378324665e-05,
      "loss": 2.101,
      "step": 940
    },
    {
      "epoch": 0.3687845980502621,
      "grad_norm": 4.8069024085998535,
      "learning_rate": 6.17803609560181e-05,
      "loss": 1.7885,
      "step": 941
    },
    {
      "epoch": 0.3691765051682751,
      "grad_norm": 5.215355396270752,
      "learning_rate": 6.158560803671168e-05,
      "loss": 2.011,
      "step": 942
    },
    {
      "epoch": 0.3695684122862882,
      "grad_norm": 4.99086856842041,
      "learning_rate": 6.139102589111435e-05,
      "loss": 2.056,
      "step": 943
    },
    {
      "epoch": 0.3699603194043012,
      "grad_norm": 5.7369208335876465,
      "learning_rate": 6.119661538425381e-05,
      "loss": 1.979,
      "step": 944
    },
    {
      "epoch": 0.3703522265223142,
      "grad_norm": 6.0851969718933105,
      "learning_rate": 6.100237738039484e-05,
      "loss": 2.321,
      "step": 945
    },
    {
      "epoch": 0.37074413364032727,
      "grad_norm": 6.86392879486084,
      "learning_rate": 6.0808312743035236e-05,
      "loss": 2.2877,
      "step": 946
    },
    {
      "epoch": 0.3711360407583403,
      "grad_norm": 7.688723564147949,
      "learning_rate": 6.061442233490211e-05,
      "loss": 2.2824,
      "step": 947
    },
    {
      "epoch": 0.3715279478763533,
      "grad_norm": 7.421009063720703,
      "learning_rate": 6.042070701794806e-05,
      "loss": 1.9596,
      "step": 948
    },
    {
      "epoch": 0.37191985499436636,
      "grad_norm": 8.9202880859375,
      "learning_rate": 6.0227167653347305e-05,
      "loss": 2.1535,
      "step": 949
    },
    {
      "epoch": 0.3723117621123794,
      "grad_norm": 14.531527519226074,
      "learning_rate": 6.0033805101491794e-05,
      "loss": 2.7632,
      "step": 950
    },
    {
      "epoch": 0.3727036692303924,
      "grad_norm": 1.4576749801635742,
      "learning_rate": 5.98406202219875e-05,
      "loss": 0.5843,
      "step": 951
    },
    {
      "epoch": 0.37309557634840546,
      "grad_norm": 1.6230075359344482,
      "learning_rate": 5.964761387365052e-05,
      "loss": 0.7447,
      "step": 952
    },
    {
      "epoch": 0.37348748346641847,
      "grad_norm": 1.865219235420227,
      "learning_rate": 5.9454786914503255e-05,
      "loss": 1.0828,
      "step": 953
    },
    {
      "epoch": 0.3738793905844315,
      "grad_norm": 1.6923391819000244,
      "learning_rate": 5.926214020177074e-05,
      "loss": 0.6706,
      "step": 954
    },
    {
      "epoch": 0.3742712977024445,
      "grad_norm": 2.008579730987549,
      "learning_rate": 5.9069674591876534e-05,
      "loss": 1.7347,
      "step": 955
    },
    {
      "epoch": 0.37466320482045756,
      "grad_norm": 2.2702107429504395,
      "learning_rate": 5.887739094043923e-05,
      "loss": 1.764,
      "step": 956
    },
    {
      "epoch": 0.3750551119384706,
      "grad_norm": 2.622749090194702,
      "learning_rate": 5.868529010226845e-05,
      "loss": 1.8989,
      "step": 957
    },
    {
      "epoch": 0.3754470190564836,
      "grad_norm": 2.871635675430298,
      "learning_rate": 5.849337293136112e-05,
      "loss": 2.2221,
      "step": 958
    },
    {
      "epoch": 0.37583892617449666,
      "grad_norm": 3.057769536972046,
      "learning_rate": 5.830164028089766e-05,
      "loss": 1.8093,
      "step": 959
    },
    {
      "epoch": 0.37623083329250967,
      "grad_norm": 3.046288013458252,
      "learning_rate": 5.811009300323818e-05,
      "loss": 1.7718,
      "step": 960
    },
    {
      "epoch": 0.3766227404105227,
      "grad_norm": 3.5015342235565186,
      "learning_rate": 5.791873194991872e-05,
      "loss": 2.5153,
      "step": 961
    },
    {
      "epoch": 0.37701464752853575,
      "grad_norm": 3.28128981590271,
      "learning_rate": 5.7727557971647427e-05,
      "loss": 2.0167,
      "step": 962
    },
    {
      "epoch": 0.37740655464654876,
      "grad_norm": 3.118090867996216,
      "learning_rate": 5.7536571918300864e-05,
      "loss": 1.848,
      "step": 963
    },
    {
      "epoch": 0.3777984617645618,
      "grad_norm": 3.401566982269287,
      "learning_rate": 5.734577463892008e-05,
      "loss": 2.0362,
      "step": 964
    },
    {
      "epoch": 0.37819036888257485,
      "grad_norm": 3.181457281112671,
      "learning_rate": 5.7155166981706956e-05,
      "loss": 1.6719,
      "step": 965
    },
    {
      "epoch": 0.37858227600058786,
      "grad_norm": 3.3688788414001465,
      "learning_rate": 5.6964749794020354e-05,
      "loss": 1.9336,
      "step": 966
    },
    {
      "epoch": 0.37897418311860087,
      "grad_norm": 3.695744037628174,
      "learning_rate": 5.6774523922372394e-05,
      "loss": 1.9646,
      "step": 967
    },
    {
      "epoch": 0.37936609023661394,
      "grad_norm": 3.42696213722229,
      "learning_rate": 5.6584490212424804e-05,
      "loss": 1.8824,
      "step": 968
    },
    {
      "epoch": 0.37975799735462695,
      "grad_norm": 3.895899772644043,
      "learning_rate": 5.639464950898491e-05,
      "loss": 1.9196,
      "step": 969
    },
    {
      "epoch": 0.38014990447263997,
      "grad_norm": 3.546597480773926,
      "learning_rate": 5.620500265600206e-05,
      "loss": 1.5168,
      "step": 970
    },
    {
      "epoch": 0.38054181159065303,
      "grad_norm": 4.170307636260986,
      "learning_rate": 5.601555049656382e-05,
      "loss": 2.1263,
      "step": 971
    },
    {
      "epoch": 0.38093371870866605,
      "grad_norm": 4.319623947143555,
      "learning_rate": 5.58262938728922e-05,
      "loss": 1.9966,
      "step": 972
    },
    {
      "epoch": 0.38132562582667906,
      "grad_norm": 3.933392286300659,
      "learning_rate": 5.563723362634008e-05,
      "loss": 1.8882,
      "step": 973
    },
    {
      "epoch": 0.38171753294469213,
      "grad_norm": 4.4646315574646,
      "learning_rate": 5.544837059738719e-05,
      "loss": 2.5457,
      "step": 974
    },
    {
      "epoch": 0.38210944006270514,
      "grad_norm": 4.428443431854248,
      "learning_rate": 5.525970562563656e-05,
      "loss": 1.9738,
      "step": 975
    },
    {
      "epoch": 0.38250134718071815,
      "grad_norm": 3.992448329925537,
      "learning_rate": 5.507123954981073e-05,
      "loss": 1.765,
      "step": 976
    },
    {
      "epoch": 0.3828932542987312,
      "grad_norm": 4.151793479919434,
      "learning_rate": 5.488297320774807e-05,
      "loss": 1.805,
      "step": 977
    },
    {
      "epoch": 0.38328516141674424,
      "grad_norm": 4.187563419342041,
      "learning_rate": 5.4694907436399e-05,
      "loss": 1.9378,
      "step": 978
    },
    {
      "epoch": 0.38367706853475725,
      "grad_norm": 4.464833736419678,
      "learning_rate": 5.4507043071822284e-05,
      "loss": 1.8628,
      "step": 979
    },
    {
      "epoch": 0.3840689756527703,
      "grad_norm": 4.567963123321533,
      "learning_rate": 5.431938094918132e-05,
      "loss": 2.1188,
      "step": 980
    },
    {
      "epoch": 0.38446088277078333,
      "grad_norm": 4.461292266845703,
      "learning_rate": 5.41319219027404e-05,
      "loss": 1.9218,
      "step": 981
    },
    {
      "epoch": 0.38485278988879634,
      "grad_norm": 4.801513671875,
      "learning_rate": 5.394466676586114e-05,
      "loss": 2.3581,
      "step": 982
    },
    {
      "epoch": 0.3852446970068094,
      "grad_norm": 4.623222351074219,
      "learning_rate": 5.375761637099854e-05,
      "loss": 2.0619,
      "step": 983
    },
    {
      "epoch": 0.3856366041248224,
      "grad_norm": 4.8148627281188965,
      "learning_rate": 5.357077154969742e-05,
      "loss": 1.9045,
      "step": 984
    },
    {
      "epoch": 0.38602851124283544,
      "grad_norm": 4.8767828941345215,
      "learning_rate": 5.3384133132588784e-05,
      "loss": 2.0655,
      "step": 985
    },
    {
      "epoch": 0.3864204183608485,
      "grad_norm": 4.712939739227295,
      "learning_rate": 5.3197701949386e-05,
      "loss": 2.2083,
      "step": 986
    },
    {
      "epoch": 0.3868123254788615,
      "grad_norm": 4.615323543548584,
      "learning_rate": 5.301147882888116e-05,
      "loss": 1.7824,
      "step": 987
    },
    {
      "epoch": 0.38720423259687453,
      "grad_norm": 4.9034504890441895,
      "learning_rate": 5.28254645989414e-05,
      "loss": 2.0276,
      "step": 988
    },
    {
      "epoch": 0.38759613971488754,
      "grad_norm": 4.976192474365234,
      "learning_rate": 5.2639660086505226e-05,
      "loss": 2.0001,
      "step": 989
    },
    {
      "epoch": 0.3879880468329006,
      "grad_norm": 5.0580363273620605,
      "learning_rate": 5.2454066117578815e-05,
      "loss": 2.346,
      "step": 990
    },
    {
      "epoch": 0.3883799539509136,
      "grad_norm": 5.31488037109375,
      "learning_rate": 5.226868351723244e-05,
      "loss": 2.4048,
      "step": 991
    },
    {
      "epoch": 0.38877186106892664,
      "grad_norm": 5.452157974243164,
      "learning_rate": 5.2083513109596616e-05,
      "loss": 2.0133,
      "step": 992
    },
    {
      "epoch": 0.3891637681869397,
      "grad_norm": 5.4536051750183105,
      "learning_rate": 5.189855571785859e-05,
      "loss": 2.2337,
      "step": 993
    },
    {
      "epoch": 0.3895556753049527,
      "grad_norm": 5.422115325927734,
      "learning_rate": 5.171381216425863e-05,
      "loss": 2.0556,
      "step": 994
    },
    {
      "epoch": 0.38994758242296573,
      "grad_norm": 5.533926963806152,
      "learning_rate": 5.152928327008635e-05,
      "loss": 1.9221,
      "step": 995
    },
    {
      "epoch": 0.3903394895409788,
      "grad_norm": 6.175473213195801,
      "learning_rate": 5.134496985567714e-05,
      "loss": 1.9078,
      "step": 996
    },
    {
      "epoch": 0.3907313966589918,
      "grad_norm": 6.327174186706543,
      "learning_rate": 5.116087274040837e-05,
      "loss": 1.5398,
      "step": 997
    },
    {
      "epoch": 0.3911233037770048,
      "grad_norm": 7.2678327560424805,
      "learning_rate": 5.0976992742695925e-05,
      "loss": 2.5995,
      "step": 998
    },
    {
      "epoch": 0.3915152108950179,
      "grad_norm": 9.469869613647461,
      "learning_rate": 5.07933306799904e-05,
      "loss": 2.3089,
      "step": 999
    },
    {
      "epoch": 0.3919071180130309,
      "grad_norm": 10.412221908569336,
      "learning_rate": 5.060988736877366e-05,
      "loss": 1.9305,
      "step": 1000
    },
    {
      "epoch": 0.3922990251310439,
      "grad_norm": 1.6813195943832397,
      "learning_rate": 5.042666362455498e-05,
      "loss": 1.2189,
      "step": 1001
    },
    {
      "epoch": 0.392690932249057,
      "grad_norm": 1.5347741842269897,
      "learning_rate": 5.024366026186755e-05,
      "loss": 0.3558,
      "step": 1002
    },
    {
      "epoch": 0.39308283936707,
      "grad_norm": 1.48504638671875,
      "learning_rate": 5.006087809426496e-05,
      "loss": 0.4201,
      "step": 1003
    },
    {
      "epoch": 0.393474746485083,
      "grad_norm": 2.156648635864258,
      "learning_rate": 4.987831793431731e-05,
      "loss": 1.7416,
      "step": 1004
    },
    {
      "epoch": 0.3938666536030961,
      "grad_norm": 2.477829694747925,
      "learning_rate": 4.9695980593607817e-05,
      "loss": 1.9278,
      "step": 1005
    },
    {
      "epoch": 0.3942585607211091,
      "grad_norm": 2.1177494525909424,
      "learning_rate": 4.9513866882729146e-05,
      "loss": 1.3218,
      "step": 1006
    },
    {
      "epoch": 0.3946504678391221,
      "grad_norm": 2.2638659477233887,
      "learning_rate": 4.9331977611279777e-05,
      "loss": 1.2143,
      "step": 1007
    },
    {
      "epoch": 0.3950423749571352,
      "grad_norm": 3.0279455184936523,
      "learning_rate": 4.9150313587860433e-05,
      "loss": 1.5786,
      "step": 1008
    },
    {
      "epoch": 0.3954342820751482,
      "grad_norm": 2.515331983566284,
      "learning_rate": 4.896887562007054e-05,
      "loss": 1.3618,
      "step": 1009
    },
    {
      "epoch": 0.3958261891931612,
      "grad_norm": 3.637824535369873,
      "learning_rate": 4.8787664514504504e-05,
      "loss": 2.0378,
      "step": 1010
    },
    {
      "epoch": 0.3962180963111743,
      "grad_norm": 3.0306451320648193,
