{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.964487905301081,
  "eval_steps": 200,
  "global_step": 180,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.016469377251672673,
      "grad_norm": 25.988508224487305,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.5961,
      "step": 1
    },
    {
      "epoch": 0.032938754503345345,
      "grad_norm": 26.565134048461914,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.595,
      "step": 2
    },
    {
      "epoch": 0.049408131755018014,
      "grad_norm": 22.700319290161133,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.5211,
      "step": 3
    },
    {
      "epoch": 0.06587750900669069,
      "grad_norm": 17.019222259521484,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.4043,
      "step": 4
    },
    {
      "epoch": 0.08234688625836335,
      "grad_norm": 15.759174346923828,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.3571,
      "step": 5
    },
    {
      "epoch": 0.09881626351003603,
      "grad_norm": 12.310270309448242,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.351,
      "step": 6
    },
    {
      "epoch": 0.1152856407617087,
      "grad_norm": 6.430344104766846,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.3183,
      "step": 7
    },
    {
      "epoch": 0.13175501801338138,
      "grad_norm": 29.118850708007812,
      "learning_rate": 8.888888888888888e-06,
      "loss": 0.2986,
      "step": 8
    },
    {
      "epoch": 0.14822439526505404,
      "grad_norm": 5.387965202331543,
      "learning_rate": 1e-05,
      "loss": 0.3031,
      "step": 9
    },
    {
      "epoch": 0.1646937725167267,
      "grad_norm": 4.416501998901367,
      "learning_rate": 1.1111111111111113e-05,
      "loss": 0.2769,
      "step": 10
    },
    {
      "epoch": 0.1811631497683994,
      "grad_norm": 3.6543190479278564,
      "learning_rate": 1.2222222222222224e-05,
      "loss": 0.2596,
      "step": 11
    },
    {
      "epoch": 0.19763252702007206,
      "grad_norm": 3.7170989513397217,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 0.2532,
      "step": 12
    },
    {
      "epoch": 0.21410190427174472,
      "grad_norm": 2.963078498840332,
      "learning_rate": 1.4444444444444446e-05,
      "loss": 0.241,
      "step": 13
    },
    {
      "epoch": 0.2305712815234174,
      "grad_norm": 3.8106493949890137,
      "learning_rate": 1.555555555555556e-05,
      "loss": 0.2355,
      "step": 14
    },
    {
      "epoch": 0.24704065877509007,
      "grad_norm": 3.3100316524505615,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 0.227,
      "step": 15
    },
    {
      "epoch": 0.26351003602676276,
      "grad_norm": 3.8181209564208984,
      "learning_rate": 1.7777777777777777e-05,
      "loss": 0.222,
      "step": 16
    },
    {
      "epoch": 0.2799794132784354,
      "grad_norm": 3.1601641178131104,
      "learning_rate": 1.888888888888889e-05,
      "loss": 0.216,
      "step": 17
    },
    {
      "epoch": 0.2964487905301081,
      "grad_norm": 3.8540680408477783,
      "learning_rate": 2e-05,
      "loss": 0.2219,
      "step": 18
    },
    {
      "epoch": 0.3129181677817808,
      "grad_norm": 2.4376087188720703,
      "learning_rate": 1.9998119704485016e-05,
      "loss": 0.2124,
      "step": 19
    },
    {
      "epoch": 0.3293875450334534,
      "grad_norm": 2.0443239212036133,
      "learning_rate": 1.9992479525042305e-05,
      "loss": 0.2072,
      "step": 20
    },
    {
      "epoch": 0.3458569222851261,
      "grad_norm": 2.7866272926330566,
      "learning_rate": 1.9983081582712684e-05,
      "loss": 0.2023,
      "step": 21
    },
    {
      "epoch": 0.3623262995367988,
      "grad_norm": 2.3178961277008057,
      "learning_rate": 1.996992941167792e-05,
      "loss": 0.2013,
      "step": 22
    },
    {
      "epoch": 0.3787956767884714,
      "grad_norm": 2.5984015464782715,
