{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 24.0,
  "eval_steps": 500,
  "global_step": 6000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.1,
      "grad_norm": 0.6611918807029724,
      "learning_rate": 5.319148936170213e-05,
      "loss": 0.6932,
      "step": 25
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9467485547065735,
      "learning_rate": 0.00010638297872340425,
      "loss": 0.6912,
      "step": 50
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9465051889419556,
      "learning_rate": 0.00015957446808510637,
      "loss": 0.6893,
      "step": 75
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5430934429168701,
      "learning_rate": 0.0002127659574468085,
      "loss": 0.6804,
      "step": 100
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6103107929229736,
      "learning_rate": 0.00026595744680851064,
      "loss": 0.6806,
      "step": 125
    },
    {
      "epoch": 0.6,
      "grad_norm": 2.3773200511932373,
      "learning_rate": 0.00031914893617021275,
      "loss": 0.6801,
      "step": 150
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8832203149795532,
      "learning_rate": 0.0003723404255319149,
      "loss": 0.6791,
      "step": 175
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3350876569747925,
      "learning_rate": 0.0003992081821181128,
      "loss": 0.6746,
      "step": 200
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5796219110488892,
      "learning_rate": 0.0003975585615308479,
      "loss": 0.6771,
      "step": 225
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.884006381034851,
      "learning_rate": 0.00039590894094358297,
      "loss": 0.6649,
      "step": 250
    },
    {
      "epoch": 1.1,
      "grad_norm": 3.286440372467041,
      "learning_rate": 0.0003942593203563181,
      "loss": 0.6388,
      "step": 275
    },
    {
      "epoch": 1.2,
      "grad_norm": 8.244946479797363,
      "learning_rate": 0.0003926096997690532,
      "loss": 0.6238,
      "step": 300
    },
    {
      "epoch": 1.3,
      "grad_norm": 4.265683650970459,
      "learning_rate": 0.00039096007918178817,
      "loss": 0.6383,
      "step": 325
    },
    {
      "epoch": 1.4,
      "grad_norm": 4.115826606750488,
      "learning_rate": 0.0003893104585945233,
      "loss": 0.6194,
      "step": 350
    },
    {
      "epoch": 1.5,
      "grad_norm": 5.694250583648682,
      "learning_rate": 0.0003876608380072583,
      "loss": 0.6324,
      "step": 375
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.463121175765991,
      "learning_rate": 0.00038601121741999343,
      "loss": 0.621,
      "step": 400
    },
    {
      "epoch": 1.7,
      "grad_norm": 4.582865238189697,
      "learning_rate": 0.0003843615968327285,
      "loss": 0.6116,
      "step": 425
    },
    {
      "epoch": 1.8,
      "grad_norm": 11.996281623840332,
      "learning_rate": 0.0003827119762454636,
      "loss": 0.6393,
      "step": 450
    },
    {
      "epoch": 1.9,
      "grad_norm": 3.0407373905181885,
      "learning_rate": 0.00038106235565819863,
      "loss": 0.628,
      "step": 475
    },
    {
      "epoch": 2.0,
      "grad_norm": 2.917588233947754,
      "learning_rate": 0.0003794127350709337,
      "loss": 0.6078,
      "step": 500
    },
    {
      "epoch": 2.1,
      "grad_norm": 4.748379707336426,
      "learning_rate": 0.0003777631144836688,
      "loss": 0.4899,
      "step": 525
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.8076977729797363,
      "learning_rate": 0.00037611349389640383,
      "loss": 0.5086,
      "step": 550
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.2440714836120605,
      "learning_rate": 0.00037446387330913894,
      "loss": 0.5327,
      "step": 575
    },
    {
      "epoch": 2.4,
      "grad_norm": 7.110438346862793,
      "learning_rate": 0.000372814252721874,
      "loss": 0.5436,
      "step": 600
    },
    {
      "epoch": 2.5,
      "grad_norm": 5.46150541305542,
      "learning_rate": 0.00037116463213460903,
      "loss": 0.5294,
      "step": 625
    },
    {
      "epoch": 2.6,
      "grad_norm": 5.136163234710693,
      "learning_rate": 0.00036951501154734414,
      "loss": 0.5245,
      "step": 650
    },
    {
      "epoch": 2.7,
      "grad_norm": 8.735346794128418,
      "learning_rate": 0.0003678653909600792,
      "loss": 0.5449,
      "step": 675
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.922825574874878,
      "learning_rate": 0.0003662157703728143,
      "loss": 0.5406,
      "step": 700
    },
    {
      "epoch": 2.9,
      "grad_norm": 7.744819641113281,
      "learning_rate": 0.00036456614978554934,
      "loss": 0.5447,
      "step": 725
    },
    {
      "epoch": 3.0,
      "grad_norm": 11.6185884475708,
      "learning_rate": 0.00036291652919828444,
      "loss": 0.5195,