      "learning_rate": 4.860668107674823e-05,
      "loss": 1.7247,
      "step": 1011
    },
    {
      "epoch": 0.3966100034291873,
      "grad_norm": 3.4333624839782715,
      "learning_rate": 4.8425926111375506e-05,
      "loss": 1.7455,
      "step": 1012
    },
    {
      "epoch": 0.3970019105472003,
      "grad_norm": 3.335279941558838,
      "learning_rate": 4.824540042194443e-05,
      "loss": 1.8399,
      "step": 1013
    },
    {
      "epoch": 0.39739381766521337,
      "grad_norm": 3.347822904586792,
      "learning_rate": 4.8065104810993856e-05,
      "loss": 2.1683,
      "step": 1014
    },
    {
      "epoch": 0.3977857247832264,
      "grad_norm": 4.273656845092773,
      "learning_rate": 4.788504008003978e-05,
      "loss": 1.8233,
      "step": 1015
    },
    {
      "epoch": 0.3981776319012394,
      "grad_norm": 3.716404676437378,
      "learning_rate": 4.770520702957182e-05,
      "loss": 2.0898,
      "step": 1016
    },
    {
      "epoch": 0.39856953901925246,
      "grad_norm": 3.41519832611084,
      "learning_rate": 4.752560645904962e-05,
      "loss": 1.5607,
      "step": 1017
    },
    {
      "epoch": 0.3989614461372655,
      "grad_norm": 3.6369516849517822,
      "learning_rate": 4.734623916689941e-05,
      "loss": 1.8506,
      "step": 1018
    },
    {
      "epoch": 0.3993533532552785,
      "grad_norm": 3.7849838733673096,
      "learning_rate": 4.716710595051022e-05,
      "loss": 2.1756,
      "step": 1019
    },
    {
      "epoch": 0.39974526037329156,
      "grad_norm": 3.8635830879211426,
      "learning_rate": 4.698820760623064e-05,
      "loss": 1.7232,
      "step": 1020
    },
    {
      "epoch": 0.40013716749130457,
      "grad_norm": 3.8572263717651367,
      "learning_rate": 4.6809544929365004e-05,
      "loss": 2.4024,
      "step": 1021
    },
    {
      "epoch": 0.4005290746093176,
      "grad_norm": 3.6128010749816895,
      "learning_rate": 4.663111871417e-05,
      "loss": 1.609,
      "step": 1022
    },
    {
      "epoch": 0.40092098172733065,
      "grad_norm": 3.8890762329101562,
      "learning_rate": 4.645292975385111e-05,
      "loss": 2.0119,
      "step": 1023
    },
    {
      "epoch": 0.40131288884534366,
      "grad_norm": 4.021796226501465,
      "learning_rate": 4.627497884055912e-05,
      "loss": 1.6486,
      "step": 1024
    },
    {
      "epoch": 0.4017047959633567,
      "grad_norm": 4.3973894119262695,
      "learning_rate": 4.609726676538652e-05,
      "loss": 2.1021,
      "step": 1025
    },
    {
      "epoch": 0.4020967030813697,
      "grad_norm": 3.8529930114746094,
      "learning_rate": 4.591979431836402e-05,
      "loss": 1.8995,
      "step": 1026
    },
    {
      "epoch": 0.40248861019938276,
      "grad_norm": 4.095380783081055,
      "learning_rate": 4.574256228845706e-05,
      "loss": 1.8838,
      "step": 1027
    },
    {
      "epoch": 0.40288051731739577,
      "grad_norm": 4.439388275146484,
      "learning_rate": 4.5565571463562365e-05,
      "loss": 2.3249,
      "step": 1028
    },
    {
      "epoch": 0.4032724244354088,
      "grad_norm": 4.336061477661133,
      "learning_rate": 4.5388822630504256e-05,
      "loss": 2.0326,
      "step": 1029
    },
    {
      "epoch": 0.40366433155342185,
      "grad_norm": 4.670348167419434,
      "learning_rate": 4.521231657503132e-05,
      "loss": 2.3664,
      "step": 1030
    },
    {
      "epoch": 0.40405623867143486,
      "grad_norm": 4.394977569580078,
      "learning_rate": 4.503605408181286e-05,
      "loss": 2.2275,
      "step": 1031
    },
    {
      "epoch": 0.4044481457894479,
      "grad_norm": 4.878340244293213,
      "learning_rate": 4.486003593443537e-05,
      "loss": 2.2219,
      "step": 1032
    },
    {
      "epoch": 0.40484005290746095,
      "grad_norm": 5.087642669677734,
      "learning_rate": 4.468426291539914e-05,
      "loss": 2.1393,
      "step": 1033
    },
    {
      "epoch": 0.40523196002547396,
      "grad_norm": 5.026920318603516,
      "learning_rate": 4.4508735806114654e-05,
      "loss": 2.2734,
      "step": 1034
    },
    {
      "epoch": 0.40562386714348697,
      "grad_norm": 4.480007648468018,
      "learning_rate": 4.433345538689929e-05,
      "loss": 2.1809,
      "step": 1035
    },
    {
      "epoch": 0.40601577426150004,
      "grad_norm": 4.747096538543701,
      "learning_rate": 4.415842243697369e-05,
      "loss": 2.3178,
      "step": 1036
    },
    {
      "epoch": 0.40640768137951305,
      "grad_norm": 4.795892715454102,
      "learning_rate": 4.39836377344583e-05,
      "loss": 2.1743,
      "step": 1037
    },
    {
      "epoch": 0.40679958849752607,
      "grad_norm": 4.661589622497559,
      "learning_rate": 4.380910205637012e-05,
      "loss": 1.7175,
      "step": 1038
    },
    {
      "epoch": 0.40719149561553913,
      "grad_norm": 4.601487159729004,
      "learning_rate": 4.363481617861893e-05,
      "loss": 1.8506,
      "step": 1039
    },
    {
      "epoch": 0.40758340273355215,
      "grad_norm": 4.862745761871338,
      "learning_rate": 4.346078087600412e-05,
      "loss": 1.9588,
      "step": 1040
    },
    {
      "epoch": 0.40797530985156516,
      "grad_norm": 5.128295421600342,
      "learning_rate": 4.3286996922211034e-05,
      "loss": 1.9782,
      "step": 1041
    },
    {
      "epoch": 0.40836721696957823,
      "grad_norm": 5.889504432678223,
      "learning_rate": 4.311346508980772e-05,
      "loss": 2.4801,
      "step": 1042
    },
    {
      "epoch": 0.40875912408759124,
      "grad_norm": 4.718349456787109,
      "learning_rate": 4.2940186150241365e-05,
      "loss": 1.6892,
      "step": 1043
    },
    {
      "epoch": 0.40915103120560425,
      "grad_norm": 5.4573869705200195,
      "learning_rate": 4.27671608738349e-05,
      "loss": 1.7853,
      "step": 1044
    },
    {
      "epoch": 0.4095429383236173,
      "grad_norm": 5.859316349029541,
      "learning_rate": 4.2594390029783534e-05,
      "loss": 1.5707,
      "step": 1045
    },
    {
      "epoch": 0.40993484544163034,
      "grad_norm": 5.9655256271362305,
      "learning_rate": 4.242187438615153e-05,
      "loss": 1.8441,
      "step": 1046
    },
    {
      "epoch": 0.41032675255964335,
      "grad_norm": 7.081624507904053,
      "learning_rate": 4.224961470986849e-05,
      "loss": 1.8047,
      "step": 1047
    },
    {
      "epoch": 0.4107186596776564,
      "grad_norm": 7.673589706420898,
      "learning_rate": 4.207761176672614e-05,
      "loss": 2.2996,
      "step": 1048
    },
    {
      "epoch": 0.41111056679566943,
      "grad_norm": 9.99533748626709,
      "learning_rate": 4.190586632137491e-05,
      "loss": 2.0573,
      "step": 1049
    },
    {
      "epoch": 0.41150247391368244,
      "grad_norm": 12.307229042053223,
      "learning_rate": 4.173437913732048e-05,
      "loss": 2.1191,
      "step": 1050
    },
    {
      "epoch": 0.4118943810316955,
      "grad_norm": 1.3487191200256348,
      "learning_rate": 4.156315097692037e-05,
      "loss": 0.9482,
      "step": 1051
    },
    {
      "epoch": 0.4122862881497085,
      "grad_norm": 1.3746885061264038,
      "learning_rate": 4.139218260138074e-05,
      "loss": 0.6559,
      "step": 1052
    },
    {
      "epoch": 0.41267819526772154,
      "grad_norm": 1.5744364261627197,
      "learning_rate": 4.12214747707527e-05,
      "loss": 0.6672,
      "step": 1053
    },
    {
      "epoch": 0.4130701023857346,
      "grad_norm": 2.52003812789917,
      "learning_rate": 4.1051028243929125e-05,
      "loss": 1.8736,
      "step": 1054
    },
    {
      "epoch": 0.4134620095037476,
      "grad_norm": 2.237959623336792,
      "learning_rate": 4.088084377864135e-05,
      "loss": 1.3843,
      "step": 1055
    },
    {
      "epoch": 0.41385391662176063,
      "grad_norm": 2.7332470417022705,
      "learning_rate": 4.07109221314556e-05,
      "loss": 1.3759,
      "step": 1056
    },
    {
      "epoch": 0.4142458237397737,
      "grad_norm": 2.6395058631896973,
      "learning_rate": 4.054126405776971e-05,
      "loss": 1.5625,
      "step": 1057
    },
    {
      "epoch": 0.4146377308577867,
      "grad_norm": 2.459700584411621,
      "learning_rate": 4.037187031180985e-05,
      "loss": 1.1902,
      "step": 1058
    },
    {
      "epoch": 0.4150296379757997,
      "grad_norm": 2.8403749465942383,
      "learning_rate": 4.020274164662707e-05,
      "loss": 1.8714,
      "step": 1059
    },
    {
      "epoch": 0.4154215450938128,
      "grad_norm": 3.7568366527557373,
      "learning_rate": 4.003387881409397e-05,
      "loss": 1.9077,
      "step": 1060
    },
    {
      "epoch": 0.4158134522118258,
      "grad_norm": 3.250598192214966,
      "learning_rate": 3.986528256490141e-05,
      "loss": 1.7914,
      "step": 1061
    },
    {
      "epoch": 0.4162053593298388,
      "grad_norm": 2.9566619396209717,
      "learning_rate": 3.969695364855511e-05,
      "loss": 1.6894,
      "step": 1062
    },
    {
      "epoch": 0.41659726644785183,
      "grad_norm": 3.480721950531006,
      "learning_rate": 3.952889281337235e-05,
      "loss": 2.0458,
      "step": 1063
    },
    {
      "epoch": 0.4169891735658649,
      "grad_norm": 3.2218005657196045,
      "learning_rate": 3.93611008064786e-05,
      "loss": 2.1689,
      "step": 1064
    },
    {
      "epoch": 0.4173810806838779,
      "grad_norm": 2.9976806640625,
      "learning_rate": 3.9193578373804364e-05,
      "loss": 1.523,
      "step": 1065
    },
    {
      "epoch": 0.4177729878018909,
      "grad_norm": 3.7460598945617676,
      "learning_rate": 3.90263262600816e-05,
      "loss": 2.0017,
      "step": 1066
    },
    {
      "epoch": 0.418164894919904,
      "grad_norm": 4.384772300720215,
      "learning_rate": 3.88593452088406e-05,
      "loss": 2.2635,
      "step": 1067
    },
    {
      "epoch": 0.418556802037917,
      "grad_norm": 4.028706073760986,
      "learning_rate": 3.869263596240661e-05,
      "loss": 2.093,
      "step": 1068
    },
    {
      "epoch": 0.41894870915593,
      "grad_norm": 3.653794765472412,
      "learning_rate": 3.8526199261896544e-05,
      "loss": 1.468,
      "step": 1069
    },
    {
      "epoch": 0.4193406162739431,
      "grad_norm": 3.621711254119873,
      "learning_rate": 3.836003584721577e-05,
      "loss": 1.6105,
      "step": 1070
    },
    {
      "epoch": 0.4197325233919561,
      "grad_norm": 4.032018184661865,
      "learning_rate": 3.8194146457054655e-05,
      "loss": 1.8875,
      "step": 1071
    },
    {
      "epoch": 0.4201244305099691,
      "grad_norm": 3.3381333351135254,
      "learning_rate": 3.802853182888543e-05,
      "loss": 1.3083,
      "step": 1072
    },
    {
      "epoch": 0.4205163376279822,
      "grad_norm": 4.465856075286865,
      "learning_rate": 3.786319269895877e-05,
      "loss": 2.3169,
      "step": 1073
    },
    {
      "epoch": 0.4209082447459952,
      "grad_norm": 4.420526027679443,
      "learning_rate": 3.769812980230074e-05,
      "loss": 1.807,
      "step": 1074
    },
    {
      "epoch": 0.4213001518640082,
      "grad_norm": 4.535737037658691,
      "learning_rate": 3.7533343872709294e-05,
      "loss": 1.6755,
      "step": 1075
    },
    {
      "epoch": 0.4216920589820213,
      "grad_norm": 5.291508197784424,
      "learning_rate": 3.736883564275112e-05,
      "loss": 2.3512,
      "step": 1076
    },
    {
      "epoch": 0.4220839661000343,
      "grad_norm": 4.205623149871826,
      "learning_rate": 3.7204605843758386e-05,
      "loss": 1.6356,
      "step": 1077
    },
    {
      "epoch": 0.4224758732180473,
      "grad_norm": 4.722237586975098,
      "learning_rate": 3.704065520582549e-05,
      "loss": 2.5431,
      "step": 1078
    },
    {
      "epoch": 0.4228677803360604,
      "grad_norm": 4.8755879402160645,
      "learning_rate": 3.6876984457805786e-05,
      "loss": 2.2964,
      "step": 1079
    },
    {
      "epoch": 0.4232596874540734,
      "grad_norm": 4.938620567321777,
      "learning_rate": 3.671359432730834e-05,
      "loss": 2.1839,
      "step": 1080
    },
    {
      "epoch": 0.4236515945720864,
      "grad_norm": 4.563158988952637,
      "learning_rate": 3.655048554069478e-05,
      "loss": 1.9773,
      "step": 1081
    },