      "learning_rate": 1.9953027957931658e-05,
      "loss": 0.1933,
      "step": 23
    },
    {
      "epoch": 0.3952650540401441,
      "grad_norm": 1.8709274530410767,
      "learning_rate": 1.9932383577419432e-05,
      "loss": 0.1945,
      "step": 24
    },
    {
      "epoch": 0.4117344312918168,
      "grad_norm": 2.1134493350982666,
      "learning_rate": 1.9908004033648452e-05,
      "loss": 0.1916,
      "step": 25
    },
    {
      "epoch": 0.42820380854348944,
      "grad_norm": 2.076616048812866,
      "learning_rate": 1.9879898494768093e-05,
      "loss": 0.1902,
      "step": 26
    },
    {
      "epoch": 0.44467318579516213,
      "grad_norm": 2.05837082862854,
      "learning_rate": 1.9848077530122083e-05,
      "loss": 0.1849,
      "step": 27
    },
    {
      "epoch": 0.4611425630468348,
      "grad_norm": 1.7282419204711914,
      "learning_rate": 1.9812553106273848e-05,
      "loss": 0.1843,
      "step": 28
    },
    {
      "epoch": 0.47761194029850745,
      "grad_norm": 2.2622146606445312,
      "learning_rate": 1.9773338582506357e-05,
      "loss": 0.1872,
      "step": 29
    },
    {
      "epoch": 0.49408131755018014,
      "grad_norm": 1.5245953798294067,
      "learning_rate": 1.973044870579824e-05,
      "loss": 0.182,
      "step": 30
    },
    {
      "epoch": 0.5105506948018528,
      "grad_norm": 1.590587854385376,
      "learning_rate": 1.9683899605278062e-05,
      "loss": 0.1767,
      "step": 31
    },
    {
      "epoch": 0.5270200720535255,
      "grad_norm": 1.6935125589370728,
      "learning_rate": 1.9633708786158803e-05,
      "loss": 0.1739,
      "step": 32
    },
    {
      "epoch": 0.5434894493051982,
      "grad_norm": 1.6206973791122437,
      "learning_rate": 1.957989512315489e-05,
      "loss": 0.1771,
      "step": 33
    },
    {
      "epoch": 0.5599588265568708,
      "grad_norm": 1.561724066734314,
      "learning_rate": 1.9522478853384154e-05,
      "loss": 0.1706,
      "step": 34
    },
    {
      "epoch": 0.5764282038085435,
      "grad_norm": 1.4770928621292114,
      "learning_rate": 1.946148156875751e-05,
      "loss": 0.1711,
      "step": 35
    },
    {
      "epoch": 0.5928975810602162,
      "grad_norm": 1.5764074325561523,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.1756,
      "step": 36
    },
    {
      "epoch": 0.6093669583118888,
      "grad_norm": 2.4715301990509033,
      "learning_rate": 1.932883704732001e-05,
      "loss": 0.1757,
      "step": 37
    },
    {
      "epoch": 0.6258363355635616,
      "grad_norm": 1.3919658660888672,
      "learning_rate": 1.9257239692688907e-05,
      "loss": 0.1658,
      "step": 38
    },
    {
      "epoch": 0.6423057128152342,
      "grad_norm": 2.575564384460449,
      "learning_rate": 1.9182161068802742e-05,
      "loss": 0.1768,
      "step": 39
    },
    {
      "epoch": 0.6587750900669068,
      "grad_norm": 1.6217772960662842,
      "learning_rate": 1.9103629409661468e-05,
      "loss": 0.1695,
      "step": 40
    },
    {
      "epoch": 0.6752444673185796,
      "grad_norm": 2.0268895626068115,
      "learning_rate": 1.902167424781038e-05,
      "loss": 0.1749,
      "step": 41
    },
    {
      "epoch": 0.6917138445702522,
      "grad_norm": 1.6548528671264648,
      "learning_rate": 1.8936326403234125e-05,
      "loss": 0.1723,
      "step": 42
    },
    {
      "epoch": 0.7081832218219248,
      "grad_norm": 1.6716668605804443,
      "learning_rate": 1.8847617971766577e-05,
      "loss": 0.1724,
      "step": 43
    },
    {
      "epoch": 0.7246525990735976,
      "grad_norm": 1.6575418710708618,
      "learning_rate": 1.8755582313020912e-05,
      "loss": 0.1701,
      "step": 44
    },
    {
      "epoch": 0.7411219763252702,
      "grad_norm": 1.429405689239502,
      "learning_rate": 1.866025403784439e-05,
      "loss": 0.1665,
      "step": 45
    },
    {
      "epoch": 0.7575913535769428,