      "step": 750
    },
    {
      "epoch": 3.1,
      "grad_norm": 3.73836088180542,
      "learning_rate": 0.00036126690861101944,
      "loss": 0.3824,
      "step": 775
    },
    {
      "epoch": 3.2,
      "grad_norm": 14.850343704223633,
      "learning_rate": 0.00035961728802375454,
      "loss": 0.4071,
      "step": 800
    },
    {
      "epoch": 3.3,
      "grad_norm": 5.7157440185546875,
      "learning_rate": 0.0003579676674364896,
      "loss": 0.3986,
      "step": 825
    },
    {
      "epoch": 3.4,
      "grad_norm": 12.418399810791016,
      "learning_rate": 0.0003563180468492247,
      "loss": 0.4282,
      "step": 850
    },
    {
      "epoch": 3.5,
      "grad_norm": 12.793001174926758,
      "learning_rate": 0.0003546684262619598,
      "loss": 0.4822,
      "step": 875
    },
    {
      "epoch": 3.6,
      "grad_norm": 6.489450931549072,
      "learning_rate": 0.00035301880567469485,
      "loss": 0.4239,
      "step": 900
    },
    {
      "epoch": 3.7,
      "grad_norm": 5.365822792053223,
      "learning_rate": 0.0003513691850874299,
      "loss": 0.421,
      "step": 925
    },
    {
      "epoch": 3.8,
      "grad_norm": 12.643745422363281,
      "learning_rate": 0.00034971956450016495,
      "loss": 0.3964,
      "step": 950
    },
    {
      "epoch": 3.9,
      "grad_norm": 14.334024429321289,
      "learning_rate": 0.00034806994391290005,
      "loss": 0.4634,
      "step": 975
    },
    {
      "epoch": 4.0,
      "grad_norm": 6.819091320037842,
      "learning_rate": 0.0003464203233256351,
      "loss": 0.4139,
      "step": 1000
    },
    {
      "epoch": 4.1,
      "grad_norm": 8.29238224029541,
      "learning_rate": 0.0003447707027383702,
      "loss": 0.2695,
      "step": 1025
    },
    {
      "epoch": 4.2,
      "grad_norm": 5.984206676483154,
      "learning_rate": 0.00034312108215110525,
      "loss": 0.2653,
      "step": 1050
    },
    {
      "epoch": 4.3,
      "grad_norm": 5.9425435066223145,
      "learning_rate": 0.0003414714615638403,
      "loss": 0.2982,
      "step": 1075
    },
    {
      "epoch": 4.4,
      "grad_norm": 7.1877593994140625,
      "learning_rate": 0.0003398218409765754,
      "loss": 0.3309,
      "step": 1100
    },
    {
      "epoch": 4.5,
      "grad_norm": 17.83046531677246,
      "learning_rate": 0.00033817222038931045,
      "loss": 0.3467,
      "step": 1125
    },
    {
      "epoch": 4.6,
      "grad_norm": 4.865128517150879,
      "learning_rate": 0.00033652259980204556,
      "loss": 0.3117,
      "step": 1150
    },
    {
      "epoch": 4.7,
      "grad_norm": 31.427154541015625,
      "learning_rate": 0.0003348729792147806,
      "loss": 0.3273,
      "step": 1175
    },
    {
      "epoch": 4.8,
      "grad_norm": 26.77984619140625,
      "learning_rate": 0.0003332233586275157,
      "loss": 0.3504,
      "step": 1200
    },
    {
      "epoch": 4.9,
      "grad_norm": 9.389993667602539,
      "learning_rate": 0.00033157373804025076,
      "loss": 0.3339,
      "step": 1225
    },
    {
      "epoch": 5.0,
      "grad_norm": 25.70441246032715,
      "learning_rate": 0.0003299241174529858,
      "loss": 0.3336,
      "step": 1250
    },
    {
      "epoch": 5.1,
      "grad_norm": 13.482085227966309,
      "learning_rate": 0.0003282744968657209,
      "loss": 0.2156,
      "step": 1275
    },
    {
      "epoch": 5.2,
      "grad_norm": 8.725810050964355,
      "learning_rate": 0.00032662487627845596,
      "loss": 0.1811,
      "step": 1300
    },
    {
      "epoch": 5.3,
      "grad_norm": 9.313215255737305,
      "learning_rate": 0.00032497525569119106,
      "loss": 0.2807,
      "step": 1325
    },
    {
      "epoch": 5.4,
      "grad_norm": 11.026411056518555,
      "learning_rate": 0.0003233256351039261,
      "loss": 0.2757,
      "step": 1350
    },
    {
      "epoch": 5.5,
      "grad_norm": 11.038985252380371,
      "learning_rate": 0.00032167601451666116,
      "loss": 0.2177,
      "step": 1375
    },
    {
      "epoch": 5.6,
      "grad_norm": 4.008651256561279,
      "learning_rate": 0.00032002639392939627,
      "loss": 0.2163,
      "step": 1400
    },
    {
      "epoch": 5.7,
      "grad_norm": 12.480770111083984,
      "learning_rate": 0.0003183767733421313,
      "loss": 0.2173,
      "step": 1425
    },
    {
      "epoch": 5.8,
      "grad_norm": 8.751969337463379,
      "learning_rate": 0.0003167271527548664,
      "loss": 0.2299,
      "step": 1450
    },
    {
      "epoch": 5.9,
      "grad_norm": 7.701971530914307,
      "learning_rate": 0.00031507753216760147,
      "loss": 0.1949,
      "step": 1475
    },
    {
      "epoch": 6.0,
      "grad_norm": 8.48027515411377,
      "learning_rate": 0.00031342791158033657,
      "loss": 0.2599,
      "step": 1500
    },
    {
      "epoch": 6.1,
      "grad_norm": 9.29404067993164,