    {
      "epoch": 0.42404350169009947,
      "grad_norm": 4.686749458312988,
      "learning_rate": 3.638765882307589e-05,
      "loss": 1.9329,
      "step": 1082
    },
    {
      "epoch": 0.4244354088081125,
      "grad_norm": 4.811989784240723,
      "learning_rate": 3.6225114898308634e-05,
      "loss": 2.1335,
      "step": 1083
    },
    {
      "epoch": 0.4248273159261255,
      "grad_norm": 4.912534713745117,
      "learning_rate": 3.6062854488992714e-05,
      "loss": 1.8844,
      "step": 1084
    },
    {
      "epoch": 0.42521922304413856,
      "grad_norm": 4.940800189971924,
      "learning_rate": 3.5900878316467454e-05,
      "loss": 1.9888,
      "step": 1085
    },
    {
      "epoch": 0.4256111301621516,
      "grad_norm": 5.001417636871338,
      "learning_rate": 3.573918710080857e-05,
      "loss": 2.0879,
      "step": 1086
    },
    {
      "epoch": 0.4260030372801646,
      "grad_norm": 5.342638969421387,
      "learning_rate": 3.5577781560825066e-05,
      "loss": 2.3173,
      "step": 1087
    },
    {
      "epoch": 0.42639494439817766,
      "grad_norm": 5.021146297454834,
      "learning_rate": 3.541666241405588e-05,
      "loss": 2.4,
      "step": 1088
    },
    {
      "epoch": 0.42678685151619067,
      "grad_norm": 5.218502521514893,
      "learning_rate": 3.5255830376766764e-05,
      "loss": 2.0347,
      "step": 1089
    },
    {
      "epoch": 0.4271787586342037,
      "grad_norm": 4.585696697235107,
      "learning_rate": 3.509528616394716e-05,
      "loss": 1.7006,
      "step": 1090
    },
    {
      "epoch": 0.42757066575221675,
      "grad_norm": 4.883055210113525,
      "learning_rate": 3.4935030489306883e-05,
      "loss": 1.6439,
      "step": 1091
    },
    {
      "epoch": 0.42796257287022976,
      "grad_norm": 4.9321441650390625,
      "learning_rate": 3.4775064065273165e-05,
      "loss": 1.7053,
      "step": 1092
    },
    {
      "epoch": 0.4283544799882428,
      "grad_norm": 5.062867164611816,
      "learning_rate": 3.4615387602987236e-05,
      "loss": 1.8006,
      "step": 1093
    },
    {
      "epoch": 0.42874638710625584,
      "grad_norm": 6.410160064697266,
      "learning_rate": 3.445600181230134e-05,
      "loss": 2.0275,
      "step": 1094
    },
    {
      "epoch": 0.42913829422426886,
      "grad_norm": 5.644032001495361,
      "learning_rate": 3.429690740177549e-05,
      "loss": 1.7195,
      "step": 1095
    },
    {
      "epoch": 0.42953020134228187,
      "grad_norm": 5.906645774841309,
      "learning_rate": 3.413810507867436e-05,
      "loss": 1.7681,
      "step": 1096
    },
    {
      "epoch": 0.4299221084602949,
      "grad_norm": 6.4579901695251465,
      "learning_rate": 3.397959554896415e-05,
      "loss": 1.9511,
      "step": 1097
    },
    {
      "epoch": 0.43031401557830795,
      "grad_norm": 8.30800724029541,
      "learning_rate": 3.3821379517309405e-05,
      "loss": 1.9165,
      "step": 1098
    },
    {
      "epoch": 0.43070592269632096,
      "grad_norm": 10.596285820007324,
      "learning_rate": 3.3663457687069924e-05,
      "loss": 2.4981,
      "step": 1099
    },
    {
      "epoch": 0.431097829814334,
      "grad_norm": 11.342329025268555,
      "learning_rate": 3.350583076029754e-05,
      "loss": 2.2741,
      "step": 1100
    },
    {
      "epoch": 0.43148973693234705,
      "grad_norm": 1.2523295879364014,
      "learning_rate": 3.334849943773323e-05,
      "loss": 0.6043,
      "step": 1101
    },
    {
      "epoch": 0.43188164405036006,
      "grad_norm": 1.139450192451477,
      "learning_rate": 3.319146441880371e-05,
      "loss": 0.2605,
      "step": 1102
    },
    {
      "epoch": 0.43227355116837307,
      "grad_norm": 1.557660460472107,
      "learning_rate": 3.3034726401618444e-05,
      "loss": 0.6299,
      "step": 1103
    },
    {
      "epoch": 0.43266545828638614,
      "grad_norm": 1.7114484310150146,
      "learning_rate": 3.28782860829667e-05,
      "loss": 1.2743,
      "step": 1104
    },
    {
      "epoch": 0.43305736540439915,
      "grad_norm": 1.475270390510559,
      "learning_rate": 3.272214415831418e-05,
      "loss": 0.593,
      "step": 1105
    },
    {
      "epoch": 0.43344927252241217,
      "grad_norm": 2.4480137825012207,
      "learning_rate": 3.2566301321800085e-05,
      "loss": 1.5764,
      "step": 1106
    },
    {
      "epoch": 0.43384117964042523,
      "grad_norm": 2.541585922241211,
      "learning_rate": 3.241075826623401e-05,
      "loss": 1.7128,
      "step": 1107
    },
    {
      "epoch": 0.43423308675843825,
      "grad_norm": 2.6549935340881348,
      "learning_rate": 3.225551568309284e-05,
      "loss": 1.5962,
      "step": 1108
    },
    {
      "epoch": 0.43462499387645126,
      "grad_norm": 2.881373882293701,
      "learning_rate": 3.210057426251773e-05,
      "loss": 1.6607,
      "step": 1109
    },
    {
      "epoch": 0.43501690099446433,
      "grad_norm": 2.859070062637329,
      "learning_rate": 3.1945934693310896e-05,
      "loss": 1.7831,
      "step": 1110
    },
    {
      "epoch": 0.43540880811247734,
      "grad_norm": 3.224935531616211,
      "learning_rate": 3.179159766293282e-05,
      "loss": 1.9259,
      "step": 1111
    },
    {
      "epoch": 0.43580071523049035,
      "grad_norm": 3.399707555770874,
      "learning_rate": 3.163756385749889e-05,
      "loss": 2.0553,
      "step": 1112
    },
    {
      "epoch": 0.4361926223485034,
      "grad_norm": 2.879889965057373,
      "learning_rate": 3.148383396177653e-05,
      "loss": 1.7597,
      "step": 1113
    },
    {
      "epoch": 0.43658452946651644,
      "grad_norm": 3.7083587646484375,
      "learning_rate": 3.133040865918213e-05,
      "loss": 2.1358,
      "step": 1114
    },
    {
      "epoch": 0.43697643658452945,
      "grad_norm": 3.7487943172454834,
      "learning_rate": 3.117728863177796e-05,
      "loss": 2.0655,
      "step": 1115
    },
    {
      "epoch": 0.4373683437025425,
      "grad_norm": 3.6311116218566895,
      "learning_rate": 3.102447456026919e-05,
      "loss": 2.0885,
      "step": 1116
    },
    {
      "epoch": 0.43776025082055553,
      "grad_norm": 3.673100471496582,
      "learning_rate": 3.0871967124000834e-05,
      "loss": 1.8562,
      "step": 1117
    },
    {
      "epoch": 0.43815215793856854,
      "grad_norm": 3.396204948425293,
      "learning_rate": 3.0719767000954714e-05,
      "loss": 1.7412,
      "step": 1118
    },
    {
      "epoch": 0.4385440650565816,
      "grad_norm": 4.07634973526001,
      "learning_rate": 3.056787486774656e-05,
      "loss": 1.6932,
      "step": 1119
    },
    {
      "epoch": 0.4389359721745946,
      "grad_norm": 3.929225206375122,
      "learning_rate": 3.041629139962283e-05,
      "loss": 2.0081,
      "step": 1120
    },
    {
      "epoch": 0.43932787929260764,
      "grad_norm": 3.8846709728240967,
      "learning_rate": 3.0265017270457775e-05,
      "loss": 2.0683,
      "step": 1121
    },
    {
      "epoch": 0.4397197864106207,
      "grad_norm": 4.118119239807129,
      "learning_rate": 3.0114053152750556e-05,
      "loss": 1.9363,
      "step": 1122
    },
    {
      "epoch": 0.4401116935286337,
      "grad_norm": 4.2815937995910645,
      "learning_rate": 2.9963399717622077e-05,
      "loss": 1.721,
      "step": 1123
    },
    {
      "epoch": 0.44050360064664673,
      "grad_norm": 3.8667311668395996,
      "learning_rate": 2.98130576348121e-05,
      "loss": 1.6522,
      "step": 1124
    },
    {
      "epoch": 0.4408955077646598,
      "grad_norm": 4.201051712036133,
      "learning_rate": 2.966302757267625e-05,
      "loss": 1.8621,
      "step": 1125
    },
    {
      "epoch": 0.4408955077646598,
      "eval_loss": 1.8709391355514526,
      "eval_runtime": 39.3819,
      "eval_samples_per_second": 54.568,
      "eval_steps_per_second": 27.297,
      "step": 1125
    },
    {
      "epoch": 0.4412874148826728,
      "grad_norm": 3.905463695526123,
      "learning_rate": 2.9513310198183065e-05,
      "loss": 1.8967,
      "step": 1126
    },
    {
      "epoch": 0.4416793220006858,
      "grad_norm": 4.339902877807617,
      "learning_rate": 2.936390617691097e-05,
      "loss": 1.9623,
      "step": 1127
    },
    {
      "epoch": 0.4420712291186989,
      "grad_norm": 4.53336763381958,
      "learning_rate": 2.9214816173045356e-05,
      "loss": 1.7501,
      "step": 1128
    },
    {
      "epoch": 0.4424631362367119,
      "grad_norm": 4.321282863616943,
      "learning_rate": 2.906604084937572e-05,
      "loss": 1.9344,
      "step": 1129
    },
    {
      "epoch": 0.4428550433547249,
      "grad_norm": 4.578629493713379,
      "learning_rate": 2.8917580867292526e-05,
      "loss": 2.4232,
      "step": 1130
    },
    {
      "epoch": 0.443246950472738,
      "grad_norm": 4.56374979019165,
      "learning_rate": 2.8769436886784408e-05,
      "loss": 2.2241,
      "step": 1131
    },
    {
      "epoch": 0.443638857590751,
      "grad_norm": 4.458689212799072,
      "learning_rate": 2.862160956643517e-05,
      "loss": 2.0046,
      "step": 1132
    },
    {
      "epoch": 0.444030764708764,
      "grad_norm": 4.124095916748047,
      "learning_rate": 2.847409956342092e-05,
      "loss": 1.3972,
      "step": 1133
    },
    {
      "epoch": 0.444422671826777,
      "grad_norm": 4.54262113571167,
      "learning_rate": 2.8326907533507074e-05,
      "loss": 1.9323,
      "step": 1134
    },
    {
      "epoch": 0.4448145789447901,
      "grad_norm": 4.80766487121582,
      "learning_rate": 2.8180034131045464e-05,
      "loss": 1.8253,
      "step": 1135
    },
    {
      "epoch": 0.4452064860628031,
      "grad_norm": 6.259877681732178,
      "learning_rate": 2.8033480008971546e-05,
      "loss": 2.2948,
      "step": 1136
    },
    {
      "epoch": 0.4455983931808161,
      "grad_norm": 4.698707103729248,
      "learning_rate": 2.7887245818801277e-05,
      "loss": 1.807,
      "step": 1137
    },
    {
      "epoch": 0.4459903002988292,
      "grad_norm": 5.418363094329834,
      "learning_rate": 2.7741332210628345e-05,
      "loss": 2.5922,
      "step": 1138
    },
    {
      "epoch": 0.4463822074168422,
      "grad_norm": 5.3498759269714355,
      "learning_rate": 2.759573983312138e-05,
      "loss": 2.1608,
      "step": 1139
    },
    {
      "epoch": 0.4467741145348552,
      "grad_norm": 5.069705486297607,
      "learning_rate": 2.7450469333520855e-05,
      "loss": 2.2433,
      "step": 1140
    },
    {
      "epoch": 0.4471660216528683,
      "grad_norm": 5.354491233825684,
      "learning_rate": 2.730552135763632e-05,
      "loss": 2.1044,
      "step": 1141
    },
    {
      "epoch": 0.4475579287708813,
      "grad_norm": 5.228254318237305,
      "learning_rate": 2.7160896549843562e-05,
      "loss": 1.9293,
      "step": 1142
    },
    {
      "epoch": 0.4479498358888943,
      "grad_norm": 4.992114543914795,
      "learning_rate": 2.701659555308169e-05,
      "loss": 1.7596,
      "step": 1143
    },
    {
      "epoch": 0.4483417430069074,
      "grad_norm": 5.3238139152526855,
      "learning_rate": 2.6872619008850274e-05,
      "loss": 1.7265,
      "step": 1144
    },
    {
      "epoch": 0.4487336501249204,
      "grad_norm": 6.184856414794922,
      "learning_rate": 2.672896755720654e-05,
      "loss": 2.0537,
      "step": 1145
    },
    {
      "epoch": 0.4491255572429334,
      "grad_norm": 6.088552474975586,
      "learning_rate": 2.6585641836762433e-05,
      "loss": 2.0862,
      "step": 1146
    },
    {
      "epoch": 0.4495174643609465,
      "grad_norm": 7.702106952667236,
      "learning_rate": 2.6442642484681944e-05,
      "loss": 2.0236,
      "step": 1147
    },
    {
      "epoch": 0.4499093714789595,
      "grad_norm": 7.141883850097656,
      "learning_rate": 2.6299970136678077e-05,
      "loss": 1.8576,
      "step": 1148
    },
    {
      "epoch": 0.4503012785969725,
      "grad_norm": 8.247967720031738,
      "learning_rate": 2.6157625427010156e-05,
      "loss": 2.0878,
      "step": 1149
    },
    {
      "epoch": 0.45069318571498557,
      "grad_norm": 11.310791969299316,
      "learning_rate": 2.6015608988480955e-05,
      "loss": 2.4854,
      "step": 1150
    },
    {