      "grad_norm": 1.5980753898620605,
      "learning_rate": 1.8561668995302668e-05,
      "loss": 0.1677,
      "step": 46
    },
    {
      "epoch": 0.7740607308286156,
      "grad_norm": 1.3052290678024292,
      "learning_rate": 1.845986425919841e-05,
      "loss": 0.1674,
      "step": 47
    },
    {
      "epoch": 0.7905301080802882,
      "grad_norm": 1.3834303617477417,
      "learning_rate": 1.8354878114129368e-05,
      "loss": 0.1621,
      "step": 48
    },
    {
      "epoch": 0.8069994853319609,
      "grad_norm": 1.314771294593811,
      "learning_rate": 1.824675004109107e-05,
      "loss": 0.1603,
      "step": 49
    },
    {
      "epoch": 0.8234688625836336,
      "grad_norm": 1.255486249923706,
      "learning_rate": 1.8135520702629677e-05,
      "loss": 0.1602,
      "step": 50
    },
    {
      "epoch": 0.8399382398353062,
      "grad_norm": 1.1701654195785522,
      "learning_rate": 1.802123192755044e-05,
      "loss": 0.1595,
      "step": 51
    },
    {
      "epoch": 0.8564076170869789,
      "grad_norm": 1.1197280883789062,
      "learning_rate": 1.7903926695187595e-05,
      "loss": 0.1589,
      "step": 52
    },
    {
      "epoch": 0.8728769943386516,
      "grad_norm": 1.26057767868042,
      "learning_rate": 1.7783649119241603e-05,
      "loss": 0.1627,
      "step": 53
    },
    {
      "epoch": 0.8893463715903243,
      "grad_norm": 1.1180353164672852,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.1577,
      "step": 54
    },
    {
      "epoch": 0.9058157488419969,
      "grad_norm": 1.1337931156158447,
      "learning_rate": 1.7534358963276606e-05,
      "loss": 0.1563,
      "step": 55
    },
    {
      "epoch": 0.9222851260936696,
      "grad_norm": 1.1284104585647583,
      "learning_rate": 1.740544013109005e-05,
      "loss": 0.1552,
      "step": 56
    },
    {
      "epoch": 0.9387545033453423,
      "grad_norm": 1.1995372772216797,
      "learning_rate": 1.7273736415730488e-05,
      "loss": 0.1571,
      "step": 57
    },
    {
      "epoch": 0.9552238805970149,
      "grad_norm": 1.189568042755127,
      "learning_rate": 1.7139297345578992e-05,
      "loss": 0.1543,
      "step": 58
    },
    {
      "epoch": 0.9716932578486875,
      "grad_norm": 1.1965142488479614,
      "learning_rate": 1.7002173477671685e-05,
      "loss": 0.1601,
      "step": 59
    },
    {
      "epoch": 0.9881626351003603,
      "grad_norm": 1.1370137929916382,
      "learning_rate": 1.686241637868734e-05,
      "loss": 0.1554,
      "step": 60
    },
    {
      "epoch": 1.004632012352033,
      "grad_norm": 1.208176612854004,
      "learning_rate": 1.6720078605555227e-05,
      "loss": 0.1458,
      "step": 61
    },
    {
      "epoch": 1.0211013896037056,
      "grad_norm": 1.1520111560821533,
      "learning_rate": 1.657521368569064e-05,
      "loss": 0.1315,
      "step": 62
    },
    {
      "epoch": 1.0375707668553782,
      "grad_norm": 1.1699644327163696,
      "learning_rate": 1.6427876096865394e-05,
      "loss": 0.1289,
      "step": 63
    },
    {
      "epoch": 1.054040144107051,
      "grad_norm": 1.2921233177185059,
      "learning_rate": 1.627812124672099e-05,
      "loss": 0.1292,
      "step": 64
    },
    {
      "epoch": 1.0705095213587237,
      "grad_norm": 1.1183018684387207,
      "learning_rate": 1.6126005451932028e-05,
      "loss": 0.1301,
      "step": 65
    },
    {
      "epoch": 1.0869788986103963,
      "grad_norm": 1.132805347442627,
      "learning_rate": 1.5971585917027864e-05,
      "loss": 0.1291,
      "step": 66
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 1.1662507057189941,
      "learning_rate": 1.5814920712880267e-05,
      "loss": 0.1296,
      "step": 67
    },
    {
      "epoch": 1.1199176531137416,
      "grad_norm": 1.230918288230896,
      "learning_rate": 1.5656068754865388e-05,
      "loss": 0.1283,
      "step": 68
    },
    {
      "epoch": 1.1363870303654142,