      "learning_rate": 0.00031177829099307157,
      "loss": 0.1376,
      "step": 1525
    },
    {
      "epoch": 6.2,
      "grad_norm": 20.137714385986328,
      "learning_rate": 0.00031012867040580667,
      "loss": 0.1647,
      "step": 1550
    },
    {
      "epoch": 6.3,
      "grad_norm": 11.394575119018555,
      "learning_rate": 0.0003084790498185417,
      "loss": 0.1565,
      "step": 1575
    },
    {
      "epoch": 6.4,
      "grad_norm": 8.214287757873535,
      "learning_rate": 0.0003068294292312768,
      "loss": 0.1739,
      "step": 1600
    },
    {
      "epoch": 6.5,
      "grad_norm": 7.779988765716553,
      "learning_rate": 0.0003051798086440119,
      "loss": 0.1403,
      "step": 1625
    },
    {
      "epoch": 6.6,
      "grad_norm": 9.421648025512695,
      "learning_rate": 0.000303530188056747,
      "loss": 0.1545,
      "step": 1650
    },
    {
      "epoch": 6.7,
      "grad_norm": 5.751734256744385,
      "learning_rate": 0.000301880567469482,
      "loss": 0.1971,
      "step": 1675
    },
    {
      "epoch": 6.8,
      "grad_norm": 23.861705780029297,
      "learning_rate": 0.0003002309468822171,
      "loss": 0.1681,
      "step": 1700
    },
    {
      "epoch": 6.9,
      "grad_norm": 18.944721221923828,
      "learning_rate": 0.0002985813262949522,
      "loss": 0.1703,
      "step": 1725
    },
    {
      "epoch": 7.0,
      "grad_norm": 14.045795440673828,
      "learning_rate": 0.00029693170570768723,
      "loss": 0.1801,
      "step": 1750
    },
    {
      "epoch": 7.1,
      "grad_norm": 6.4620137214660645,
      "learning_rate": 0.00029528208512042233,
      "loss": 0.1253,
      "step": 1775
    },
    {
      "epoch": 7.2,
      "grad_norm": 4.318169593811035,
      "learning_rate": 0.0002936324645331574,
      "loss": 0.1397,
      "step": 1800
    },
    {
      "epoch": 7.3,
      "grad_norm": 24.91462516784668,
      "learning_rate": 0.00029198284394589243,
      "loss": 0.1259,
      "step": 1825
    },
    {
      "epoch": 7.4,
      "grad_norm": 23.614572525024414,
      "learning_rate": 0.00029033322335862753,
      "loss": 0.1293,
      "step": 1850
    },
    {
      "epoch": 7.5,
      "grad_norm": 3.60048508644104,
      "learning_rate": 0.0002886836027713626,
      "loss": 0.138,
      "step": 1875
    },
    {
      "epoch": 7.6,
      "grad_norm": 16.62705421447754,
      "learning_rate": 0.0002870339821840977,
      "loss": 0.1733,
      "step": 1900
    },
    {
      "epoch": 7.7,
      "grad_norm": 33.79671859741211,
      "learning_rate": 0.00028538436159683273,
      "loss": 0.217,
      "step": 1925
    },
    {
      "epoch": 7.8,
      "grad_norm": 9.69206428527832,
      "learning_rate": 0.00028373474100956784,
      "loss": 0.1289,
      "step": 1950
    },
    {
      "epoch": 7.9,
      "grad_norm": 18.655046463012695,
      "learning_rate": 0.0002820851204223029,
      "loss": 0.1168,
      "step": 1975
    },
    {
      "epoch": 8.0,
      "grad_norm": 8.4110746383667,
      "learning_rate": 0.00028043549983503794,
      "loss": 0.1471,
      "step": 2000
    },
    {
      "epoch": 8.1,
      "grad_norm": 14.473428726196289,
      "learning_rate": 0.00027878587924777304,
      "loss": 0.0774,
      "step": 2025
    },
    {
      "epoch": 8.2,
      "grad_norm": 1.2464979887008667,
      "learning_rate": 0.0002771362586605081,
      "loss": 0.0715,
      "step": 2050
    },
    {
      "epoch": 8.3,
      "grad_norm": 26.74981689453125,
      "learning_rate": 0.0002754866380732432,
      "loss": 0.1289,
      "step": 2075
    },
    {
      "epoch": 8.4,
      "grad_norm": 13.958703994750977,
      "learning_rate": 0.00027383701748597824,
      "loss": 0.1294,
      "step": 2100
    },
    {
      "epoch": 8.5,
      "grad_norm": 10.394835472106934,
      "learning_rate": 0.0002721873968987133,
      "loss": 0.0987,
      "step": 2125
    },
    {
      "epoch": 8.6,
      "grad_norm": 9.387774467468262,
      "learning_rate": 0.00027053777631144834,
      "loss": 0.1269,
      "step": 2150
    },
    {
      "epoch": 8.7,
      "grad_norm": 4.988718509674072,
      "learning_rate": 0.00026888815572418344,
      "loss": 0.1374,
      "step": 2175
    },
    {
      "epoch": 8.8,
      "grad_norm": 12.66832447052002,
      "learning_rate": 0.00026723853513691855,
      "loss": 0.0997,
      "step": 2200
    },
    {
      "epoch": 8.9,
      "grad_norm": 19.10486602783203,
      "learning_rate": 0.0002655889145496536,
      "loss": 0.1325,
      "step": 2225
    },
    {
      "epoch": 9.0,
      "grad_norm": 3.2943971157073975,
      "learning_rate": 0.0002639392939623887,
      "loss": 0.1385,
      "step": 2250
    },
    {
      "epoch": 9.1,
      "grad_norm": 11.490874290466309,
      "learning_rate": 0.0002622896733751237,
      "loss": 0.0573,
      "step": 2275
