{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 1479,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.002028397565922921,
      "grad_norm": 5.8069485921507455,
      "learning_rate": 6.756756756756757e-08,
      "loss": 0.8777,
      "step": 1
    },
    {
      "epoch": 0.004056795131845842,
      "grad_norm": 5.850878998479943,
      "learning_rate": 1.3513513513513515e-07,
      "loss": 0.8494,
      "step": 2
    },
    {
      "epoch": 0.006085192697768763,
      "grad_norm": 5.740125451612867,
      "learning_rate": 2.0270270270270273e-07,
      "loss": 0.8494,
      "step": 3
    },
    {
      "epoch": 0.008113590263691683,
      "grad_norm": 5.844789276823164,
      "learning_rate": 2.702702702702703e-07,
      "loss": 0.8642,
      "step": 4
    },
    {
      "epoch": 0.010141987829614604,
      "grad_norm": 5.883573839642794,
      "learning_rate": 3.378378378378379e-07,
      "loss": 0.8593,
      "step": 5
    },
    {
      "epoch": 0.012170385395537525,
      "grad_norm": 5.777450352442993,
      "learning_rate": 4.0540540540540546e-07,
      "loss": 0.8564,
      "step": 6
    },
    {
      "epoch": 0.014198782961460446,
      "grad_norm": 5.770126637722235,
      "learning_rate": 4.7297297297297305e-07,
      "loss": 0.8638,
      "step": 7
    },
    {
      "epoch": 0.016227180527383367,
      "grad_norm": 5.628642387143134,
      "learning_rate": 5.405405405405406e-07,
      "loss": 0.8358,
      "step": 8
    },
    {
      "epoch": 0.018255578093306288,
      "grad_norm": 5.766706356615061,
      "learning_rate": 6.081081081081082e-07,
      "loss": 0.8596,
      "step": 9
    },
    {
      "epoch": 0.02028397565922921,
      "grad_norm": 5.427973662475351,
      "learning_rate": 6.756756756756758e-07,
      "loss": 0.8546,
      "step": 10
    },
    {
      "epoch": 0.02231237322515213,
      "grad_norm": 5.312769960474894,
      "learning_rate": 7.432432432432434e-07,
      "loss": 0.8614,
      "step": 11
    },
    {
      "epoch": 0.02434077079107505,
      "grad_norm": 5.429098069478728,
      "learning_rate": 8.108108108108109e-07,
      "loss": 0.8695,
      "step": 12
    },
    {
      "epoch": 0.02636916835699797,
      "grad_norm": 4.330815436989572,
      "learning_rate": 8.783783783783785e-07,
      "loss": 0.8163,
      "step": 13
    },
    {
      "epoch": 0.028397565922920892,
      "grad_norm": 4.330665121243353,
      "learning_rate": 9.459459459459461e-07,
      "loss": 0.7996,
      "step": 14
    },
    {
      "epoch": 0.030425963488843813,
      "grad_norm": 4.034587881930939,
      "learning_rate": 1.0135135135135136e-06,
      "loss": 0.806,
      "step": 15
    },
    {
      "epoch": 0.032454361054766734,
      "grad_norm": 3.977426411014994,
      "learning_rate": 1.0810810810810812e-06,
      "loss": 0.8158,
      "step": 16
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 2.317932740634394,
      "learning_rate": 1.148648648648649e-06,
      "loss": 0.7731,
      "step": 17
    },
    {
      "epoch": 0.036511156186612576,
      "grad_norm": 2.2030246894922905,
      "learning_rate": 1.2162162162162164e-06,
      "loss": 0.7645,
      "step": 18
    },
    {
      "epoch": 0.038539553752535496,
      "grad_norm": 2.149577637207114,
      "learning_rate": 1.2837837837837838e-06,
      "loss": 0.7879,
      "step": 19
    },
    {
      "epoch": 0.04056795131845842,
      "grad_norm": 1.9820351548369777,
      "learning_rate": 1.3513513513513515e-06,
      "loss": 0.754,
      "step": 20
    },
    {
      "epoch": 0.04259634888438134,
      "grad_norm": 1.8835614715083024,
      "learning_rate": 1.418918918918919e-06,
      "loss": 0.7749,
      "step": 21
    },
    {
      "epoch": 0.04462474645030426,
      "grad_norm": 1.8849807905906102,
      "learning_rate": 1.4864864864864868e-06,
      "loss": 0.7579,
      "step": 22
    },
    {
      "epoch": 0.04665314401622718,
      "grad_norm": 1.8422847573333128,