      "grad_norm": 1.111428141593933,
      "learning_rate": 1.5495089780708062e-05,
      "loss": 0.1264,
      "step": 69
    },
    {
      "epoch": 1.152856407617087,
      "grad_norm": 1.118282437324524,
      "learning_rate": 1.5332044328016916e-05,
      "loss": 0.1326,
      "step": 70
    },
    {
      "epoch": 1.1693257848687597,
      "grad_norm": 1.0411953926086426,
      "learning_rate": 1.5166993711518631e-05,
      "loss": 0.1273,
      "step": 71
    },
    {
      "epoch": 1.1857951621204323,
      "grad_norm": 1.0174416303634644,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.1236,
      "step": 72
    },
    {
      "epoch": 1.202264539372105,
      "grad_norm": 1.1264220476150513,
      "learning_rate": 1.4831125992966386e-05,
      "loss": 0.1262,
      "step": 73
    },
    {
      "epoch": 1.2187339166237776,
      "grad_norm": 1.1279516220092773,
      "learning_rate": 1.4660435197025391e-05,
      "loss": 0.1259,
      "step": 74
    },
    {
      "epoch": 1.2352032938754502,
      "grad_norm": 0.9918281435966492,
      "learning_rate": 1.4487991802004625e-05,
      "loss": 0.1263,
      "step": 75
    },
    {
      "epoch": 1.2516726711271229,
      "grad_norm": 1.0115736722946167,
      "learning_rate": 1.4313860656812537e-05,
      "loss": 0.1281,
      "step": 76
    },
    {
      "epoch": 1.2681420483787957,
      "grad_norm": 0.993411123752594,
      "learning_rate": 1.4138107245051394e-05,
      "loss": 0.1287,
      "step": 77
    },
    {
      "epoch": 1.2846114256304684,
      "grad_norm": 1.0473469495773315,
      "learning_rate": 1.396079766039157e-05,
      "loss": 0.1272,
      "step": 78
    },
    {
      "epoch": 1.301080802882141,
      "grad_norm": 0.9457677006721497,
      "learning_rate": 1.3781998581716427e-05,
      "loss": 0.125,
      "step": 79
    },
    {
      "epoch": 1.3175501801338136,
      "grad_norm": 0.9976021647453308,
      "learning_rate": 1.3601777248047105e-05,
      "loss": 0.1256,
      "step": 80
    },
    {
      "epoch": 1.3340195573854863,
      "grad_norm": 1.0927587747573853,
      "learning_rate": 1.342020143325669e-05,
      "loss": 0.1265,
      "step": 81
    },
    {
      "epoch": 1.3504889346371591,
      "grad_norm": 1.0087074041366577,
      "learning_rate": 1.3237339420583213e-05,
      "loss": 0.1252,
      "step": 82
    },
    {
      "epoch": 1.3669583118888318,
      "grad_norm": 1.046044945716858,
      "learning_rate": 1.3053259976951134e-05,
      "loss": 0.1247,
      "step": 83
    },
    {
      "epoch": 1.3834276891405044,
      "grad_norm": 1.0403574705123901,
      "learning_rate": 1.2868032327110904e-05,
      "loss": 0.1254,
      "step": 84
    },
    {
      "epoch": 1.399897066392177,
      "grad_norm": 0.9709586501121521,
      "learning_rate": 1.2681726127606374e-05,
      "loss": 0.1295,
      "step": 85
    },
    {
      "epoch": 1.4163664436438497,
      "grad_norm": 0.9234594106674194,
      "learning_rate": 1.2494411440579814e-05,
      "loss": 0.1226,
      "step": 86
    },
    {
      "epoch": 1.4328358208955223,
      "grad_norm": 1.01310133934021,
      "learning_rate": 1.2306158707424402e-05,
      "loss": 0.1243,
      "step": 87
    },
    {
      "epoch": 1.449305198147195,
      "grad_norm": 0.977340579032898,
      "learning_rate": 1.211703872229411e-05,
      "loss": 0.1252,
      "step": 88
    },
    {
      "epoch": 1.4657745753988678,
      "grad_norm": 0.9476578235626221,
      "learning_rate": 1.1927122605480899e-05,
      "loss": 0.1235,
      "step": 89
    },
    {
      "epoch": 1.4822439526505404,
      "grad_norm": 0.9919003248214722,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.1238,
      "step": 90
    },
    {
      "epoch": 1.498713329902213,
      "grad_norm": 0.9731943011283875,
      "learning_rate": 1.1545187928078407e-05,
      "loss": 0.1269,
      "step": 91
    },