    },
    {
      "epoch": 9.2,
      "grad_norm": 11.961288452148438,
      "learning_rate": 0.0002606400527878588,
      "loss": 0.1128,
      "step": 2300
    },
    {
      "epoch": 9.3,
      "grad_norm": 0.7256277203559875,
      "learning_rate": 0.00025899043220059385,
      "loss": 0.0673,
      "step": 2325
    },
    {
      "epoch": 9.4,
      "grad_norm": 9.857927322387695,
      "learning_rate": 0.00025734081161332895,
      "loss": 0.2072,
      "step": 2350
    },
    {
      "epoch": 9.5,
      "grad_norm": 34.927734375,
      "learning_rate": 0.000255691191026064,
      "loss": 0.221,
      "step": 2375
    },
    {
      "epoch": 9.6,
      "grad_norm": 28.461849212646484,
      "learning_rate": 0.0002540415704387991,
      "loss": 0.1263,
      "step": 2400
    },
    {
      "epoch": 9.7,
      "grad_norm": 9.591217041015625,
      "learning_rate": 0.00025239194985153415,
      "loss": 0.099,
      "step": 2425
    },
    {
      "epoch": 9.8,
      "grad_norm": 9.780253410339355,
      "learning_rate": 0.0002507423292642692,
      "loss": 0.0754,
      "step": 2450
    },
    {
      "epoch": 9.9,
      "grad_norm": 14.394485473632812,
      "learning_rate": 0.0002490927086770043,
      "loss": 0.0909,
      "step": 2475
    },
    {
      "epoch": 10.0,
      "grad_norm": 4.217277526855469,
      "learning_rate": 0.00024744308808973936,
      "loss": 0.1495,
      "step": 2500
    },
    {
      "epoch": 10.1,
      "grad_norm": 0.08438724279403687,
      "learning_rate": 0.00024579346750247446,
      "loss": 0.0662,
      "step": 2525
    },
    {
      "epoch": 10.2,
      "grad_norm": 0.9595862030982971,
      "learning_rate": 0.0002441438469152095,
      "loss": 0.084,
      "step": 2550
    },
    {
      "epoch": 10.3,
      "grad_norm": 42.32374954223633,
      "learning_rate": 0.00024249422632794456,
      "loss": 0.0541,
      "step": 2575
    },
    {
      "epoch": 10.4,
      "grad_norm": 24.792821884155273,
      "learning_rate": 0.00024084460574067963,
      "loss": 0.121,
      "step": 2600
    },
    {
      "epoch": 10.5,
      "grad_norm": 2.7049331665039062,
      "learning_rate": 0.0002391949851534147,
      "loss": 0.1551,
      "step": 2625
    },
    {
      "epoch": 10.6,
      "grad_norm": 2.788362741470337,
      "learning_rate": 0.0002375453645661498,
      "loss": 0.0877,
      "step": 2650
    },
    {
      "epoch": 10.7,
      "grad_norm": 0.4999215006828308,
      "learning_rate": 0.00023589574397888486,
      "loss": 0.1263,
      "step": 2675
    },
    {
      "epoch": 10.8,
      "grad_norm": 16.476675033569336,
      "learning_rate": 0.00023424612339161997,
      "loss": 0.0883,
      "step": 2700
    },
    {
      "epoch": 10.9,
      "grad_norm": 38.192726135253906,
      "learning_rate": 0.000232596502804355,
      "loss": 0.0843,
      "step": 2725
    },
    {
      "epoch": 11.0,
      "grad_norm": 18.09952735900879,
      "learning_rate": 0.00023094688221709007,
      "loss": 0.0957,
      "step": 2750
    },
    {
      "epoch": 11.1,
      "grad_norm": 34.254417419433594,
      "learning_rate": 0.00022929726162982514,
      "loss": 0.0856,
      "step": 2775
    },
    {
      "epoch": 11.2,
      "grad_norm": 11.744216918945312,
      "learning_rate": 0.00022764764104256022,
      "loss": 0.0654,
      "step": 2800
    },
    {
      "epoch": 11.3,
      "grad_norm": 0.1285558044910431,
      "learning_rate": 0.0002259980204552953,
      "loss": 0.0638,
      "step": 2825
    },
    {
      "epoch": 11.4,
      "grad_norm": 20.57583999633789,
      "learning_rate": 0.00022434839986803037,
      "loss": 0.079,
      "step": 2850
    },
    {
      "epoch": 11.5,
      "grad_norm": 0.7192332148551941,
      "learning_rate": 0.00022269877928076542,
      "loss": 0.0854,
      "step": 2875
    },
    {
      "epoch": 11.6,
      "grad_norm": 31.341829299926758,
      "learning_rate": 0.0002210491586935005,
      "loss": 0.063,
      "step": 2900
    },
    {
      "epoch": 11.7,
      "grad_norm": 13.460247039794922,
      "learning_rate": 0.00021939953810623557,
      "loss": 0.1358,
      "step": 2925
    },
    {
      "epoch": 11.8,
      "grad_norm": 1.671036720275879,
      "learning_rate": 0.00021774991751897065,
      "loss": 0.0806,
      "step": 2950
    },
    {
      "epoch": 11.9,
      "grad_norm": 22.026491165161133,
      "learning_rate": 0.00021610029693170573,
      "loss": 0.0912,
      "step": 2975
    },
    {
      "epoch": 12.0,
      "grad_norm": 46.78192138671875,
      "learning_rate": 0.0002144506763444408,
      "loss": 0.0763,
      "step": 3000
    },
    {
      "epoch": 12.1,
      "grad_norm": 6.397649765014648,
      "learning_rate": 0.00021280105575717582,
      "loss": 0.0613,
      "step": 3025
    },
    {
      "epoch": 12.2,