      "epoch": 0.4510850928329986,
      "grad_norm": 1.2903518676757812,
      "learning_rate": 2.5873921452433915e-05,
      "loss": 0.7109,
      "step": 1151
    },
    {
      "epoch": 0.4514769999510116,
      "grad_norm": 1.3248708248138428,
      "learning_rate": 2.57325634487503e-05,
      "loss": 0.3858,
      "step": 1152
    },
    {
      "epoch": 0.45186890706902466,
      "grad_norm": 1.9522415399551392,
      "learning_rate": 2.5591535605846383e-05,
      "loss": 1.4758,
      "step": 1153
    },
    {
      "epoch": 0.4522608141870377,
      "grad_norm": 2.146930694580078,
      "learning_rate": 2.5450838550670808e-05,
      "loss": 1.2781,
      "step": 1154
    },
    {
      "epoch": 0.4526527213050507,
      "grad_norm": 2.3952581882476807,
      "learning_rate": 2.5310472908701555e-05,
      "loss": 1.6293,
      "step": 1155
    },
    {
      "epoch": 0.45304462842306376,
      "grad_norm": 2.829385757446289,
      "learning_rate": 2.5170439303943294e-05,
      "loss": 1.5638,
      "step": 1156
    },
    {
      "epoch": 0.45343653554107677,
      "grad_norm": 2.968790054321289,
      "learning_rate": 2.503073835892471e-05,
      "loss": 2.2704,
      "step": 1157
    },
    {
      "epoch": 0.4538284426590898,
      "grad_norm": 3.252389669418335,
      "learning_rate": 2.4891370694695517e-05,
      "loss": 1.9568,
      "step": 1158
    },
    {
      "epoch": 0.45422034977710285,
      "grad_norm": 3.182776689529419,
      "learning_rate": 2.4752336930823837e-05,
      "loss": 1.9977,
      "step": 1159
    },
    {
      "epoch": 0.45461225689511586,
      "grad_norm": 3.20595121383667,
      "learning_rate": 2.4613637685393432e-05,
      "loss": 1.7874,
      "step": 1160
    },
    {
      "epoch": 0.4550041640131289,
      "grad_norm": 3.2108423709869385,
      "learning_rate": 2.4475273575000936e-05,
      "loss": 1.5715,
      "step": 1161
    },
    {
      "epoch": 0.45539607113114194,
      "grad_norm": 3.374884605407715,
      "learning_rate": 2.4337245214753103e-05,
      "loss": 1.6547,
      "step": 1162
    },
    {
      "epoch": 0.45578797824915496,
      "grad_norm": 3.325153350830078,
      "learning_rate": 2.4199553218264093e-05,
      "loss": 1.552,
      "step": 1163
    },
    {
      "epoch": 0.45617988536716797,
      "grad_norm": 3.4900574684143066,
      "learning_rate": 2.4062198197652752e-05,
      "loss": 1.9581,
      "step": 1164
    },
    {
      "epoch": 0.45657179248518104,
      "grad_norm": 3.3368542194366455,
      "learning_rate": 2.3925180763539844e-05,
      "loss": 1.8785,
      "step": 1165
    },
    {
      "epoch": 0.45696369960319405,
      "grad_norm": 3.696206569671631,
      "learning_rate": 2.3788501525045438e-05,
      "loss": 1.9671,
      "step": 1166
    },
    {
      "epoch": 0.45735560672120706,
      "grad_norm": 3.8370039463043213,
      "learning_rate": 2.3652161089786086e-05,
      "loss": 1.9951,
      "step": 1167
    },
    {
      "epoch": 0.4577475138392201,
      "grad_norm": 3.591837167739868,
      "learning_rate": 2.351616006387214e-05,
      "loss": 1.4722,
      "step": 1168
    },
    {
      "epoch": 0.45813942095723315,
      "grad_norm": 3.6157119274139404,
      "learning_rate": 2.3380499051905137e-05,
      "loss": 1.7826,
      "step": 1169
    },
    {
      "epoch": 0.45853132807524616,
      "grad_norm": 4.350569725036621,
      "learning_rate": 2.324517865697501e-05,
      "loss": 2.1727,
      "step": 1170
    },
    {
      "epoch": 0.45892323519325917,
      "grad_norm": 4.022909641265869,
      "learning_rate": 2.3110199480657525e-05,
      "loss": 1.9128,
      "step": 1171
    },
    {
      "epoch": 0.45931514231127224,
      "grad_norm": 4.371360778808594,
      "learning_rate": 2.2975562123011495e-05,
      "loss": 1.9332,
      "step": 1172
    },
    {
      "epoch": 0.45970704942928525,
      "grad_norm": 4.438855171203613,
      "learning_rate": 2.2841267182576143e-05,
      "loss": 2.2052,
      "step": 1173
    },
    {
      "epoch": 0.46009895654729827,
      "grad_norm": 4.477270126342773,
      "learning_rate": 2.2707315256368433e-05,
      "loss": 2.0621,
      "step": 1174
    },
    {
      "epoch": 0.46049086366531133,
      "grad_norm": 4.10659646987915,
      "learning_rate": 2.2573706939880555e-05,
      "loss": 1.7448,
      "step": 1175
    },
    {
      "epoch": 0.46088277078332435,
      "grad_norm": 4.038181304931641,
      "learning_rate": 2.2440442827077045e-05,
      "loss": 1.502,
      "step": 1176
    },
    {
      "epoch": 0.46127467790133736,
      "grad_norm": 4.006118297576904,
      "learning_rate": 2.230752351039228e-05,
      "loss": 1.6319,
      "step": 1177
    },
    {
      "epoch": 0.46166658501935043,
      "grad_norm": 4.865692615509033,
      "learning_rate": 2.2174949580727832e-05,
      "loss": 1.7084,
      "step": 1178
    },
    {
      "epoch": 0.46205849213736344,
      "grad_norm": 4.602259635925293,
      "learning_rate": 2.2042721627449846e-05,
      "loss": 2.38,
      "step": 1179
    },
    {
      "epoch": 0.46245039925537645,
      "grad_norm": 4.572423934936523,
      "learning_rate": 2.1910840238386398e-05,
      "loss": 2.0662,
      "step": 1180
    },
    {
      "epoch": 0.4628423063733895,
      "grad_norm": 4.433712482452393,
      "learning_rate": 2.1779305999824884e-05,
      "loss": 1.9092,
      "step": 1181
    },
    {
      "epoch": 0.46323421349140254,
      "grad_norm": 4.5985612869262695,
      "learning_rate": 2.164811949650942e-05,
      "loss": 1.9287,
      "step": 1182
    },
    {
      "epoch": 0.46362612060941555,
      "grad_norm": 4.545965194702148,
      "learning_rate": 2.1517281311638217e-05,
      "loss": 1.8265,
      "step": 1183
    },
    {
      "epoch": 0.4640180277274286,
      "grad_norm": 4.913460731506348,
      "learning_rate": 2.1386792026861103e-05,
      "loss": 2.1133,
      "step": 1184
    },
    {
      "epoch": 0.46440993484544163,
      "grad_norm": 4.976416110992432,
      "learning_rate": 2.125665222227675e-05,
      "loss": 1.859,
      "step": 1185
    },
    {
      "epoch": 0.46480184196345464,
      "grad_norm": 4.737184524536133,
      "learning_rate": 2.112686247643024e-05,
      "loss": 1.8176,
      "step": 1186
    },
    {
      "epoch": 0.4651937490814677,
      "grad_norm": 4.664793968200684,
      "learning_rate": 2.09974233663104e-05,
      "loss": 2.0407,
      "step": 1187
    },
    {
      "epoch": 0.4655856561994807,
      "grad_norm": 5.029167652130127,
      "learning_rate": 2.0868335467347366e-05,
      "loss": 1.8795,
      "step": 1188
    },
    {
      "epoch": 0.46597756331749374,
      "grad_norm": 5.17645263671875,
      "learning_rate": 2.073959935340988e-05,
      "loss": 2.1018,
      "step": 1189
    },
    {
      "epoch": 0.4663694704355068,
      "grad_norm": 5.135799884796143,
      "learning_rate": 2.06112155968028e-05,
      "loss": 1.762,
      "step": 1190
    },
    {
      "epoch": 0.4667613775535198,
      "grad_norm": 5.43034553527832,
      "learning_rate": 2.0483184768264596e-05,
      "loss": 2.226,
      "step": 1191
    },
    {
      "epoch": 0.46715328467153283,
      "grad_norm": 6.0636820793151855,
      "learning_rate": 2.035550743696468e-05,
      "loss": 1.9962,
      "step": 1192
    },
    {
      "epoch": 0.4675451917895459,
      "grad_norm": 5.623441219329834,
      "learning_rate": 2.022818417050113e-05,
      "loss": 1.5376,
      "step": 1193
    },
    {
      "epoch": 0.4679370989075589,
      "grad_norm": 6.107014179229736,
      "learning_rate": 2.0101215534897855e-05,
      "loss": 2.2256,
      "step": 1194
    },
    {
      "epoch": 0.4683290060255719,
      "grad_norm": 6.131252765655518,
      "learning_rate": 1.99746020946023e-05,
      "loss": 2.2417,
      "step": 1195
    },
    {
      "epoch": 0.468720913143585,
      "grad_norm": 6.999653339385986,
      "learning_rate": 1.9848344412482854e-05,
      "loss": 1.7416,
      "step": 1196
    },
    {
      "epoch": 0.469112820261598,
      "grad_norm": 6.954730033874512,
      "learning_rate": 1.9722443049826344e-05,
      "loss": 1.8519,
      "step": 1197
    },
    {
      "epoch": 0.469504727379611,
      "grad_norm": 8.254525184631348,
      "learning_rate": 1.9596898566335576e-05,
      "loss": 1.9218,
      "step": 1198
    },
    {
      "epoch": 0.4698966344976241,
      "grad_norm": 9.411231994628906,
      "learning_rate": 1.9471711520126824e-05,
      "loss": 1.8943,
      "step": 1199
    },
    {
      "epoch": 0.4702885416156371,
      "grad_norm": 12.048138618469238,
      "learning_rate": 1.9346882467727325e-05,
      "loss": 2.0123,
      "step": 1200
    },
    {
      "epoch": 0.4706804487336501,
      "grad_norm": 1.2816966772079468,
      "learning_rate": 1.9222411964072884e-05,
      "loss": 0.9805,
      "step": 1201
    },
    {
      "epoch": 0.4710723558516632,
      "grad_norm": 1.9703985452651978,
      "learning_rate": 1.9098300562505266e-05,
      "loss": 1.5275,
      "step": 1202
    },
    {
      "epoch": 0.4714642629696762,
      "grad_norm": 1.6411155462265015,
      "learning_rate": 1.8974548814769944e-05,
      "loss": 1.1657,
      "step": 1203
    },
    {
      "epoch": 0.4718561700876892,
      "grad_norm": 2.3778491020202637,
      "learning_rate": 1.8851157271013442e-05,
      "loss": 1.4481,
      "step": 1204
    },
    {
      "epoch": 0.4722480772057022,
      "grad_norm": 2.4556612968444824,
      "learning_rate": 1.872812647978095e-05,
      "loss": 1.6667,
      "step": 1205
    },
    {
      "epoch": 0.4726399843237153,
      "grad_norm": 2.251343250274658,
      "learning_rate": 1.8605456988014015e-05,
      "loss": 1.2396,
      "step": 1206
    },
    {
      "epoch": 0.4730318914417283,
      "grad_norm": 2.91467022895813,
      "learning_rate": 1.8483149341047923e-05,
      "loss": 1.7779,
      "step": 1207
    },
    {
      "epoch": 0.4734237985597413,
      "grad_norm": 2.7803587913513184,
      "learning_rate": 1.8361204082609352e-05,
      "loss": 1.8255,
      "step": 1208
    },
    {
      "epoch": 0.4738157056777544,
      "grad_norm": 2.550692081451416,
      "learning_rate": 1.8239621754813995e-05,
      "loss": 1.4439,
      "step": 1209
    },
    {
      "epoch": 0.4742076127957674,
      "grad_norm": 3.0735573768615723,
      "learning_rate": 1.811840289816409e-05,
      "loss": 1.3542,
      "step": 1210
    },
    {
      "epoch": 0.4745995199137804,
      "grad_norm": 3.1861467361450195,
      "learning_rate": 1.799754805154603e-05,
      "loss": 1.7391,
      "step": 1211
    },
    {
      "epoch": 0.4749914270317935,
      "grad_norm": 3.4716169834136963,
      "learning_rate": 1.787705775222802e-05,
      "loss": 1.6906,
      "step": 1212
    },
    {
      "epoch": 0.4753833341498065,
      "grad_norm": 3.5379340648651123,
      "learning_rate": 1.775693253585763e-05,
      "loss": 2.2569,
      "step": 1213
    },
    {
      "epoch": 0.4757752412678195,
      "grad_norm": 3.4137425422668457,
      "learning_rate": 1.763717293645939e-05,
      "loss": 1.7515,
      "step": 1214
    },
    {
      "epoch": 0.4761671483858326,
      "grad_norm": 3.6310174465179443,
      "learning_rate": 1.7517779486432495e-05,
      "loss": 2.3349,
      "step": 1215
    },
    {
      "epoch": 0.4765590555038456,
      "grad_norm": 3.9528141021728516,
      "learning_rate": 1.7398752716548395e-05,
      "loss": 2.5629,
      "step": 1216
    },
    {
      "epoch": 0.4769509626218586,
      "grad_norm": 3.596297025680542,
      "learning_rate": 1.728009315594843e-05,
      "loss": 1.8068,
      "step": 1217
    },
    {
      "epoch": 0.47734286973987167,
      "grad_norm": 3.315270185470581,
      "learning_rate": 1.716180133214149e-05,
      "loss": 1.6211,
      "step": 1218
    },
    {
      "epoch": 0.4777347768578847,
      "grad_norm": 3.9681286811828613,
      "learning_rate": 1.704387777100165e-05,
      "loss": 1.8875,
      "step": 1219
    },
    {
      "epoch": 0.4781266839758977,
      "grad_norm": 3.8074514865875244,
      "learning_rate": 1.6926322996765897e-05,
      "loss": 1.8768,
      "step": 1220
    },
    {
      "epoch": 0.47851859109391076,