    {
      "epoch": 1.5151827071538857,
      "grad_norm": 0.9613791108131409,
      "learning_rate": 1.1353312997501313e-05,
      "loss": 0.1278,
      "step": 92
    },
    {
      "epoch": 1.5316520844055583,
      "grad_norm": 0.9685577750205994,
      "learning_rate": 1.1160929141252303e-05,
      "loss": 0.1255,
      "step": 93
    },
    {
      "epoch": 1.5481214616572312,
      "grad_norm": 0.9065141081809998,
      "learning_rate": 1.0968108707031792e-05,
      "loss": 0.1229,
      "step": 94
    },
    {
      "epoch": 1.5645908389089036,
      "grad_norm": 0.9448639750480652,
      "learning_rate": 1.077492420671931e-05,
      "loss": 0.1228,
      "step": 95
    },
    {
      "epoch": 1.5810602161605765,
      "grad_norm": 0.91968834400177,
      "learning_rate": 1.0581448289104759e-05,
      "loss": 0.1238,
      "step": 96
    },
    {
      "epoch": 1.597529593412249,
      "grad_norm": 0.9806727170944214,
      "learning_rate": 1.038775371256817e-05,
      "loss": 0.1256,
      "step": 97
    },
    {
      "epoch": 1.6139989706639217,
      "grad_norm": 0.9167839884757996,
      "learning_rate": 1.0193913317718245e-05,
      "loss": 0.1249,
      "step": 98
    },
    {
      "epoch": 1.6304683479155946,
      "grad_norm": 0.9122770428657532,
      "learning_rate": 1e-05,
      "loss": 0.1254,
      "step": 99
    },
    {
      "epoch": 1.646937725167267,
      "grad_norm": 0.9491678476333618,
      "learning_rate": 9.806086682281759e-06,
      "loss": 0.1216,
      "step": 100
    },
    {
      "epoch": 1.6634071024189399,
      "grad_norm": 0.8831928968429565,
      "learning_rate": 9.612246287431832e-06,
      "loss": 0.1212,
      "step": 101
    },
    {
      "epoch": 1.6798764796706125,
      "grad_norm": 0.8818361759185791,
      "learning_rate": 9.418551710895243e-06,
      "loss": 0.1206,
      "step": 102
    },
    {
      "epoch": 1.6963458569222851,
      "grad_norm": 0.9016006588935852,
      "learning_rate": 9.225075793280693e-06,
      "loss": 0.1218,
      "step": 103
    },
    {
      "epoch": 1.7128152341739578,
      "grad_norm": 0.9079029560089111,
      "learning_rate": 9.03189129296821e-06,
      "loss": 0.1198,
      "step": 104
    },
    {
      "epoch": 1.7292846114256304,
      "grad_norm": 0.9520952105522156,
      "learning_rate": 8.839070858747697e-06,
      "loss": 0.123,
      "step": 105
    },
    {
      "epoch": 1.7457539886773032,
      "grad_norm": 0.9180439114570618,
      "learning_rate": 8.646687002498692e-06,
      "loss": 0.1213,
      "step": 106
    },
    {
      "epoch": 1.7622233659289757,
      "grad_norm": 0.8695424199104309,
      "learning_rate": 8.454812071921597e-06,
      "loss": 0.1205,
      "step": 107
    },
    {
      "epoch": 1.7786927431806485,
      "grad_norm": 0.8801153302192688,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.1169,
      "step": 108
    },
    {
      "epoch": 1.7951621204323212,
      "grad_norm": 0.8642145395278931,
      "learning_rate": 8.072877394519103e-06,
      "loss": 0.1176,
      "step": 109
    },
    {
      "epoch": 1.8116314976839938,
      "grad_norm": 0.8979960680007935,
      "learning_rate": 7.882961277705897e-06,
      "loss": 0.1204,
      "step": 110
    },
    {
      "epoch": 1.8281008749356666,
      "grad_norm": 0.820036768913269,
      "learning_rate": 7.6938412925756e-06,
      "loss": 0.1213,
      "step": 111
    },
    {
      "epoch": 1.844570252187339,
      "grad_norm": 0.8929469585418701,
      "learning_rate": 7.505588559420188e-06,
      "loss": 0.1186,
      "step": 112
    },
    {
      "epoch": 1.861039629439012,
      "grad_norm": 0.8560171127319336,
      "learning_rate": 7.3182738723936255e-06,
      "loss": 0.1207,
      "step": 113
    },
    {
      "epoch": 1.8775090066906845,
      "grad_norm": 0.8577896952629089,
      "learning_rate": 7.131967672889101e-06,
      "loss": 0.119,