      "learning_rate": 1.5540540540540541e-06,
      "loss": 0.7208,
      "step": 23
    },
    {
      "epoch": 0.0486815415821501,
      "grad_norm": 2.575729722750571,
      "learning_rate": 1.6216216216216219e-06,
      "loss": 0.7213,
      "step": 24
    },
    {
      "epoch": 0.05070993914807302,
      "grad_norm": 2.9152060029759546,
      "learning_rate": 1.6891891891891894e-06,
      "loss": 0.7257,
      "step": 25
    },
    {
      "epoch": 0.05273833671399594,
      "grad_norm": 2.9188577138990977,
      "learning_rate": 1.756756756756757e-06,
      "loss": 0.7203,
      "step": 26
    },
    {
      "epoch": 0.05476673427991886,
      "grad_norm": 2.848731869351055,
      "learning_rate": 1.8243243243243245e-06,
      "loss": 0.7276,
      "step": 27
    },
    {
      "epoch": 0.056795131845841784,
      "grad_norm": 2.436708068419815,
      "learning_rate": 1.8918918918918922e-06,
      "loss": 0.6903,
      "step": 28
    },
    {
      "epoch": 0.058823529411764705,
      "grad_norm": 2.084454619241589,
      "learning_rate": 1.9594594594594595e-06,
      "loss": 0.6985,
      "step": 29
    },
    {
      "epoch": 0.060851926977687626,
      "grad_norm": 1.8495938896153428,
      "learning_rate": 2.0270270270270273e-06,
      "loss": 0.6946,
      "step": 30
    },
    {
      "epoch": 0.06288032454361055,
      "grad_norm": 1.180656210059394,
      "learning_rate": 2.0945945945945946e-06,
      "loss": 0.6879,
      "step": 31
    },
    {
      "epoch": 0.06490872210953347,
      "grad_norm": 1.0492261248520411,
      "learning_rate": 2.1621621621621623e-06,
      "loss": 0.6734,
      "step": 32
    },
    {
      "epoch": 0.06693711967545639,
      "grad_norm": 1.0644912626466227,
      "learning_rate": 2.22972972972973e-06,
      "loss": 0.667,
      "step": 33
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 1.105170724025192,
      "learning_rate": 2.297297297297298e-06,
      "loss": 0.6736,
      "step": 34
    },
    {
      "epoch": 0.07099391480730223,
      "grad_norm": 1.1208333621093622,
      "learning_rate": 2.364864864864865e-06,
      "loss": 0.6692,
      "step": 35
    },
    {
      "epoch": 0.07302231237322515,
      "grad_norm": 0.9603646415019718,
      "learning_rate": 2.432432432432433e-06,
      "loss": 0.6576,
      "step": 36
    },
    {
      "epoch": 0.07505070993914807,
      "grad_norm": 0.8703465050866678,
      "learning_rate": 2.5e-06,
      "loss": 0.6692,
      "step": 37
    },
    {
      "epoch": 0.07707910750507099,
      "grad_norm": 0.7374675628711646,
      "learning_rate": 2.5675675675675675e-06,
      "loss": 0.6557,
      "step": 38
    },
    {
      "epoch": 0.07910750507099391,
      "grad_norm": 0.8089195153377154,
      "learning_rate": 2.6351351351351353e-06,
      "loss": 0.6633,
      "step": 39
    },
    {
      "epoch": 0.08113590263691683,
      "grad_norm": 0.6903817963282142,
      "learning_rate": 2.702702702702703e-06,
      "loss": 0.6717,
      "step": 40
    },
    {
      "epoch": 0.08316430020283976,
      "grad_norm": 0.7049880557365092,
      "learning_rate": 2.7702702702702703e-06,
      "loss": 0.6615,
      "step": 41
    },
    {
      "epoch": 0.08519269776876268,
      "grad_norm": 0.759300803344897,
      "learning_rate": 2.837837837837838e-06,
      "loss": 0.6589,
      "step": 42
    },
    {
      "epoch": 0.0872210953346856,
      "grad_norm": 0.7539925917198705,
      "learning_rate": 2.9054054054054054e-06,
      "loss": 0.6577,
      "step": 43
    },
    {
      "epoch": 0.08924949290060852,
      "grad_norm": 0.6601002411580605,
      "learning_rate": 2.9729729729729736e-06,
      "loss": 0.6064,
      "step": 44
    },
    {
      "epoch": 0.09127789046653144,
      "grad_norm": 0.600667410945786,
      "learning_rate": 3.040540540540541e-06,
      "loss": 0.6082,
      "step": 45
    },
    {
      "epoch": 0.09330628803245436,
      "grad_norm": 0.504593987361291,