      "grad_norm": 3.6715331077575684,
      "learning_rate": 1.6809137532031704e-05,
      "loss": 1.8684,
      "step": 1221
    },
    {
      "epoch": 0.4789104982119238,
      "grad_norm": 3.994828462600708,
      "learning_rate": 1.6692321897754758e-05,
      "loss": 2.1018,
      "step": 1222
    },
    {
      "epoch": 0.4793024053299368,
      "grad_norm": 4.163514614105225,
      "learning_rate": 1.65758766132467e-05,
      "loss": 1.7621,
      "step": 1223
    },
    {
      "epoch": 0.47969431244794986,
      "grad_norm": 4.0636444091796875,
      "learning_rate": 1.6459802196172668e-05,
      "loss": 1.5892,
      "step": 1224
    },
    {
      "epoch": 0.48008621956596287,
      "grad_norm": 4.426535606384277,
      "learning_rate": 1.634409916254914e-05,
      "loss": 1.9036,
      "step": 1225
    },
    {
      "epoch": 0.4804781266839759,
      "grad_norm": 4.42534875869751,
      "learning_rate": 1.622876802674158e-05,
      "loss": 2.0746,
      "step": 1226
    },
    {
      "epoch": 0.48087003380198895,
      "grad_norm": 4.2113847732543945,
      "learning_rate": 1.6113809301462125e-05,
      "loss": 2.0476,
      "step": 1227
    },
    {
      "epoch": 0.48126194092000196,
      "grad_norm": 4.861171722412109,
      "learning_rate": 1.599922349776738e-05,
      "loss": 2.1532,
      "step": 1228
    },
    {
      "epoch": 0.481653848038015,
      "grad_norm": 4.383703231811523,
      "learning_rate": 1.5885011125056047e-05,
      "loss": 1.9094,
      "step": 1229
    },
    {
      "epoch": 0.48204575515602804,
      "grad_norm": 4.693561553955078,
      "learning_rate": 1.5771172691066794e-05,
      "loss": 1.8299,
      "step": 1230
    },
    {
      "epoch": 0.48243766227404106,
      "grad_norm": 4.714425563812256,
      "learning_rate": 1.565770870187585e-05,
      "loss": 2.1893,
      "step": 1231
    },
    {
      "epoch": 0.48282956939205407,
      "grad_norm": 4.604955196380615,
      "learning_rate": 1.5544619661894864e-05,
      "loss": 2.0113,
      "step": 1232
    },
    {
      "epoch": 0.48322147651006714,
      "grad_norm": 5.0616912841796875,
      "learning_rate": 1.543190607386861e-05,
      "loss": 2.173,
      "step": 1233
    },
    {
      "epoch": 0.48361338362808015,
      "grad_norm": 5.034733772277832,
      "learning_rate": 1.5319568438872745e-05,
      "loss": 2.4819,
      "step": 1234
    },
    {
      "epoch": 0.48400529074609316,
      "grad_norm": 4.723273277282715,
      "learning_rate": 1.520760725631164e-05,
      "loss": 1.8429,
      "step": 1235
    },
    {
      "epoch": 0.48439719786410623,
      "grad_norm": 5.021108150482178,
      "learning_rate": 1.5096023023916094e-05,
      "loss": 1.9197,
      "step": 1236
    },
    {
      "epoch": 0.48478910498211925,
      "grad_norm": 4.8129096031188965,
      "learning_rate": 1.498481623774115e-05,
      "loss": 1.9067,
      "step": 1237
    },
    {
      "epoch": 0.48518101210013226,
      "grad_norm": 5.344303607940674,
      "learning_rate": 1.4873987392163947e-05,
      "loss": 2.1829,
      "step": 1238
    },
    {
      "epoch": 0.4855729192181453,
      "grad_norm": 4.596015453338623,
      "learning_rate": 1.4763536979881354e-05,
      "loss": 1.9883,
      "step": 1239
    },
    {
      "epoch": 0.48596482633615834,
      "grad_norm": 5.126211643218994,
      "learning_rate": 1.4653465491908003e-05,
      "loss": 2.0296,
      "step": 1240
    },
    {
      "epoch": 0.48635673345417135,
      "grad_norm": 5.316787242889404,
      "learning_rate": 1.4543773417573925e-05,
      "loss": 2.3697,
      "step": 1241
    },
    {
      "epoch": 0.48674864057218437,
      "grad_norm": 5.3284125328063965,
      "learning_rate": 1.4434461244522458e-05,
      "loss": 2.0194,
      "step": 1242
    },
    {
      "epoch": 0.48714054769019743,
      "grad_norm": 5.978663921356201,
      "learning_rate": 1.4325529458708065e-05,
      "loss": 2.2342,
      "step": 1243
    },
    {
      "epoch": 0.48753245480821045,
      "grad_norm": 6.219578742980957,
      "learning_rate": 1.4216978544394177e-05,
      "loss": 1.5737,
      "step": 1244
    },
    {
      "epoch": 0.48792436192622346,
      "grad_norm": 5.139858722686768,
      "learning_rate": 1.4108808984151023e-05,
      "loss": 1.3506,
      "step": 1245
    },
    {
      "epoch": 0.48831626904423653,
      "grad_norm": 6.140778541564941,
      "learning_rate": 1.4001021258853509e-05,
      "loss": 1.9014,
      "step": 1246
    },
    {
      "epoch": 0.48870817616224954,
      "grad_norm": 7.050278186798096,
      "learning_rate": 1.3893615847679065e-05,
      "loss": 2.0598,
      "step": 1247
    },
    {
      "epoch": 0.48910008328026255,
      "grad_norm": 7.652315616607666,
      "learning_rate": 1.3786593228105494e-05,
      "loss": 2.0894,
      "step": 1248
    },
    {
      "epoch": 0.4894919903982756,
      "grad_norm": 7.891330718994141,
      "learning_rate": 1.3679953875908957e-05,
      "loss": 1.4682,
      "step": 1249
    },
    {
      "epoch": 0.48988389751628864,
      "grad_norm": 11.646049499511719,
      "learning_rate": 1.3573698265161683e-05,
      "loss": 1.6784,
      "step": 1250
    },
    {
      "epoch": 0.49027580463430165,
      "grad_norm": 1.5593860149383545,
      "learning_rate": 1.3467826868229994e-05,
      "loss": 1.4679,
      "step": 1251
    },
    {
      "epoch": 0.4906677117523147,
      "grad_norm": 1.4130744934082031,
      "learning_rate": 1.3362340155772146e-05,
      "loss": 0.5222,
      "step": 1252
    },
    {
      "epoch": 0.49105961887032773,
      "grad_norm": 1.9753338098526,
      "learning_rate": 1.3257238596736266e-05,
      "loss": 1.3745,
      "step": 1253
    },
    {
      "epoch": 0.49145152598834074,
      "grad_norm": 2.440312147140503,
      "learning_rate": 1.3152522658358245e-05,
      "loss": 1.9949,
      "step": 1254
    },
    {
      "epoch": 0.4918434331063538,
      "grad_norm": 1.9646104574203491,
      "learning_rate": 1.3048192806159721e-05,
      "loss": 1.1746,
      "step": 1255
    },
    {
      "epoch": 0.4922353402243668,
      "grad_norm": 2.7373812198638916,
      "learning_rate": 1.2944249503945894e-05,
      "loss": 1.795,
      "step": 1256
    },
    {
      "epoch": 0.49262724734237984,
      "grad_norm": 3.2771518230438232,
      "learning_rate": 1.2840693213803545e-05,
      "loss": 1.6479,
      "step": 1257
    },
    {
      "epoch": 0.4930191544603929,
      "grad_norm": 2.7557213306427,
      "learning_rate": 1.2737524396099032e-05,
      "loss": 1.7249,
      "step": 1258
    },
    {
      "epoch": 0.4934110615784059,
      "grad_norm": 2.8898143768310547,
      "learning_rate": 1.2634743509476088e-05,
      "loss": 1.7679,
      "step": 1259
    },
    {
      "epoch": 0.49380296869641893,
      "grad_norm": 3.359405755996704,
      "learning_rate": 1.2532351010853916e-05,
      "loss": 1.8335,
      "step": 1260
    },
    {
      "epoch": 0.494194875814432,
      "grad_norm": 3.7645535469055176,
      "learning_rate": 1.243034735542512e-05,
      "loss": 2.0702,
      "step": 1261
    },
    {
      "epoch": 0.494586782932445,
      "grad_norm": 3.5672197341918945,
      "learning_rate": 1.2328732996653669e-05,
      "loss": 1.6744,
      "step": 1262
    },
    {
      "epoch": 0.494978690050458,
      "grad_norm": 2.9757351875305176,
      "learning_rate": 1.2227508386272878e-05,
      "loss": 1.6618,
      "step": 1263
    },
    {
      "epoch": 0.4953705971684711,
      "grad_norm": 3.334315061569214,
      "learning_rate": 1.212667397428342e-05,
      "loss": 1.5938,
      "step": 1264
    },
    {
      "epoch": 0.4957625042864841,
      "grad_norm": 3.7320988178253174,
      "learning_rate": 1.2026230208951306e-05,
      "loss": 2.0401,
      "step": 1265
    },
    {
      "epoch": 0.4961544114044971,
      "grad_norm": 3.2607672214508057,
      "learning_rate": 1.1926177536805905e-05,
      "loss": 1.5748,
      "step": 1266
    },
    {
      "epoch": 0.4965463185225102,
      "grad_norm": 4.308067321777344,
      "learning_rate": 1.1826516402637989e-05,
      "loss": 1.8738,
      "step": 1267
    },
    {
      "epoch": 0.4969382256405232,
      "grad_norm": 3.7249412536621094,
      "learning_rate": 1.1727247249497685e-05,
      "loss": 1.6869,
      "step": 1268
    },
    {
      "epoch": 0.4973301327585362,
      "grad_norm": 3.4116594791412354,
      "learning_rate": 1.1628370518692533e-05,
      "loss": 1.5012,
      "step": 1269
    },
    {
      "epoch": 0.4977220398765493,
      "grad_norm": 4.3949384689331055,
      "learning_rate": 1.152988664978556e-05,
      "loss": 2.0143,
      "step": 1270
    },
    {
      "epoch": 0.4981139469945623,
      "grad_norm": 3.9335439205169678,
      "learning_rate": 1.1431796080593283e-05,
      "loss": 1.8934,
      "step": 1271
    },
    {
      "epoch": 0.4985058541125753,
      "grad_norm": 4.108706951141357,
      "learning_rate": 1.1334099247183783e-05,
      "loss": 1.9592,
      "step": 1272
    },
    {
      "epoch": 0.4988977612305884,
      "grad_norm": 3.5353474617004395,
      "learning_rate": 1.1236796583874787e-05,
      "loss": 1.4046,
      "step": 1273
    },
    {
      "epoch": 0.4992896683486014,
      "grad_norm": 3.8804337978363037,
      "learning_rate": 1.1139888523231678e-05,
      "loss": 1.4882,
      "step": 1274
    },
    {
      "epoch": 0.4996815754666144,
      "grad_norm": 3.8991992473602295,
      "learning_rate": 1.1043375496065611e-05,
      "loss": 1.6514,
      "step": 1275
    },
    {
      "epoch": 0.5000734825846275,
      "grad_norm": 4.6726202964782715,
      "learning_rate": 1.0947257931431642e-05,
      "loss": 2.0945,
      "step": 1276
    },
    {
      "epoch": 0.5004653897026404,
      "grad_norm": 4.43438720703125,
      "learning_rate": 1.0851536256626705e-05,
      "loss": 1.879,
      "step": 1277
    },
    {
      "epoch": 0.5008572968206535,
      "grad_norm": 4.0653791427612305,
      "learning_rate": 1.0756210897187812e-05,
      "loss": 1.5855,
      "step": 1278
    },
    {
      "epoch": 0.5012492039386666,
      "grad_norm": 4.888524055480957,
      "learning_rate": 1.0661282276890127e-05,
      "loss": 2.1269,
      "step": 1279
    },
    {
      "epoch": 0.5016411110566795,
      "grad_norm": 4.573784351348877,
      "learning_rate": 1.0566750817745074e-05,
      "loss": 2.1842,
      "step": 1280
    },
    {
      "epoch": 0.5020330181746926,
      "grad_norm": 4.8447184562683105,
      "learning_rate": 1.0472616939998492e-05,
      "loss": 2.2226,
      "step": 1281
    },
    {
      "epoch": 0.5024249252927057,
      "grad_norm": 4.745467185974121,
      "learning_rate": 1.0378881062128731e-05,
      "loss": 2.3902,
      "step": 1282
    },
    {
      "epoch": 0.5028168324107186,
      "grad_norm": 4.509251117706299,
      "learning_rate": 1.0285543600844804e-05,
      "loss": 1.9423,
      "step": 1283
    },
    {
      "epoch": 0.5032087395287317,
      "grad_norm": 4.820037364959717,
      "learning_rate": 1.019260497108453e-05,
      "loss": 1.9476,
      "step": 1284
    },
    {
      "epoch": 0.5036006466467448,
      "grad_norm": 4.707878112792969,
      "learning_rate": 1.010006558601274e-05,
      "loss": 1.5921,
      "step": 1285
    },
    {
      "epoch": 0.5039925537647577,
      "grad_norm": 4.771453857421875,
      "learning_rate": 1.000792585701934e-05,
      "loss": 2.1841,
      "step": 1286
    },
    {
      "epoch": 0.5043844608827708,
      "grad_norm": 4.900398254394531,
      "learning_rate": 9.91618619371757e-06,
      "loss": 2.09,
      "step": 1287
    },
    {
      "epoch": 0.5047763680007838,
      "grad_norm": 4.940152645111084,
      "learning_rate": 9.82484700394215e-06,
      "loss": 1.8391,
      "step": 1288
    },
    {
      "epoch": 0.5051682751187968,
      "grad_norm": 4.643918991088867,
      "learning_rate": 9.73390869374743e-06,
      "loss": 1.9663,
      "step": 1289
    },
    {
      "epoch": 0.5055601822368099,
      "grad_norm": 5.533751964569092,
      "learning_rate": 9.643371667405698e-06,
      "loss": 2.3462,
      "step": 1290
    },
    {
      "epoch": 0.5059520893548229,
      "grad_norm": 4.972218990325928,