      "step": 114
    },
    {
      "epoch": 1.8939783839423572,
      "grad_norm": 0.8649168610572815,
      "learning_rate": 6.94674002304887e-06,
      "loss": 0.1171,
      "step": 115
    },
    {
      "epoch": 1.9104477611940298,
      "grad_norm": 0.8647825121879578,
      "learning_rate": 6.762660579416791e-06,
      "loss": 0.119,
      "step": 116
    },
    {
      "epoch": 1.9269171384457024,
      "grad_norm": 0.8913580775260925,
      "learning_rate": 6.579798566743314e-06,
      "loss": 0.1178,
      "step": 117
    },
    {
      "epoch": 1.9433865156973753,
      "grad_norm": 0.8711913824081421,
      "learning_rate": 6.3982227519528986e-06,
      "loss": 0.1177,
      "step": 118
    },
    {
      "epoch": 1.9598558929490477,
      "grad_norm": 0.887152910232544,
      "learning_rate": 6.218001418283577e-06,
      "loss": 0.1208,
      "step": 119
    },
    {
      "epoch": 1.9763252702007206,
      "grad_norm": 0.9313440918922424,
      "learning_rate": 6.039202339608432e-06,
      "loss": 0.1174,
      "step": 120
    },
    {
      "epoch": 1.9927946474523932,
      "grad_norm": 0.971422553062439,
      "learning_rate": 5.8618927549486095e-06,
      "loss": 0.1205,
      "step": 121
    },
    {
      "epoch": 2.009264024704066,
      "grad_norm": 0.9346736669540405,
      "learning_rate": 5.686139343187468e-06,
      "loss": 0.1045,
      "step": 122
    },
    {
      "epoch": 2.0257334019557387,
      "grad_norm": 0.9971660375595093,
      "learning_rate": 5.512008197995379e-06,
      "loss": 0.0927,
      "step": 123
    },
    {
      "epoch": 2.042202779207411,
      "grad_norm": 0.8701250553131104,
      "learning_rate": 5.339564802974615e-06,
      "loss": 0.0917,
      "step": 124
    },
    {
      "epoch": 2.058672156459084,
      "grad_norm": 0.9550381898880005,
      "learning_rate": 5.168874007033615e-06,
      "loss": 0.089,
      "step": 125
    },
    {
      "epoch": 2.0751415337107564,
      "grad_norm": 0.985183835029602,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0878,
      "step": 126
    },
    {
      "epoch": 2.0916109109624292,
      "grad_norm": 1.021644949913025,
      "learning_rate": 4.8330062884813714e-06,
      "loss": 0.0877,
      "step": 127
    },
    {
      "epoch": 2.108080288214102,
      "grad_norm": 0.967758297920227,
      "learning_rate": 4.66795567198309e-06,
      "loss": 0.0888,
      "step": 128
    },
    {
      "epoch": 2.1245496654657745,
      "grad_norm": 0.9479052424430847,
      "learning_rate": 4.504910219291941e-06,
      "loss": 0.0893,
      "step": 129
    },
    {
      "epoch": 2.1410190427174474,
      "grad_norm": 0.8988873362541199,
      "learning_rate": 4.343931245134616e-06,
      "loss": 0.0885,
      "step": 130
    },
    {
      "epoch": 2.1574884199691198,
      "grad_norm": 0.8895907402038574,
      "learning_rate": 4.185079287119733e-06,
      "loss": 0.0875,
      "step": 131
    },
    {
      "epoch": 2.1739577972207926,
      "grad_norm": 0.8918712139129639,
      "learning_rate": 4.028414082972141e-06,
      "loss": 0.0874,
      "step": 132
    },
    {
      "epoch": 2.1904271744724655,
      "grad_norm": 0.8613003492355347,
      "learning_rate": 3.873994548067972e-06,
      "loss": 0.0864,
      "step": 133
    },
    {
      "epoch": 2.206896551724138,
      "grad_norm": 0.8328737020492554,
      "learning_rate": 3.7218787532790167e-06,
      "loss": 0.0869,
      "step": 134
    },
    {
      "epoch": 2.2233659289758108,
      "grad_norm": 0.8785966038703918,
      "learning_rate": 3.5721239031346067e-06,
      "loss": 0.0866,
      "step": 135
    },
    {
      "epoch": 2.239835306227483,
      "grad_norm": 0.9229673743247986,
      "learning_rate": 3.424786314309365e-06,
      "loss": 0.0865,
      "step": 136
    },
    {
      "epoch": 2.256304683479156,
      "grad_norm": 0.9559910893440247,
      "learning_rate": 3.279921394444776e-06,