      "grad_norm": 7.179838180541992,
      "learning_rate": 0.00021115143516991093,
      "loss": 0.0496,
      "step": 3050
    },
    {
      "epoch": 12.3,
      "grad_norm": 0.021825680509209633,
      "learning_rate": 0.000209501814582646,
      "loss": 0.0633,
      "step": 3075
    },
    {
      "epoch": 12.4,
      "grad_norm": 6.6774187088012695,
      "learning_rate": 0.00020785219399538108,
      "loss": 0.0765,
      "step": 3100
    },
    {
      "epoch": 12.5,
      "grad_norm": 0.13217756152153015,
      "learning_rate": 0.00020620257340811616,
      "loss": 0.0696,
      "step": 3125
    },
    {
      "epoch": 12.6,
      "grad_norm": 2.5293831825256348,
      "learning_rate": 0.00020455295282085123,
      "loss": 0.0419,
      "step": 3150
    },
    {
      "epoch": 12.7,
      "grad_norm": 1.1612874269485474,
      "learning_rate": 0.00020290333223358626,
      "loss": 0.1046,
      "step": 3175
    },
    {
      "epoch": 12.8,
      "grad_norm": 40.93427658081055,
      "learning_rate": 0.00020125371164632133,
      "loss": 0.0488,
      "step": 3200
    },
    {
      "epoch": 12.9,
      "grad_norm": 1.446478009223938,
      "learning_rate": 0.0001996040910590564,
      "loss": 0.062,
      "step": 3225
    },
    {
      "epoch": 13.0,
      "grad_norm": 28.893821716308594,
      "learning_rate": 0.00019795447047179148,
      "loss": 0.0913,
      "step": 3250
    },
    {
      "epoch": 13.1,
      "grad_norm": 23.31914520263672,
      "learning_rate": 0.0001963048498845266,
      "loss": 0.061,
      "step": 3275
    },
    {
      "epoch": 13.2,
      "grad_norm": 0.26242795586586,
      "learning_rate": 0.00019465522929726164,
      "loss": 0.0563,
      "step": 3300
    },
    {
      "epoch": 13.3,
      "grad_norm": 0.08909507840871811,
      "learning_rate": 0.00019300560870999671,
      "loss": 0.0387,
      "step": 3325
    },
    {
      "epoch": 13.4,
      "grad_norm": 24.825326919555664,
      "learning_rate": 0.0001913559881227318,
      "loss": 0.0776,
      "step": 3350
    },
    {
      "epoch": 13.5,
      "grad_norm": 39.43446731567383,
      "learning_rate": 0.00018970636753546684,
      "loss": 0.0846,
      "step": 3375
    },
    {
      "epoch": 13.6,
      "grad_norm": 32.98988723754883,
      "learning_rate": 0.00018805674694820192,
      "loss": 0.0512,
      "step": 3400
    },
    {
      "epoch": 13.7,
      "grad_norm": 4.053821563720703,
      "learning_rate": 0.000186407126360937,
      "loss": 0.1576,
      "step": 3425
    },
    {
      "epoch": 13.8,
      "grad_norm": 0.2785554528236389,
      "learning_rate": 0.00018475750577367207,
      "loss": 0.0728,
      "step": 3450
    },
    {
      "epoch": 13.9,
      "grad_norm": 11.39714527130127,
      "learning_rate": 0.00018310788518640715,
      "loss": 0.0776,
      "step": 3475
    },
    {
      "epoch": 14.0,
      "grad_norm": 3.7018072605133057,
      "learning_rate": 0.00018145826459914222,
      "loss": 0.0736,
      "step": 3500
    },
    {
      "epoch": 14.1,
      "grad_norm": 0.26053619384765625,
      "learning_rate": 0.00017980864401187727,
      "loss": 0.0598,
      "step": 3525
    },
    {
      "epoch": 14.2,
      "grad_norm": 6.454179763793945,
      "learning_rate": 0.00017815902342461235,
      "loss": 0.0435,
      "step": 3550
    },
    {
      "epoch": 14.3,
      "grad_norm": 34.47985076904297,
      "learning_rate": 0.00017650940283734742,
      "loss": 0.0467,
      "step": 3575
    },
    {
      "epoch": 14.4,
      "grad_norm": 11.76491928100586,
      "learning_rate": 0.00017485978225008247,
      "loss": 0.1298,
      "step": 3600
    },
    {
      "epoch": 14.5,
      "grad_norm": 26.64131736755371,
      "learning_rate": 0.00017321016166281755,
      "loss": 0.1313,
      "step": 3625
    },
    {
      "epoch": 14.6,
      "grad_norm": 5.510906219482422,
      "learning_rate": 0.00017156054107555263,
      "loss": 0.0421,
      "step": 3650
    },
    {
      "epoch": 14.7,
      "grad_norm": 0.04721131548285484,
      "learning_rate": 0.0001699109204882877,
      "loss": 0.0569,
      "step": 3675
    },
    {
      "epoch": 14.8,
      "grad_norm": 0.6616227030754089,
      "learning_rate": 0.00016826129990102278,
      "loss": 0.0618,
      "step": 3700
    },
    {
      "epoch": 14.9,
      "grad_norm": 0.05287986248731613,
      "learning_rate": 0.00016661167931375785,
      "loss": 0.0574,
      "step": 3725
    },
    {
      "epoch": 15.0,
      "grad_norm": 0.49707508087158203,
      "learning_rate": 0.0001649620587264929,
      "loss": 0.094,
      "step": 3750
    },
    {
      "epoch": 15.1,
      "grad_norm": 23.649728775024414,
      "learning_rate": 0.00016331243813922798,
      "loss": 0.0512,
      "step": 3775
    },
    {
      "epoch": 15.2,
      "grad_norm": 0.01772051490843296,