      "learning_rate": 3.1081081081081082e-06,
      "loss": 0.6027,
      "step": 46
    },
    {
      "epoch": 0.09533468559837728,
      "grad_norm": 0.5762802035682117,
      "learning_rate": 3.1756756756756755e-06,
      "loss": 0.6146,
      "step": 47
    },
    {
      "epoch": 0.0973630831643002,
      "grad_norm": 0.6358624840319193,
      "learning_rate": 3.2432432432432437e-06,
      "loss": 0.6098,
      "step": 48
    },
    {
      "epoch": 0.09939148073022312,
      "grad_norm": 0.6245519717815737,
      "learning_rate": 3.310810810810811e-06,
      "loss": 0.6215,
      "step": 49
    },
    {
      "epoch": 0.10141987829614604,
      "grad_norm": 0.53567589129345,
      "learning_rate": 3.3783783783783788e-06,
      "loss": 0.6109,
      "step": 50
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.4739400749676762,
      "learning_rate": 3.445945945945946e-06,
      "loss": 0.6035,
      "step": 51
    },
    {
      "epoch": 0.10547667342799188,
      "grad_norm": 0.5311730473676546,
      "learning_rate": 3.513513513513514e-06,
      "loss": 0.5938,
      "step": 52
    },
    {
      "epoch": 0.1075050709939148,
      "grad_norm": 0.565834908168601,
      "learning_rate": 3.5810810810810816e-06,
      "loss": 0.5948,
      "step": 53
    },
    {
      "epoch": 0.10953346855983773,
      "grad_norm": 0.4438937106056284,
      "learning_rate": 3.648648648648649e-06,
      "loss": 0.6032,
      "step": 54
    },
    {
      "epoch": 0.11156186612576065,
      "grad_norm": 0.42680551935314914,
      "learning_rate": 3.7162162162162162e-06,
      "loss": 0.5888,
      "step": 55
    },
    {
      "epoch": 0.11359026369168357,
      "grad_norm": 0.4948715940286881,
      "learning_rate": 3.7837837837837844e-06,
      "loss": 0.58,
      "step": 56
    },
    {
      "epoch": 0.11561866125760649,
      "grad_norm": 0.41844867666892843,
      "learning_rate": 3.851351351351352e-06,
      "loss": 0.564,
      "step": 57
    },
    {
      "epoch": 0.11764705882352941,
      "grad_norm": 0.427151716250513,
      "learning_rate": 3.918918918918919e-06,
      "loss": 0.5896,
      "step": 58
    },
    {
      "epoch": 0.11967545638945233,
      "grad_norm": 0.4012818562990369,
      "learning_rate": 3.986486486486487e-06,
      "loss": 0.5712,
      "step": 59
    },
    {
      "epoch": 0.12170385395537525,
      "grad_norm": 0.38130473336019244,
      "learning_rate": 4.0540540540540545e-06,
      "loss": 0.583,
      "step": 60
    },
    {
      "epoch": 0.12373225152129817,
      "grad_norm": 0.3584898967159039,
      "learning_rate": 4.121621621621622e-06,
      "loss": 0.5587,
      "step": 61
    },
    {
      "epoch": 0.1257606490872211,
      "grad_norm": 0.3289735953666129,
      "learning_rate": 4.189189189189189e-06,
      "loss": 0.597,
      "step": 62
    },
    {
      "epoch": 0.12778904665314403,
      "grad_norm": 0.33167305013329107,
      "learning_rate": 4.256756756756757e-06,
      "loss": 0.5765,
      "step": 63
    },
    {
      "epoch": 0.12981744421906694,
      "grad_norm": 0.39291328408963805,
      "learning_rate": 4.324324324324325e-06,
      "loss": 0.5993,
      "step": 64
    },
    {
      "epoch": 0.13184584178498987,
      "grad_norm": 0.3556050665824897,
      "learning_rate": 4.391891891891892e-06,
      "loss": 0.5874,
      "step": 65
    },
    {
      "epoch": 0.13387423935091278,
      "grad_norm": 0.299607045836003,
      "learning_rate": 4.45945945945946e-06,
      "loss": 0.5537,
      "step": 66
    },
    {
      "epoch": 0.1359026369168357,
      "grad_norm": 0.33825456987697766,
      "learning_rate": 4.527027027027027e-06,
      "loss": 0.5672,
      "step": 67
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.36427391514007756,
      "learning_rate": 4.594594594594596e-06,
      "loss": 0.5826,
      "step": 68
    },
    {
      "epoch": 0.13995943204868155,
      "grad_norm": 0.3480006590565837,
      "learning_rate": 4.6621621621621625e-06,