      "learning_rate": 9.553236327405246e-06,
      "loss": 2.0053,
      "step": 1291
    },
    {
      "epoch": 0.5063439964728359,
      "grad_norm": 5.573946952819824,
      "learning_rate": 9.463503074448677e-06,
      "loss": 2.0105,
      "step": 1292
    },
    {
      "epoch": 0.506735903590849,
      "grad_norm": 5.260864734649658,
      "learning_rate": 9.374172307451068e-06,
      "loss": 1.5386,
      "step": 1293
    },
    {
      "epoch": 0.507127810708862,
      "grad_norm": 5.5153985023498535,
      "learning_rate": 9.285244423538197e-06,
      "loss": 1.9529,
      "step": 1294
    },
    {
      "epoch": 0.507519717826875,
      "grad_norm": 6.7475409507751465,
      "learning_rate": 9.196719818044886e-06,
      "loss": 1.8721,
      "step": 1295
    },
    {
      "epoch": 0.5079116249448881,
      "grad_norm": 6.267107009887695,
      "learning_rate": 9.108598884513053e-06,
      "loss": 1.3398,
      "step": 1296
    },
    {
      "epoch": 0.5083035320629011,
      "grad_norm": 7.718843936920166,
      "learning_rate": 9.020882014690136e-06,
      "loss": 1.5895,
      "step": 1297
    },
    {
      "epoch": 0.5086954391809141,
      "grad_norm": 7.658002853393555,
      "learning_rate": 8.933569598527247e-06,
      "loss": 2.0302,
      "step": 1298
    },
    {
      "epoch": 0.5090873462989272,
      "grad_norm": 6.787975311279297,
      "learning_rate": 8.846662024177477e-06,
      "loss": 1.0236,
      "step": 1299
    },
    {
      "epoch": 0.5094792534169402,
      "grad_norm": 11.165488243103027,
      "learning_rate": 8.760159677994172e-06,
      "loss": 1.9742,
      "step": 1300
    },
    {
      "epoch": 0.5098711605349532,
      "grad_norm": 1.6676123142242432,
      "learning_rate": 8.674062944529216e-06,
      "loss": 1.3251,
      "step": 1301
    },
    {
      "epoch": 0.5102630676529663,
      "grad_norm": 1.0167549848556519,
      "learning_rate": 8.588372206531292e-06,
      "loss": 0.5012,
      "step": 1302
    },
    {
      "epoch": 0.5106549747709793,
      "grad_norm": 1.505925178527832,
      "learning_rate": 8.503087844944213e-06,
      "loss": 0.4747,
      "step": 1303
    },
    {
      "epoch": 0.5110468818889923,
      "grad_norm": 1.7298166751861572,
      "learning_rate": 8.418210238905256e-06,
      "loss": 1.046,
      "step": 1304
    },
    {
      "epoch": 0.5114387890070053,
      "grad_norm": 1.8829516172409058,
      "learning_rate": 8.333739765743398e-06,
      "loss": 1.2416,
      "step": 1305
    },
    {
      "epoch": 0.5118306961250184,
      "grad_norm": 1.9784975051879883,
      "learning_rate": 8.249676800977658e-06,
      "loss": 1.1391,
      "step": 1306
    },
    {
      "epoch": 0.5122226032430314,
      "grad_norm": 2.633385181427002,
      "learning_rate": 8.16602171831553e-06,
      "loss": 1.9154,
      "step": 1307
    },
    {
      "epoch": 0.5126145103610444,
      "grad_norm": 3.1572091579437256,
      "learning_rate": 8.082774889651168e-06,
      "loss": 2.1728,
      "step": 1308
    },
    {
      "epoch": 0.5130064174790575,
      "grad_norm": 2.82183837890625,
      "learning_rate": 7.999936685063835e-06,
      "loss": 1.5539,
      "step": 1309
    },
    {
      "epoch": 0.5133983245970705,
      "grad_norm": 2.987389087677002,
      "learning_rate": 7.91750747281621e-06,
      "loss": 1.4952,
      "step": 1310
    },
    {
      "epoch": 0.5137902317150835,
      "grad_norm": 4.090176105499268,
      "learning_rate": 7.835487619352811e-06,
      "loss": 2.2207,
      "step": 1311
    },
    {
      "epoch": 0.5141821388330966,
      "grad_norm": 3.3717334270477295,
      "learning_rate": 7.753877489298244e-06,
      "loss": 2.0226,
      "step": 1312
    },
    {
      "epoch": 0.5145740459511096,
      "grad_norm": 3.818284034729004,
      "learning_rate": 7.67267744545579e-06,
      "loss": 2.192,
      "step": 1313
    },
    {
      "epoch": 0.5149659530691226,
      "grad_norm": 3.6257965564727783,
      "learning_rate": 7.591887848805545e-06,
      "loss": 1.8098,
      "step": 1314
    },
    {
      "epoch": 0.5153578601871357,
      "grad_norm": 3.547513246536255,
      "learning_rate": 7.5115090585029966e-06,
      "loss": 2.0897,
      "step": 1315
    },
    {
      "epoch": 0.5157497673051487,
      "grad_norm": 3.8446054458618164,
      "learning_rate": 7.431541431877342e-06,
      "loss": 1.8845,
      "step": 1316
    },
    {
      "epoch": 0.5161416744231617,
      "grad_norm": 3.5873358249664307,
      "learning_rate": 7.351985324429933e-06,
      "loss": 1.8746,
      "step": 1317
    },
    {
      "epoch": 0.5165335815411748,
      "grad_norm": 4.027246475219727,
      "learning_rate": 7.272841089832694e-06,
      "loss": 2.2741,
      "step": 1318
    },
    {
      "epoch": 0.5169254886591877,
      "grad_norm": 3.4317448139190674,
      "learning_rate": 7.194109079926514e-06,
      "loss": 1.5461,
      "step": 1319
    },
    {
      "epoch": 0.5173173957772008,
      "grad_norm": 3.683546304702759,
      "learning_rate": 7.115789644719728e-06,
      "loss": 1.5832,
      "step": 1320
    },
    {
      "epoch": 0.5177093028952139,
      "grad_norm": 4.593991756439209,
      "learning_rate": 7.037883132386547e-06,
      "loss": 1.6827,
      "step": 1321
    },
    {
      "epoch": 0.5181012100132268,
      "grad_norm": 3.7813708782196045,
      "learning_rate": 6.960389889265517e-06,
      "loss": 1.8778,
      "step": 1322
    },
    {
      "epoch": 0.5184931171312399,
      "grad_norm": 4.44172477722168,
      "learning_rate": 6.883310259857944e-06,
      "loss": 1.7601,
      "step": 1323
    },
    {
      "epoch": 0.518885024249253,
      "grad_norm": 4.222709655761719,
      "learning_rate": 6.806644586826383e-06,
      "loss": 1.9261,
      "step": 1324
    },
    {
      "epoch": 0.5192769313672659,
      "grad_norm": 4.293867111206055,
      "learning_rate": 6.730393210993147e-06,
      "loss": 1.9647,
      "step": 1325
    },
    {
      "epoch": 0.519668838485279,
      "grad_norm": 4.473932266235352,
      "learning_rate": 6.654556471338746e-06,
      "loss": 2.0327,
      "step": 1326
    },
    {
      "epoch": 0.5200607456032921,
      "grad_norm": 4.121243000030518,
      "learning_rate": 6.579134705000412e-06,
      "loss": 1.4838,
      "step": 1327
    },
    {
      "epoch": 0.520452652721305,
      "grad_norm": 4.534484386444092,
      "learning_rate": 6.504128247270546e-06,
      "loss": 2.0137,
      "step": 1328
    },
    {
      "epoch": 0.5208445598393181,
      "grad_norm": 4.7986741065979,
      "learning_rate": 6.429537431595312e-06,
      "loss": 2.1107,
      "step": 1329
    },
    {
      "epoch": 0.5212364669573311,
      "grad_norm": 5.030558109283447,
      "learning_rate": 6.355362589573077e-06,
      "loss": 2.2379,
      "step": 1330
    },
    {
      "epoch": 0.5216283740753441,
      "grad_norm": 4.254178047180176,
      "learning_rate": 6.2816040509530165e-06,
      "loss": 1.8851,
      "step": 1331
    },
    {
      "epoch": 0.5220202811933572,
      "grad_norm": 4.246702194213867,
      "learning_rate": 6.2082621436335475e-06,
      "loss": 1.9072,
      "step": 1332
    },
    {
      "epoch": 0.5224121883113702,
      "grad_norm": 4.843648433685303,
      "learning_rate": 6.135337193660962e-06,
      "loss": 2.3032,
      "step": 1333
    },
    {
      "epoch": 0.5228040954293832,
      "grad_norm": 5.311872959136963,
      "learning_rate": 6.062829525227909e-06,
      "loss": 2.0073,
      "step": 1334
    },
    {
      "epoch": 0.5231960025473963,
      "grad_norm": 4.763339996337891,
      "learning_rate": 5.990739460672024e-06,
      "loss": 1.9776,
      "step": 1335
    },
    {
      "epoch": 0.5235879096654092,
      "grad_norm": 4.935484409332275,
      "learning_rate": 5.9190673204744255e-06,
      "loss": 2.0388,
      "step": 1336
    },
    {
      "epoch": 0.5239798167834223,
      "grad_norm": 4.658010959625244,
      "learning_rate": 5.84781342325833e-06,
      "loss": 2.355,
      "step": 1337
    },
    {
      "epoch": 0.5243717239014354,
      "grad_norm": 4.550975322723389,
      "learning_rate": 5.77697808578761e-06,
      "loss": 1.5731,
      "step": 1338
    },
    {
      "epoch": 0.5247636310194483,
      "grad_norm": 5.154825687408447,
      "learning_rate": 5.706561622965467e-06,
      "loss": 2.2845,
      "step": 1339
    },
    {
      "epoch": 0.5251555381374614,
      "grad_norm": 4.9441633224487305,
      "learning_rate": 5.636564347832907e-06,
      "loss": 1.6571,
      "step": 1340
    },
    {
      "epoch": 0.5255474452554745,
      "grad_norm": 5.694057941436768,
      "learning_rate": 5.566986571567401e-06,
      "loss": 1.7631,
      "step": 1341
    },
    {
      "epoch": 0.5259393523734874,
      "grad_norm": 5.547728538513184,
      "learning_rate": 5.497828603481569e-06,
      "loss": 2.4085,
      "step": 1342
    },
    {
      "epoch": 0.5263312594915005,
      "grad_norm": 5.311732769012451,
      "learning_rate": 5.429090751021704e-06,
      "loss": 2.1665,
      "step": 1343
    },
    {
      "epoch": 0.5267231666095136,
      "grad_norm": 5.8789167404174805,
      "learning_rate": 5.3607733197664436e-06,
      "loss": 1.8624,
      "step": 1344
    },
    {
      "epoch": 0.5271150737275265,
      "grad_norm": 5.627329349517822,
      "learning_rate": 5.2928766134254345e-06,
      "loss": 1.7578,
      "step": 1345
    },
    {
      "epoch": 0.5275069808455396,
      "grad_norm": 5.913056373596191,
      "learning_rate": 5.225400933837954e-06,
      "loss": 1.6379,
      "step": 1346
    },
    {
      "epoch": 0.5278988879635527,
      "grad_norm": 6.612465858459473,
      "learning_rate": 5.158346580971573e-06,
      "loss": 1.7578,
      "step": 1347
    },
    {
      "epoch": 0.5282907950815656,
      "grad_norm": 7.556913375854492,
      "learning_rate": 5.091713852920854e-06,
      "loss": 1.5792,
      "step": 1348
    },
    {
      "epoch": 0.5286827021995787,
      "grad_norm": 9.047926902770996,
      "learning_rate": 5.025503045905933e-06,
      "loss": 1.9531,
      "step": 1349
    },
    {
      "epoch": 0.5290746093175918,
      "grad_norm": 17.35049819946289,
      "learning_rate": 4.959714454271369e-06,
      "loss": 2.2852,
      "step": 1350
    },
    {
      "epoch": 0.5294665164356047,
      "grad_norm": 1.3906933069229126,
      "learning_rate": 4.8943483704846475e-06,
      "loss": 0.9542,
      "step": 1351
    },
    {
      "epoch": 0.5298584235536178,
      "grad_norm": 1.4756219387054443,
      "learning_rate": 4.829405085134997e-06,
      "loss": 0.3774,
      "step": 1352
    },
    {
      "epoch": 0.5302503306716309,
      "grad_norm": 1.1823430061340332,
      "learning_rate": 4.764884886932086e-06,
      "loss": 0.3929,
      "step": 1353
    },
    {
      "epoch": 0.5306422377896438,
      "grad_norm": 2.0313920974731445,
      "learning_rate": 4.700788062704687e-06,
      "loss": 1.1821,
      "step": 1354
    },
    {
      "epoch": 0.5310341449076569,
      "grad_norm": 1.7937896251678467,
      "learning_rate": 4.6371148973994525e-06,
      "loss": 0.8926,
      "step": 1355
    },
    {
      "epoch": 0.53142605202567,
      "grad_norm": 2.956267833709717,
      "learning_rate": 4.573865674079625e-06,
      "loss": 1.9447,
      "step": 1356
    },
    {
      "epoch": 0.5318179591436829,
      "grad_norm": 2.713437557220459,
      "learning_rate": 4.511040673923828e-06,
      "loss": 2.0029,
      "step": 1357
    },
    {
      "epoch": 0.532209866261696,
      "grad_norm": 2.6672322750091553,
      "learning_rate": 4.448640176224694e-06,
      "loss": 1.9151,
      "step": 1358
    },
    {
      "epoch": 0.532601773379709,
      "grad_norm": 2.6359589099884033,
      "learning_rate": 4.386664458387779e-06,
      "loss": 1.5954,
      "step": 1359
    },
    {
      "epoch": 0.532993680497722,
      "grad_norm": 3.18007493019104,
      "learning_rate": 4.325113795930203e-06,
      "loss": 1.9893,
      "step": 1360
    },
    {
      "epoch": 0.5333855876157351,
      "grad_norm": 2.9305989742279053,
      "learning_rate": 4.263988462479484e-06,
      "loss": 1.9884,
      "step": 1361
    },
    {
      "epoch": 0.5337774947337481,