      "learning_rate": 0.00016166281755196306,
      "loss": 0.0719,
      "step": 3800
    },
    {
      "epoch": 15.3,
      "grad_norm": 1.5259039402008057,
      "learning_rate": 0.00016001319696469813,
      "loss": 0.0615,
      "step": 3825
    },
    {
      "epoch": 15.4,
      "grad_norm": 0.026406478136777878,
      "learning_rate": 0.0001583635763774332,
      "loss": 0.0651,
      "step": 3850
    },
    {
      "epoch": 15.5,
      "grad_norm": 27.12177085876465,
      "learning_rate": 0.00015671395579016829,
      "loss": 0.0458,
      "step": 3875
    },
    {
      "epoch": 15.6,
      "grad_norm": 9.838811874389648,
      "learning_rate": 0.00015506433520290334,
      "loss": 0.0662,
      "step": 3900
    },
    {
      "epoch": 15.7,
      "grad_norm": 0.08516795933246613,
      "learning_rate": 0.0001534147146156384,
      "loss": 0.0302,
      "step": 3925
    },
    {
      "epoch": 15.8,
      "grad_norm": 1.1224850416183472,
      "learning_rate": 0.0001517650940283735,
      "loss": 0.0637,
      "step": 3950
    },
    {
      "epoch": 15.9,
      "grad_norm": 0.14810702204704285,
      "learning_rate": 0.00015011547344110854,
      "loss": 0.0296,
      "step": 3975
    },
    {
      "epoch": 16.0,
      "grad_norm": 0.046288371086120605,
      "learning_rate": 0.00014846585285384361,
      "loss": 0.0368,
      "step": 4000
    },
    {
      "epoch": 16.1,
      "grad_norm": 0.008764918893575668,
      "learning_rate": 0.0001468162322665787,
      "loss": 0.0272,
      "step": 4025
    },
    {
      "epoch": 16.2,
      "grad_norm": 0.39770299196243286,
      "learning_rate": 0.00014516661167931377,
      "loss": 0.0344,
      "step": 4050
    },
    {
      "epoch": 16.3,
      "grad_norm": 23.780405044555664,
      "learning_rate": 0.00014351699109204884,
      "loss": 0.0617,
      "step": 4075
    },
    {
      "epoch": 16.4,
      "grad_norm": 0.023103665560483932,
      "learning_rate": 0.00014186737050478392,
      "loss": 0.0182,
      "step": 4100
    },
    {
      "epoch": 16.5,
      "grad_norm": 0.00998806394636631,
      "learning_rate": 0.00014021774991751897,
      "loss": 0.042,
      "step": 4125
    },
    {
      "epoch": 16.6,
      "grad_norm": 43.775169372558594,
      "learning_rate": 0.00013856812933025404,
      "loss": 0.0182,
      "step": 4150
    },
    {
      "epoch": 16.7,
      "grad_norm": 0.008157053031027317,
      "learning_rate": 0.00013691850874298912,
      "loss": 0.0377,
      "step": 4175
    },
    {
      "epoch": 16.8,
      "grad_norm": 0.011455570347607136,
      "learning_rate": 0.00013526888815572417,
      "loss": 0.0255,
      "step": 4200
    },
    {
      "epoch": 16.9,
      "grad_norm": 34.8545036315918,
      "learning_rate": 0.00013361926756845927,
      "loss": 0.0486,
      "step": 4225
    },
    {
      "epoch": 17.0,
      "grad_norm": 0.006066238507628441,
      "learning_rate": 0.00013196964698119435,
      "loss": 0.0821,
      "step": 4250
    },
    {
      "epoch": 17.1,
      "grad_norm": 43.46882247924805,
      "learning_rate": 0.0001303200263939294,
      "loss": 0.0333,
      "step": 4275
    },
    {
      "epoch": 17.2,
      "grad_norm": 2.3177573680877686,
      "learning_rate": 0.00012867040580666448,
      "loss": 0.0477,
      "step": 4300
    },
    {
      "epoch": 17.3,
      "grad_norm": 0.2264157235622406,
      "learning_rate": 0.00012702078521939955,
      "loss": 0.0373,
      "step": 4325
    },
    {
      "epoch": 17.4,
      "grad_norm": 68.17675018310547,
      "learning_rate": 0.0001253711646321346,
      "loss": 0.036,
      "step": 4350
    },
    {
      "epoch": 17.5,
      "grad_norm": 1.5359147787094116,
      "learning_rate": 0.00012372154404486968,
      "loss": 0.0437,
      "step": 4375
    },
    {
      "epoch": 17.6,
      "grad_norm": 0.008654219098389149,
      "learning_rate": 0.00012207192345760475,
      "loss": 0.0295,
      "step": 4400
    },
    {
      "epoch": 17.7,
      "grad_norm": 0.0074989828281104565,
      "learning_rate": 0.00012042230287033982,
      "loss": 0.0229,
      "step": 4425
    },
    {
      "epoch": 17.8,
      "grad_norm": 0.03568067401647568,
      "learning_rate": 0.0001187726822830749,
      "loss": 0.0636,
      "step": 4450
    },
    {
      "epoch": 17.9,
      "grad_norm": 0.0151940006762743,
      "learning_rate": 0.00011712306169580998,
      "loss": 0.0669,
      "step": 4475
    },
    {
      "epoch": 18.0,
      "grad_norm": 0.04830991476774216,
      "learning_rate": 0.00011547344110854503,
      "loss": 0.0388,
      "step": 4500
    },
    {
      "epoch": 18.1,
      "grad_norm": 1.1794861555099487,
      "learning_rate": 0.00011382382052128011,
      "loss": 0.0069,
      "step": 4525
    },
    {
      "epoch": 18.2,
      "grad_norm": 13.07977294921875,