      "loss": 0.0865,
      "step": 137
    },
    {
      "epoch": 2.2727740607308284,
      "grad_norm": 0.9126923680305481,
      "learning_rate": 3.1375836213126653e-06,
      "loss": 0.0853,
      "step": 138
    },
    {
      "epoch": 2.2892434379825013,
      "grad_norm": 0.8650088310241699,
      "learning_rate": 2.9978265223283152e-06,
      "loss": 0.0859,
      "step": 139
    },
    {
      "epoch": 2.305712815234174,
      "grad_norm": 0.8619425296783447,
      "learning_rate": 2.8607026544210115e-06,
      "loss": 0.0854,
      "step": 140
    },
    {
      "epoch": 2.3221821924858466,
      "grad_norm": 0.930046558380127,
      "learning_rate": 2.726263584269513e-06,
      "loss": 0.088,
      "step": 141
    },
    {
      "epoch": 2.3386515697375194,
      "grad_norm": 0.8604719638824463,
      "learning_rate": 2.594559868909956e-06,
      "loss": 0.0853,
      "step": 142
    },
    {
      "epoch": 2.355120946989192,
      "grad_norm": 0.8624026775360107,
      "learning_rate": 2.4656410367233928e-06,
      "loss": 0.0856,
      "step": 143
    },
    {
      "epoch": 2.3715903242408647,
      "grad_norm": 0.8655623197555542,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.087,
      "step": 144
    },
    {
      "epoch": 2.388059701492537,
      "grad_norm": 0.8891249895095825,
      "learning_rate": 2.2163508807584e-06,
      "loss": 0.0875,
      "step": 145
    },
    {
      "epoch": 2.40452907874421,
      "grad_norm": 0.8799765706062317,
      "learning_rate": 2.0960733048124082e-06,
      "loss": 0.0861,
      "step": 146
    },
    {
      "epoch": 2.420998455995883,
      "grad_norm": 0.8611019849777222,
      "learning_rate": 1.9787680724495617e-06,
      "loss": 0.0849,
      "step": 147
    },
    {
      "epoch": 2.4374678332475552,
      "grad_norm": 0.8410822749137878,
      "learning_rate": 1.8644792973703252e-06,
      "loss": 0.085,
      "step": 148
    },
    {
      "epoch": 2.453937210499228,
      "grad_norm": 0.9146021604537964,
      "learning_rate": 1.7532499589089324e-06,
      "loss": 0.0852,
      "step": 149
    },
    {
      "epoch": 2.4704065877509005,
      "grad_norm": 0.875053346157074,
      "learning_rate": 1.6451218858706374e-06,
      "loss": 0.0852,
      "step": 150
    },
    {
      "epoch": 2.4868759650025734,
      "grad_norm": 0.8440834283828735,
      "learning_rate": 1.5401357408015893e-06,
      "loss": 0.086,
      "step": 151
    },
    {
      "epoch": 2.5033453422542458,
      "grad_norm": 0.8311520218849182,
      "learning_rate": 1.4383310046973365e-06,
      "loss": 0.0849,
      "step": 152
    },
    {
      "epoch": 2.5198147195059186,
      "grad_norm": 0.8406962752342224,
      "learning_rate": 1.339745962155613e-06,
      "loss": 0.083,
      "step": 153
    },
    {
      "epoch": 2.5362840967575915,
      "grad_norm": 0.8209373354911804,
      "learning_rate": 1.2444176869790925e-06,
      "loss": 0.0813,
      "step": 154
    },
    {
      "epoch": 2.552753474009264,
      "grad_norm": 0.8779584169387817,
      "learning_rate": 1.152382028233422e-06,
      "loss": 0.0868,
      "step": 155
    },
    {
      "epoch": 2.5692228512609367,
      "grad_norm": 0.8810158967971802,
      "learning_rate": 1.0636735967658785e-06,
      "loss": 0.0865,
      "step": 156
    },
    {
      "epoch": 2.5856922285126096,
      "grad_norm": 0.865736722946167,
      "learning_rate": 9.783257521896228e-07,
      "loss": 0.0846,
      "step": 157
    },
    {
      "epoch": 2.602161605764282,
      "grad_norm": 0.845588207244873,
      "learning_rate": 8.963705903385344e-07,
      "loss": 0.0847,
      "step": 158
    },
    {
      "epoch": 2.618630983015955,
      "grad_norm": 0.815373420715332,
      "learning_rate": 8.178389311972612e-07,
      "loss": 0.0845,
      "step": 159
    },
    {
      "epoch": 2.6351003602676273,
      "grad_norm": 0.7676054835319519,