      "grad_norm": 3.3843114376068115,
      "learning_rate": 4.203288729772326e-06,
      "loss": 2.1543,
      "step": 1362
    },
    {
      "epoch": 0.5341694018517611,
      "grad_norm": 3.555039882659912,
      "learning_rate": 4.143014867653383e-06,
      "loss": 1.9616,
      "step": 1363
    },
    {
      "epoch": 0.5345613089697742,
      "grad_norm": 3.1563668251037598,
      "learning_rate": 4.083167144074073e-06,
      "loss": 1.7015,
      "step": 1364
    },
    {
      "epoch": 0.5349532160877872,
      "grad_norm": 3.5887413024902344,
      "learning_rate": 4.023745825091407e-06,
      "loss": 1.6368,
      "step": 1365
    },
    {
      "epoch": 0.5353451232058002,
      "grad_norm": 3.533641815185547,
      "learning_rate": 3.964751174866765e-06,
      "loss": 2.0182,
      "step": 1366
    },
    {
      "epoch": 0.5357370303238133,
      "grad_norm": 3.702850341796875,
      "learning_rate": 3.906183455664725e-06,
      "loss": 1.9329,
      "step": 1367
    },
    {
      "epoch": 0.5361289374418263,
      "grad_norm": 4.071200847625732,
      "learning_rate": 3.84804292785198e-06,
      "loss": 1.9623,
      "step": 1368
    },
    {
      "epoch": 0.5365208445598393,
      "grad_norm": 3.3908069133758545,
      "learning_rate": 3.7903298498960572e-06,
      "loss": 1.6228,
      "step": 1369
    },
    {
      "epoch": 0.5369127516778524,
      "grad_norm": 3.7330358028411865,
      "learning_rate": 3.7330444783642338e-06,
      "loss": 1.5727,
      "step": 1370
    },
    {
      "epoch": 0.5373046587958654,
      "grad_norm": 4.013121604919434,
      "learning_rate": 3.676187067922421e-06,
      "loss": 2.0572,
      "step": 1371
    },
    {
      "epoch": 0.5376965659138784,
      "grad_norm": 3.6991512775421143,
      "learning_rate": 3.619757871333973e-06,
      "loss": 1.6562,
      "step": 1372
    },
    {
      "epoch": 0.5380884730318914,
      "grad_norm": 3.6997110843658447,
      "learning_rate": 3.563757139458579e-06,
      "loss": 1.5704,
      "step": 1373
    },
    {
      "epoch": 0.5384803801499045,
      "grad_norm": 4.688897609710693,
      "learning_rate": 3.5081851212512175e-06,
      "loss": 2.1387,
      "step": 1374
    },
    {
      "epoch": 0.5388722872679175,
      "grad_norm": 4.277041912078857,
      "learning_rate": 3.4530420637609363e-06,
      "loss": 2.0295,
      "step": 1375
    },
    {
      "epoch": 0.5392641943859305,
      "grad_norm": 4.16665506362915,
      "learning_rate": 3.3983282121298086e-06,
      "loss": 1.6446,
      "step": 1376
    },
    {
      "epoch": 0.5396561015039436,
      "grad_norm": 4.408489227294922,
      "learning_rate": 3.3440438095919126e-06,
      "loss": 2.1116,
      "step": 1377
    },
    {
      "epoch": 0.5400480086219566,
      "grad_norm": 4.414140224456787,
      "learning_rate": 3.290189097472096e-06,
      "loss": 2.0354,
      "step": 1378
    },
    {
      "epoch": 0.5404399157399696,
      "grad_norm": 4.812220096588135,
      "learning_rate": 3.236764315185037e-06,
      "loss": 2.1071,
      "step": 1379
    },
    {
      "epoch": 0.5408318228579827,
      "grad_norm": 4.569962978363037,
      "learning_rate": 3.1837697002341293e-06,
      "loss": 2.0292,
      "step": 1380
    },
    {
      "epoch": 0.5412237299759957,
      "grad_norm": 4.310214042663574,
      "learning_rate": 3.131205488210409e-06,
      "loss": 1.7732,
      "step": 1381
    },
    {
      "epoch": 0.5416156370940087,
      "grad_norm": 4.937644958496094,
      "learning_rate": 3.0790719127915646e-06,
      "loss": 2.1476,
      "step": 1382
    },
    {
      "epoch": 0.5420075442120218,
      "grad_norm": 4.747223854064941,
      "learning_rate": 3.0273692057408265e-06,
      "loss": 2.4155,
      "step": 1383
    },
    {
      "epoch": 0.5423994513300348,
      "grad_norm": 4.54340934753418,
      "learning_rate": 2.976097596905969e-06,
      "loss": 1.9156,
      "step": 1384
    },
    {
      "epoch": 0.5427913584480478,
      "grad_norm": 4.812037467956543,
      "learning_rate": 2.9252573142183326e-06,
      "loss": 1.954,
      "step": 1385
    },
    {
      "epoch": 0.5431832655660609,
      "grad_norm": 4.922847270965576,
      "learning_rate": 2.874848583691714e-06,
      "loss": 2.2959,
      "step": 1386
    },
    {
      "epoch": 0.5435751726840738,
      "grad_norm": 5.3064751625061035,
      "learning_rate": 2.8248716294214774e-06,
      "loss": 2.0994,
      "step": 1387
    },
    {
      "epoch": 0.5439670798020869,
      "grad_norm": 5.1081223487854,
      "learning_rate": 2.7753266735834338e-06,
      "loss": 2.2288,
      "step": 1388
    },
    {
      "epoch": 0.5443589869201,
      "grad_norm": 5.7183942794799805,
      "learning_rate": 2.7262139364329643e-06,
      "loss": 2.2891,
      "step": 1389
    },
    {
      "epoch": 0.5447508940381129,
      "grad_norm": 5.003108501434326,
      "learning_rate": 2.677533636303964e-06,
      "loss": 2.18,
      "step": 1390
    },
    {
      "epoch": 0.545142801156126,
      "grad_norm": 5.089848041534424,
      "learning_rate": 2.6292859896079213e-06,
      "loss": 2.1952,
      "step": 1391
    },
    {
      "epoch": 0.5455347082741391,
      "grad_norm": 4.803051471710205,
      "learning_rate": 2.581471210832931e-06,
      "loss": 1.5949,
      "step": 1392
    },
    {
      "epoch": 0.545926615392152,
      "grad_norm": 5.185389995574951,
      "learning_rate": 2.5340895125427364e-06,
      "loss": 2.1897,
      "step": 1393
    },
    {
      "epoch": 0.5463185225101651,
      "grad_norm": 5.408966064453125,
      "learning_rate": 2.4871411053757898e-06,
      "loss": 2.1241,
      "step": 1394
    },
    {
      "epoch": 0.5467104296281782,
      "grad_norm": 5.5310378074646,
      "learning_rate": 2.440626198044327e-06,
      "loss": 2.0659,
      "step": 1395
    },
    {
      "epoch": 0.5471023367461911,
      "grad_norm": 6.339371681213379,
      "learning_rate": 2.394544997333437e-06,
      "loss": 1.6685,
      "step": 1396
    },
    {
      "epoch": 0.5474942438642042,
      "grad_norm": 7.643269062042236,
      "learning_rate": 2.3488977081001394e-06,
      "loss": 2.1111,
      "step": 1397
    },
    {
      "epoch": 0.5478861509822173,
      "grad_norm": 7.755165100097656,
      "learning_rate": 2.3036845332724543e-06,
      "loss": 2.0256,
      "step": 1398
    },
    {
      "epoch": 0.5482780581002302,
      "grad_norm": 8.814311981201172,
      "learning_rate": 2.2589056738485324e-06,
      "loss": 1.6392,
      "step": 1399
    },
    {
      "epoch": 0.5486699652182433,
      "grad_norm": 13.345653533935547,
      "learning_rate": 2.2145613288957478e-06,
      "loss": 2.4261,
      "step": 1400
    },
    {
      "epoch": 0.5490618723362564,
      "grad_norm": 1.234164834022522,
      "learning_rate": 2.170651695549786e-06,
      "loss": 0.9774,
      "step": 1401
    },
    {
      "epoch": 0.5494537794542693,
      "grad_norm": 1.5545003414154053,
      "learning_rate": 2.1271769690138332e-06,
      "loss": 0.6109,
      "step": 1402
    },
    {
      "epoch": 0.5498456865722824,
      "grad_norm": 1.610559344291687,
      "learning_rate": 2.084137342557646e-06,
      "loss": 0.8989,
      "step": 1403
    },
    {
      "epoch": 0.5502375936902953,
      "grad_norm": 1.8761323690414429,
      "learning_rate": 2.0415330075166937e-06,
      "loss": 0.9448,
      "step": 1404
    },
    {
      "epoch": 0.5506295008083084,
      "grad_norm": 2.3050780296325684,
      "learning_rate": 1.9993641532913833e-06,
      "loss": 1.4417,
      "step": 1405
    },
    {
      "epoch": 0.5510214079263215,
      "grad_norm": 1.8930176496505737,
      "learning_rate": 1.9576309673461357e-06,
      "loss": 1.2358,
      "step": 1406
    },
    {
      "epoch": 0.5514133150443344,
      "grad_norm": 2.2075400352478027,
      "learning_rate": 1.916333635208556e-06,
      "loss": 1.1255,
      "step": 1407
    },
    {
      "epoch": 0.5518052221623475,
      "grad_norm": 2.6658432483673096,
      "learning_rate": 1.8754723404686425e-06,
      "loss": 1.471,
      "step": 1408
    },
    {
      "epoch": 0.5521971292803606,
      "grad_norm": 2.6441118717193604,
      "learning_rate": 1.8350472647780116e-06,
      "loss": 1.6123,
      "step": 1409
    },
    {
      "epoch": 0.5525890363983735,
      "grad_norm": 3.158393621444702,
      "learning_rate": 1.7950585878489856e-06,
      "loss": 1.8184,
      "step": 1410
    },
    {
      "epoch": 0.5529809435163866,
      "grad_norm": 3.430659532546997,
      "learning_rate": 1.7555064874538397e-06,
      "loss": 1.8103,
      "step": 1411
    },
    {
      "epoch": 0.5533728506343997,
      "grad_norm": 3.3335251808166504,
      "learning_rate": 1.7163911394240672e-06,
      "loss": 1.7969,
      "step": 1412
    },
    {
      "epoch": 0.5537647577524126,
      "grad_norm": 3.4480910301208496,
      "learning_rate": 1.6777127176495043e-06,
      "loss": 2.1869,
      "step": 1413
    },
    {
      "epoch": 0.5541566648704257,
      "grad_norm": 3.6782875061035156,
      "learning_rate": 1.6394713940776296e-06,
      "loss": 1.9936,
      "step": 1414
    },
    {
      "epoch": 0.5545485719884388,
      "grad_norm": 3.5967817306518555,
      "learning_rate": 1.6016673387127646e-06,
      "loss": 2.0893,
      "step": 1415
    },
    {
      "epoch": 0.5549404791064517,
      "grad_norm": 3.932044267654419,
      "learning_rate": 1.5643007196153302e-06,
      "loss": 2.0624,
      "step": 1416
    },
    {
      "epoch": 0.5553323862244648,
      "grad_norm": 4.180208683013916,
      "learning_rate": 1.5273717029010925e-06,
      "loss": 1.834,
      "step": 1417
    },
    {
      "epoch": 0.5557242933424779,
      "grad_norm": 3.9711670875549316,
      "learning_rate": 1.4908804527404286e-06,
      "loss": 1.9743,
      "step": 1418
    },
    {
      "epoch": 0.5561162004604908,
      "grad_norm": 3.842787742614746,
      "learning_rate": 1.4548271313575835e-06,
      "loss": 1.7233,
      "step": 1419
    },
    {
      "epoch": 0.5565081075785039,
      "grad_norm": 3.580495595932007,
      "learning_rate": 1.4192118990299707e-06,
      "loss": 1.9272,
      "step": 1420
    },
    {
      "epoch": 0.556900014696517,
      "grad_norm": 4.917481422424316,
      "learning_rate": 1.3840349140874619e-06,
      "loss": 2.0819,
      "step": 1421
    },
    {
      "epoch": 0.5572919218145299,
      "grad_norm": 3.9283082485198975,
      "learning_rate": 1.3492963329116537e-06,
      "loss": 1.8576,
      "step": 1422
    },
    {
      "epoch": 0.557683828932543,
      "grad_norm": 4.298456192016602,
      "learning_rate": 1.3149963099352014e-06,
      "loss": 2.1071,
      "step": 1423
    },
    {
      "epoch": 0.558075736050556,
      "grad_norm": 3.6550862789154053,
      "learning_rate": 1.2811349976411202e-06,
      "loss": 1.4015,
      "step": 1424
    },
    {
      "epoch": 0.558467643168569,
      "grad_norm": 4.322780132293701,
      "learning_rate": 1.2477125465620853e-06,
      "loss": 1.7561,
      "step": 1425
    },
    {
      "epoch": 0.5588595502865821,
      "grad_norm": 4.455068111419678,
      "learning_rate": 1.2147291052798216e-06,
      "loss": 1.9483,
      "step": 1426
    },
    {
      "epoch": 0.5592514574045951,
      "grad_norm": 4.613096237182617,
      "learning_rate": 1.1821848204243814e-06,
      "loss": 1.8634,
      "step": 1427
    },
    {
      "epoch": 0.5596433645226081,
      "grad_norm": 4.524038791656494,
      "learning_rate": 1.1500798366735233e-06,
      "loss": 1.7424,
      "step": 1428
    },
    {
      "epoch": 0.5600352716406212,
      "grad_norm": 4.7389445304870605,
      "learning_rate": 1.1184142967520794e-06,
      "loss": 1.9841,
      "step": 1429
    },
    {
      "epoch": 0.5604271787586342,
      "grad_norm": 4.588124752044678,
      "learning_rate": 1.0871883414312777e-06,
      "loss": 2.1098,
      "step": 1430
    },
    {
      "epoch": 0.5608190858766472,
      "grad_norm": 4.397920608520508,
      "learning_rate": 1.0564021095281652e-06,
      "loss": 1.9755,
      "step": 1431
    },
    {
      "epoch": 0.5612109929946603,
      "grad_norm": 4.566826820373535,
      "learning_rate": 1.0260557379049519e-06,