      "learning_rate": 0.00011217419993401519,
      "loss": 0.0315,
      "step": 4550
    },
    {
      "epoch": 18.3,
      "grad_norm": 0.011115381494164467,
      "learning_rate": 0.00011052457934675025,
      "loss": 0.0144,
      "step": 4575
    },
    {
      "epoch": 18.4,
      "grad_norm": 25.42391014099121,
      "learning_rate": 0.00010887495875948532,
      "loss": 0.0198,
      "step": 4600
    },
    {
      "epoch": 18.5,
      "grad_norm": 44.36895751953125,
      "learning_rate": 0.0001072253381722204,
      "loss": 0.0297,
      "step": 4625
    },
    {
      "epoch": 18.6,
      "grad_norm": 6.828146457672119,
      "learning_rate": 0.00010557571758495546,
      "loss": 0.0348,
      "step": 4650
    },
    {
      "epoch": 18.7,
      "grad_norm": 38.9465446472168,
      "learning_rate": 0.00010392609699769054,
      "loss": 0.0336,
      "step": 4675
    },
    {
      "epoch": 18.8,
      "grad_norm": 0.015516690909862518,
      "learning_rate": 0.00010227647641042562,
      "loss": 0.032,
      "step": 4700
    },
    {
      "epoch": 18.9,
      "grad_norm": 5.419480323791504,
      "learning_rate": 0.00010062685582316067,
      "loss": 0.0169,
      "step": 4725
    },
    {
      "epoch": 19.0,
      "grad_norm": 0.7849720120429993,
      "learning_rate": 9.897723523589574e-05,
      "loss": 0.0321,
      "step": 4750
    },
    {
      "epoch": 19.1,
      "grad_norm": 25.438196182250977,
      "learning_rate": 9.732761464863082e-05,
      "loss": 0.046,
      "step": 4775
    },
    {
      "epoch": 19.2,
      "grad_norm": 0.010676453821361065,
      "learning_rate": 9.56779940613659e-05,
      "loss": 0.0111,
      "step": 4800
    },
    {
      "epoch": 19.3,
      "grad_norm": 0.011585243977606297,
      "learning_rate": 9.402837347410096e-05,
      "loss": 0.0154,
      "step": 4825
    },
    {
      "epoch": 19.4,
      "grad_norm": 0.012879762798547745,
      "learning_rate": 9.237875288683603e-05,
      "loss": 0.02,
      "step": 4850
    },
    {
      "epoch": 19.5,
      "grad_norm": 0.4159695506095886,
      "learning_rate": 9.072913229957111e-05,
      "loss": 0.0277,
      "step": 4875
    },
    {
      "epoch": 19.6,
      "grad_norm": 0.013872765935957432,
      "learning_rate": 8.907951171230617e-05,
      "loss": 0.0069,
      "step": 4900
    },
    {
      "epoch": 19.7,
      "grad_norm": 3.4260928630828857,
      "learning_rate": 8.742989112504124e-05,
      "loss": 0.0319,
      "step": 4925
    },
    {
      "epoch": 19.8,
      "grad_norm": 0.043105900287628174,
      "learning_rate": 8.578027053777631e-05,
      "loss": 0.0445,
      "step": 4950
    },
    {
      "epoch": 19.9,
      "grad_norm": 10.97398853302002,
      "learning_rate": 8.413064995051139e-05,
      "loss": 0.0185,
      "step": 4975
    },
    {
      "epoch": 20.0,
      "grad_norm": 12.879217147827148,
      "learning_rate": 8.248102936324645e-05,
      "loss": 0.0197,
      "step": 5000
    },
    {
      "epoch": 20.1,
      "grad_norm": 64.58011627197266,
      "learning_rate": 8.083140877598153e-05,
      "loss": 0.0228,
      "step": 5025
    },
    {
      "epoch": 20.2,
      "grad_norm": 43.28678512573242,
      "learning_rate": 7.91817881887166e-05,
      "loss": 0.0311,
      "step": 5050
    },
    {
      "epoch": 20.3,
      "grad_norm": 0.005558234639465809,
      "learning_rate": 7.753216760145167e-05,
      "loss": 0.0283,
      "step": 5075
    },
    {
      "epoch": 20.4,
      "grad_norm": 0.19123849272727966,
      "learning_rate": 7.588254701418674e-05,
      "loss": 0.0445,
      "step": 5100
    },
    {
      "epoch": 20.5,
      "grad_norm": 5.734357833862305,
      "learning_rate": 7.423292642692181e-05,
      "loss": 0.0145,
      "step": 5125
    },
    {
      "epoch": 20.6,
      "grad_norm": 0.005894747097045183,
      "learning_rate": 7.258330583965688e-05,
      "loss": 0.0233,
      "step": 5150
    },
    {
      "epoch": 20.7,
      "grad_norm": 0.018981292843818665,
      "learning_rate": 7.093368525239196e-05,
      "loss": 0.0273,
      "step": 5175
    },
    {
      "epoch": 20.8,
      "grad_norm": 0.09156472235918045,
      "learning_rate": 6.928406466512702e-05,
      "loss": 0.0233,
      "step": 5200
    },
    {
      "epoch": 20.9,
      "grad_norm": 0.8908875584602356,
      "learning_rate": 6.763444407786209e-05,
      "loss": 0.0221,
      "step": 5225
    },
    {
      "epoch": 21.0,
      "grad_norm": 6.745031833648682,
      "learning_rate": 6.598482349059718e-05,
      "loss": 0.0278,
      "step": 5250
    },
    {
      "epoch": 21.1,
      "grad_norm": 0.009193326346576214,
      "learning_rate": 6.433520290333224e-05,
      "loss": 0.0226,
      "step": 5275
    },
    {
      "epoch": 21.2,
      "grad_norm": 2.408499002456665,
      "learning_rate": 6.26855823160673e-05,