      "learning_rate": 7.427603073110967e-07,
      "loss": 0.0833,
      "step": 160
    },
    {
      "epoch": 2.6515697375193,
      "grad_norm": 0.7454907894134521,
      "learning_rate": 6.711629526799946e-07,
      "loss": 0.0819,
      "step": 161
    },
    {
      "epoch": 2.6680391147709726,
      "grad_norm": 0.7519494891166687,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.0847,
      "step": 162
    },
    {
      "epoch": 2.6845084920226454,
      "grad_norm": 0.7272472977638245,
      "learning_rate": 5.385184312424973e-07,
      "loss": 0.0823,
      "step": 163
    },
    {
      "epoch": 2.7009778692743183,
      "grad_norm": 0.7485241293907166,
      "learning_rate": 4.775211466158469e-07,
      "loss": 0.0865,
      "step": 164
    },
    {
      "epoch": 2.7174472465259907,
      "grad_norm": 0.7801215052604675,
      "learning_rate": 4.2010487684511105e-07,
      "loss": 0.0858,
      "step": 165
    },
    {
      "epoch": 2.7339166237776635,
      "grad_norm": 0.7571792602539062,
      "learning_rate": 3.662912138411967e-07,
      "loss": 0.083,
      "step": 166
    },
    {
      "epoch": 2.750386001029336,
      "grad_norm": 0.7191453576087952,
      "learning_rate": 3.161003947219421e-07,
      "loss": 0.0833,
      "step": 167
    },
    {
      "epoch": 2.766855378281009,
      "grad_norm": 0.7716163992881775,
      "learning_rate": 2.6955129420176193e-07,
      "loss": 0.0841,
      "step": 168
    },
    {
      "epoch": 2.783324755532681,
      "grad_norm": 0.7539470195770264,
      "learning_rate": 2.2666141749364434e-07,
      "loss": 0.0836,
      "step": 169
    },
    {
      "epoch": 2.799794132784354,
      "grad_norm": 0.7242315411567688,
      "learning_rate": 1.874468937261531e-07,
      "loss": 0.0821,
      "step": 170
    },
    {
      "epoch": 2.816263510036027,
      "grad_norm": 0.7200111746788025,
      "learning_rate": 1.519224698779198e-07,
      "loss": 0.0837,
      "step": 171
    },
    {
      "epoch": 2.8327328872876993,
      "grad_norm": 0.7370088696479797,
      "learning_rate": 1.201015052319099e-07,
      "loss": 0.084,
      "step": 172
    },
    {
      "epoch": 2.849202264539372,
      "grad_norm": 0.7719192504882812,
      "learning_rate": 9.199596635154684e-08,
      "loss": 0.0837,
      "step": 173
    },
    {
      "epoch": 2.8656716417910446,
      "grad_norm": 0.7894994020462036,
      "learning_rate": 6.761642258056977e-08,
      "loss": 0.0835,
      "step": 174
    },
    {
      "epoch": 2.8821410190427175,
      "grad_norm": 0.7965743541717529,
      "learning_rate": 4.6972042068341714e-08,
      "loss": 0.0823,
      "step": 175
    },
    {
      "epoch": 2.89861039629439,
      "grad_norm": 0.7946169376373291,
      "learning_rate": 3.0070588322079765e-08,
      "loss": 0.0806,
      "step": 176
    },
    {
      "epoch": 2.9150797735460627,
      "grad_norm": 0.8272018432617188,
      "learning_rate": 1.6918417287318245e-08,
      "loss": 0.0828,
      "step": 177
    },
    {
      "epoch": 2.9315491507977356,
      "grad_norm": 0.8682279586791992,
      "learning_rate": 7.520474957699586e-09,
      "loss": 0.0849,
      "step": 178
    },
    {
      "epoch": 2.948018528049408,
      "grad_norm": 0.834924578666687,
      "learning_rate": 1.8802955149865854e-09,
      "loss": 0.0837,
      "step": 179
    },
    {
      "epoch": 2.964487905301081,
      "grad_norm": 0.8157399892807007,
      "learning_rate": 0.0,
      "loss": 0.0827,
      "step": 180
    },
    {
      "epoch": 2.964487905301081,
      "step": 180,
      "total_flos": 8.05255292248916e+18,
      "train_loss": 0.14365306976768705,
      "train_runtime": 16152.8942,
      "train_samples_per_second": 5.774,
      "train_steps_per_second": 0.011
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 8.05255292248916e+18,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}