      "loss": 1.8584,
      "step": 1432
    },
    {
      "epoch": 0.5616029001126733,
      "grad_norm": 4.860532283782959,
      "learning_rate": 9.96149361468457e-07,
      "loss": 2.0773,
      "step": 1433
    },
    {
      "epoch": 0.5619948072306863,
      "grad_norm": 4.719742774963379,
      "learning_rate": 9.66683113169431e-07,
      "loss": 1.9875,
      "step": 1434
    },
    {
      "epoch": 0.5623867143486994,
      "grad_norm": 4.66433572769165,
      "learning_rate": 9.376571240020227e-07,
      "loss": 1.7513,
      "step": 1435
    },
    {
      "epoch": 0.5627786214667124,
      "grad_norm": 4.908422470092773,
      "learning_rate": 9.090715230031688e-07,
      "loss": 2.1587,
      "step": 1436
    },
    {
      "epoch": 0.5631705285847254,
      "grad_norm": 5.3385748863220215,
      "learning_rate": 8.809264372520609e-07,
      "loss": 2.0308,
      "step": 1437
    },
    {
      "epoch": 0.5635624357027385,
      "grad_norm": 5.4692840576171875,
      "learning_rate": 8.532219918695128e-07,
      "loss": 2.0405,
      "step": 1438
    },
    {
      "epoch": 0.5639543428207515,
      "grad_norm": 5.025905609130859,
      "learning_rate": 8.259583100174606e-07,
      "loss": 1.8819,
      "step": 1439
    },
    {
      "epoch": 0.5643462499387645,
      "grad_norm": 4.762441635131836,
      "learning_rate": 7.991355128984079e-07,
      "loss": 1.7986,
      "step": 1440
    },
    {
      "epoch": 0.5647381570567775,
      "grad_norm": 4.794429302215576,
      "learning_rate": 7.727537197548707e-07,
      "loss": 1.8309,
      "step": 1441
    },
    {
      "epoch": 0.5651300641747906,
      "grad_norm": 4.900668621063232,
      "learning_rate": 7.468130478688218e-07,
      "loss": 1.6892,
      "step": 1442
    },
    {
      "epoch": 0.5655219712928036,
      "grad_norm": 5.534628868103027,
      "learning_rate": 7.213136125612586e-07,
      "loss": 1.8748,
      "step": 1443
    },
    {
      "epoch": 0.5659138784108166,
      "grad_norm": 5.953134536743164,
      "learning_rate": 6.962555271915805e-07,
      "loss": 2.0223,
      "step": 1444
    },
    {
      "epoch": 0.5663057855288297,
      "grad_norm": 5.934267997741699,
      "learning_rate": 6.716389031571568e-07,
      "loss": 2.1515,
      "step": 1445
    },
    {
      "epoch": 0.5666976926468427,
      "grad_norm": 6.634733200073242,
      "learning_rate": 6.474638498928265e-07,
      "loss": 1.918,
      "step": 1446
    },
    {
      "epoch": 0.5670895997648557,
      "grad_norm": 6.247616291046143,
      "learning_rate": 6.237304748703543e-07,
      "loss": 1.7692,
      "step": 1447
    },
    {
      "epoch": 0.5674815068828688,
      "grad_norm": 6.24067497253418,
      "learning_rate": 6.004388835980423e-07,
      "loss": 1.2688,
      "step": 1448
    },
    {
      "epoch": 0.5678734140008818,
      "grad_norm": 9.73441219329834,
      "learning_rate": 5.77589179620186e-07,
      "loss": 2.596,
      "step": 1449
    },
    {
      "epoch": 0.5682653211188948,
      "grad_norm": 8.940555572509766,
      "learning_rate": 5.55181464516652e-07,
      "loss": 1.7786,
      "step": 1450
    },
    {
      "epoch": 0.5686572282369079,
      "grad_norm": 1.2116379737854004,
      "learning_rate": 5.332158379024122e-07,
      "loss": 0.8872,
      "step": 1451
    },
    {
      "epoch": 0.5690491353549209,
      "grad_norm": 1.2707936763763428,
      "learning_rate": 5.116923974270993e-07,
      "loss": 0.5679,
      "step": 1452
    },
    {
      "epoch": 0.5694410424729339,
      "grad_norm": 1.5529366731643677,
      "learning_rate": 4.906112387745965e-07,
      "loss": 0.7058,
      "step": 1453
    },
    {
      "epoch": 0.569832949590947,
      "grad_norm": 1.6745681762695312,
      "learning_rate": 4.6997245566257064e-07,
      "loss": 1.1799,
      "step": 1454
    },
    {
      "epoch": 0.57022485670896,
      "grad_norm": 2.2588400840759277,
      "learning_rate": 4.497761398421063e-07,
      "loss": 1.4141,
      "step": 1455
    },
    {
      "epoch": 0.570616763826973,
      "grad_norm": 2.2598230838775635,
      "learning_rate": 4.3002238109723927e-07,
      "loss": 1.5724,
      "step": 1456
    },
    {
      "epoch": 0.5710086709449861,
      "grad_norm": 3.0852503776550293,
      "learning_rate": 4.107112672446123e-07,
      "loss": 2.0041,
      "step": 1457
    },
    {
      "epoch": 0.571400578062999,
      "grad_norm": 2.7703232765197754,
      "learning_rate": 3.9184288413306456e-07,
      "loss": 1.9703,
      "step": 1458
    },
    {
      "epoch": 0.5717924851810121,
      "grad_norm": 3.482165813446045,
      "learning_rate": 3.734173156432208e-07,
      "loss": 2.2029,
      "step": 1459
    },
    {
      "epoch": 0.5721843922990252,
      "grad_norm": 2.7262227535247803,
      "learning_rate": 3.554346436871581e-07,
      "loss": 1.5757,
      "step": 1460
    },
    {
      "epoch": 0.5725762994170381,
      "grad_norm": 2.9860002994537354,
      "learning_rate": 3.3789494820803957e-07,
      "loss": 1.6978,
      "step": 1461
    },
    {
      "epoch": 0.5729682065350512,
      "grad_norm": 3.410658121109009,
      "learning_rate": 3.2079830717972606e-07,
      "loss": 2.1118,
      "step": 1462
    },
    {
      "epoch": 0.5733601136530643,
      "grad_norm": 3.4254398345947266,
      "learning_rate": 3.041447966064648e-07,
      "loss": 2.051,
      "step": 1463
    },
    {
      "epoch": 0.5737520207710772,
      "grad_norm": 3.6203994750976562,
      "learning_rate": 2.8793449052254563e-07,
      "loss": 1.8162,
      "step": 1464
    },
    {
      "epoch": 0.5741439278890903,
      "grad_norm": 3.4394781589508057,
      "learning_rate": 2.721674609919345e-07,
      "loss": 1.8449,
      "step": 1465
    },
    {
      "epoch": 0.5745358350071034,
      "grad_norm": 3.5789806842803955,
      "learning_rate": 2.568437781080069e-07,
      "loss": 1.9301,
      "step": 1466
    },
    {
      "epoch": 0.5749277421251163,
      "grad_norm": 3.7726640701293945,
      "learning_rate": 2.4196350999320384e-07,
      "loss": 1.6961,
      "step": 1467
    },
    {
      "epoch": 0.5753196492431294,
      "grad_norm": 3.3344314098358154,
      "learning_rate": 2.275267227987321e-07,
      "loss": 1.8552,
      "step": 1468
    },
    {
      "epoch": 0.5757115563611425,
      "grad_norm": 3.9859609603881836,
      "learning_rate": 2.135334807042866e-07,
      "loss": 1.8618,
      "step": 1469
    },
    {
      "epoch": 0.5761034634791554,
      "grad_norm": 3.8872029781341553,
      "learning_rate": 1.9998384591773944e-07,
      "loss": 1.8088,
      "step": 1470
    },
    {
      "epoch": 0.5764953705971685,
      "grad_norm": 3.877041816711426,
      "learning_rate": 1.8687787867489592e-07,
      "loss": 1.5552,
      "step": 1471
    },
    {
      "epoch": 0.5768872777151816,
      "grad_norm": 4.494336128234863,
      "learning_rate": 1.7421563723919454e-07,
      "loss": 2.2397,
      "step": 1472
    },
    {
      "epoch": 0.5772791848331945,
      "grad_norm": 4.406491279602051,
      "learning_rate": 1.6199717790145174e-07,
      "loss": 2.1255,
      "step": 1473
    },
    {
      "epoch": 0.5776710919512076,
      "grad_norm": 4.008056163787842,
      "learning_rate": 1.5022255497962879e-07,
      "loss": 1.84,
      "step": 1474
    },
    {
      "epoch": 0.5780629990692205,
      "grad_norm": 4.361143112182617,
      "learning_rate": 1.3889182081860962e-07,
      "loss": 2.1469,
      "step": 1475
    },
    {
      "epoch": 0.5784549061872336,
      "grad_norm": 4.139244556427002,
      "learning_rate": 1.2800502578991235e-07,
      "loss": 1.8226,
      "step": 1476
    },
    {
      "epoch": 0.5788468133052467,
      "grad_norm": 4.491837024688721,
      "learning_rate": 1.1756221829148928e-07,
      "loss": 2.0087,
      "step": 1477
    },
    {
      "epoch": 0.5792387204232596,
      "grad_norm": 4.200547695159912,
      "learning_rate": 1.0756344474753821e-07,
      "loss": 1.7487,
      "step": 1478
    },
    {
      "epoch": 0.5796306275412727,
      "grad_norm": 4.534806251525879,
      "learning_rate": 9.800874960826933e-08,
      "loss": 2.0131,
      "step": 1479
    },
    {
      "epoch": 0.5800225346592858,
      "grad_norm": 4.71172571182251,
      "learning_rate": 8.889817534969425e-08,
      "loss": 2.2639,
      "step": 1480
    },
    {
      "epoch": 0.5804144417772987,
      "grad_norm": 4.4803900718688965,
      "learning_rate": 8.023176247348163e-08,
      "loss": 1.9476,
      "step": 1481
    },
    {
      "epoch": 0.5808063488953118,
      "grad_norm": 4.689031600952148,
      "learning_rate": 7.200954950673522e-08,
      "loss": 2.2785,
      "step": 1482
    },
    {
      "epoch": 0.5811982560133249,
      "grad_norm": 4.634332656860352,
      "learning_rate": 6.423157300184946e-08,
      "loss": 2.4399,
      "step": 1483
    },
    {
      "epoch": 0.5815901631313378,
      "grad_norm": 5.102287292480469,
      "learning_rate": 5.6897867536331864e-08,
      "loss": 2.2395,
      "step": 1484
    },
    {
      "epoch": 0.5819820702493509,
      "grad_norm": 4.995741844177246,
      "learning_rate": 5.000846571264761e-08,
      "loss": 2.1091,
      "step": 1485
    },
    {
      "epoch": 0.582373977367364,
      "grad_norm": 4.875114440917969,
      "learning_rate": 4.35633981580974e-08,
      "loss": 2.1574,
      "step": 1486
    },
    {
      "epoch": 0.5827658844853769,
      "grad_norm": 4.840078353881836,
      "learning_rate": 3.756269352462871e-08,
      "loss": 2.1515,
      "step": 1487
    },
    {
      "epoch": 0.58315779160339,
      "grad_norm": 5.044447898864746,
      "learning_rate": 3.20063784888025e-08,
      "loss": 2.2336,
      "step": 1488
    },
    {
      "epoch": 0.583549698721403,
      "grad_norm": 5.046810150146484,
      "learning_rate": 2.6894477751548964e-08,
      "loss": 2.1112,
      "step": 1489
    },
    {
      "epoch": 0.583941605839416,
      "grad_norm": 4.999392509460449,
      "learning_rate": 2.222701403818972e-08,
      "loss": 2.16,
      "step": 1490
    },
    {
      "epoch": 0.5843335129574291,
      "grad_norm": 4.994463920593262,
      "learning_rate": 1.8004008098226887e-08,
      "loss": 1.9523,
      "step": 1491
    },
    {
      "epoch": 0.5847254200754421,
      "grad_norm": 4.6696600914001465,
      "learning_rate": 1.4225478705309769e-08,
      "loss": 1.477,
      "step": 1492
    },
    {
      "epoch": 0.5851173271934551,
      "grad_norm": 5.579797744750977,
      "learning_rate": 1.0891442657134932e-08,
      "loss": 1.9963,
      "step": 1493
    },
    {
      "epoch": 0.5855092343114682,
      "grad_norm": 6.292161464691162,
      "learning_rate": 8.001914775401798e-09,
      "loss": 2.2439,
      "step": 1494
    },
    {
      "epoch": 0.5859011414294812,
      "grad_norm": 5.767752170562744,
      "learning_rate": 5.5569079056794206e-09,
      "loss": 1.9108,
      "step": 1495
    },
    {
      "epoch": 0.5862930485474942,
      "grad_norm": 5.550935745239258,
      "learning_rate": 3.5564329174064824e-09,
      "loss": 1.4977,
      "step": 1496
    },
    {
      "epoch": 0.5866849556655073,
      "grad_norm": 6.293521881103516,
      "learning_rate": 2.0004987038246824e-09,
      "loss": 1.9414,
      "step": 1497
    },
    {
      "epoch": 0.5870768627835203,
      "grad_norm": 7.220438480377197,
      "learning_rate": 8.891121819565306e-10,
      "loss": 1.8714,
      "step": 1498
    },
    {
      "epoch": 0.5874687699015333,
      "grad_norm": 7.746157646179199,
      "learning_rate": 2.2227829252763344e-10,
      "loss": 1.712,
      "step": 1499
    },
    {
      "epoch": 0.5878606770195464,
      "grad_norm": 8.74112319946289,
      "learning_rate": 0.0,
      "loss": 1.5,
      "step": 1500
    },
    {
      "epoch": 0.5878606770195464,
      "eval_loss": 1.8475549221038818,
      "eval_runtime": 37.9424,
      "eval_samples_per_second": 56.638,
      "eval_steps_per_second": 28.332,
      "step": 1500
    }
  ],
  "logging_steps": 1,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 375,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.400512277761229e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}