      "loss": 0.0202,
      "step": 5300
    },
    {
      "epoch": 21.3,
      "grad_norm": 0.006645245011895895,
      "learning_rate": 6.103596172880238e-05,
      "loss": 0.0234,
      "step": 5325
    },
    {
      "epoch": 21.4,
      "grad_norm": 0.006036018021404743,
      "learning_rate": 5.938634114153745e-05,
      "loss": 0.0333,
      "step": 5350
    },
    {
      "epoch": 21.5,
      "grad_norm": 0.25791531801223755,
      "learning_rate": 5.7736720554272516e-05,
      "loss": 0.012,
      "step": 5375
    },
    {
      "epoch": 21.6,
      "grad_norm": 0.9368523359298706,
      "learning_rate": 5.608709996700759e-05,
      "loss": 0.0078,
      "step": 5400
    },
    {
      "epoch": 21.7,
      "grad_norm": 0.0046651544980704784,
      "learning_rate": 5.443747937974266e-05,
      "loss": 0.0096,
      "step": 5425
    },
    {
      "epoch": 21.8,
      "grad_norm": 0.004647469613701105,
      "learning_rate": 5.278785879247773e-05,
      "loss": 0.0299,
      "step": 5450
    },
    {
      "epoch": 21.9,
      "grad_norm": 0.004732856526970863,
      "learning_rate": 5.113823820521281e-05,
      "loss": 0.0185,
      "step": 5475
    },
    {
      "epoch": 22.0,
      "grad_norm": 0.007009522989392281,
      "learning_rate": 4.948861761794787e-05,
      "loss": 0.0228,
      "step": 5500
    },
    {
      "epoch": 22.1,
      "grad_norm": 25.634002685546875,
      "learning_rate": 4.783899703068295e-05,
      "loss": 0.011,
      "step": 5525
    },
    {
      "epoch": 22.2,
      "grad_norm": 0.009100685827434063,
      "learning_rate": 4.618937644341802e-05,
      "loss": 0.0056,
      "step": 5550
    },
    {
      "epoch": 22.3,
      "grad_norm": 0.004046417307108641,
      "learning_rate": 4.453975585615309e-05,
      "loss": 0.0084,
      "step": 5575
    },
    {
      "epoch": 22.4,
      "grad_norm": 0.025503572076559067,
      "learning_rate": 4.2890135268888156e-05,
      "loss": 0.0066,
      "step": 5600
    },
    {
      "epoch": 22.5,
      "grad_norm": 0.005815221928060055,
      "learning_rate": 4.1240514681623226e-05,
      "loss": 0.0407,
      "step": 5625
    },
    {
      "epoch": 22.6,
      "grad_norm": 0.06112132593989372,
      "learning_rate": 3.95908940943583e-05,
      "loss": 0.0112,
      "step": 5650
    },
    {
      "epoch": 22.7,
      "grad_norm": 0.01705116033554077,
      "learning_rate": 3.794127350709337e-05,
      "loss": 0.0008,
      "step": 5675
    },
    {
      "epoch": 22.8,
      "grad_norm": 0.005267620086669922,
      "learning_rate": 3.629165291982844e-05,
      "loss": 0.0131,
      "step": 5700
    },
    {
      "epoch": 22.9,
      "grad_norm": 0.0046350546181201935,
      "learning_rate": 3.464203233256351e-05,
      "loss": 0.0205,
      "step": 5725
    },
    {
      "epoch": 23.0,
      "grad_norm": 0.0063110594637691975,
      "learning_rate": 3.299241174529859e-05,
      "loss": 0.0072,
      "step": 5750
    },
    {
      "epoch": 23.1,
      "grad_norm": 0.004843282513320446,
      "learning_rate": 3.134279115803365e-05,
      "loss": 0.033,
      "step": 5775
    },
    {
      "epoch": 23.2,
      "grad_norm": 0.004899237770587206,
      "learning_rate": 2.9693170570768723e-05,
      "loss": 0.002,
      "step": 5800
    },
    {
      "epoch": 23.3,
      "grad_norm": 0.0040695276111364365,
      "learning_rate": 2.8043549983503796e-05,
      "loss": 0.0185,
      "step": 5825
    },
    {
      "epoch": 23.4,
      "grad_norm": 46.083438873291016,
      "learning_rate": 2.6393929396238866e-05,
      "loss": 0.0149,
      "step": 5850
    },
    {
      "epoch": 23.5,
      "grad_norm": 0.009059540927410126,
      "learning_rate": 2.4744308808973936e-05,
      "loss": 0.0264,
      "step": 5875
    },
    {
      "epoch": 23.6,
      "grad_norm": 11.57484245300293,
      "learning_rate": 2.309468822170901e-05,
      "loss": 0.0145,
      "step": 5900
    },
    {
      "epoch": 23.7,
      "grad_norm": 0.00815950334072113,
      "learning_rate": 2.1445067634444078e-05,
      "loss": 0.015,
      "step": 5925
    },
    {
      "epoch": 23.8,
      "grad_norm": 0.005930441431701183,
      "learning_rate": 1.979544704717915e-05,
      "loss": 0.0187,
      "step": 5950
    },
    {
      "epoch": 23.9,
      "grad_norm": 0.006135927978903055,
      "learning_rate": 1.814582645991422e-05,
      "loss": 0.0143,
      "step": 5975
    },
    {
      "epoch": 24.0,
      "grad_norm": 0.004939633421599865,
      "learning_rate": 1.6496205872649294e-05,
      "loss": 0.0088,
      "step": 6000
    }
  ],
  "logging_steps": 25,
  "max_steps": 6250,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 25,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 304916889600000.0,
  "train_batch_size": 20,
  "trial_name": null,
  "trial_params": null
}