      "loss": 0.5587,
      "step": 69
    },
    {
      "epoch": 0.14198782961460446,
      "grad_norm": 0.33274230238458097,
      "learning_rate": 4.72972972972973e-06,
      "loss": 0.5704,
      "step": 70
    },
    {
      "epoch": 0.1440162271805274,
      "grad_norm": 0.35265750924523864,
      "learning_rate": 4.797297297297297e-06,
      "loss": 0.5946,
      "step": 71
    },
    {
      "epoch": 0.1460446247464503,
      "grad_norm": 0.32456231204496,
      "learning_rate": 4.864864864864866e-06,
      "loss": 0.5789,
      "step": 72
    },
    {
      "epoch": 0.14807302231237324,
      "grad_norm": 0.38178368906159404,
      "learning_rate": 4.932432432432433e-06,
      "loss": 0.5786,
      "step": 73
    },
    {
      "epoch": 0.15010141987829614,
      "grad_norm": 0.31763733013666917,
      "learning_rate": 5e-06,
      "loss": 0.5598,
      "step": 74
    },
    {
      "epoch": 0.15212981744421908,
      "grad_norm": 0.3559984526138506,
      "learning_rate": 5.067567567567568e-06,
      "loss": 0.5779,
      "step": 75
    },
    {
      "epoch": 0.15415821501014199,
      "grad_norm": 0.3238301359641807,
      "learning_rate": 5.135135135135135e-06,
      "loss": 0.5653,
      "step": 76
    },
    {
      "epoch": 0.15618661257606492,
      "grad_norm": 0.32836773252992385,
      "learning_rate": 5.202702702702704e-06,
      "loss": 0.5698,
      "step": 77
    },
    {
      "epoch": 0.15821501014198783,
      "grad_norm": 0.3319454869938327,
      "learning_rate": 5.2702702702702705e-06,
      "loss": 0.5488,
      "step": 78
    },
    {
      "epoch": 0.16024340770791076,
      "grad_norm": 0.3282389699169831,
      "learning_rate": 5.337837837837838e-06,
      "loss": 0.546,
      "step": 79
    },
    {
      "epoch": 0.16227180527383367,
      "grad_norm": 0.3064535858153653,
      "learning_rate": 5.405405405405406e-06,
      "loss": 0.5623,
      "step": 80
    },
    {
      "epoch": 0.1643002028397566,
      "grad_norm": 0.30914059780681985,
      "learning_rate": 5.472972972972973e-06,
      "loss": 0.5674,
      "step": 81
    },
    {
      "epoch": 0.1663286004056795,
      "grad_norm": 0.31042095893039234,
      "learning_rate": 5.540540540540541e-06,
      "loss": 0.5369,
      "step": 82
    },
    {
      "epoch": 0.16835699797160245,
      "grad_norm": 0.3444384763824436,
      "learning_rate": 5.608108108108109e-06,
      "loss": 0.5464,
      "step": 83
    },
    {
      "epoch": 0.17038539553752535,
      "grad_norm": 0.27804971328669237,
      "learning_rate": 5.675675675675676e-06,
      "loss": 0.532,
      "step": 84
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.3349279863100278,
      "learning_rate": 5.743243243243244e-06,
      "loss": 0.5791,
      "step": 85
    },
    {
      "epoch": 0.1744421906693712,
      "grad_norm": 0.3036927573939666,
      "learning_rate": 5.810810810810811e-06,
      "loss": 0.5706,
      "step": 86
    },
    {
      "epoch": 0.17647058823529413,
      "grad_norm": 0.2924064472075852,
      "learning_rate": 5.8783783783783786e-06,
      "loss": 0.5683,
      "step": 87
    },
    {
      "epoch": 0.17849898580121704,
      "grad_norm": 0.30869319799766737,
      "learning_rate": 5.945945945945947e-06,
      "loss": 0.5298,
      "step": 88
    },
    {
      "epoch": 0.18052738336713997,
      "grad_norm": 0.3111326564310403,
      "learning_rate": 6.013513513513514e-06,
      "loss": 0.5571,
      "step": 89
    },
    {
      "epoch": 0.18255578093306288,
      "grad_norm": 0.31781718910741386,
      "learning_rate": 6.081081081081082e-06,
      "loss": 0.5343,
      "step": 90
    },
    {
      "epoch": 0.1845841784989858,
      "grad_norm": 0.3022442165108836,
      "learning_rate": 6.1486486486486495e-06,
      "loss": 0.5398,
      "step": 91
    },
    {
      "epoch": 0.18661257606490872,
      "grad_norm": 0.31275116130800373,
      "learning_rate": 6.2162162162162164e-06,
      "loss": 0.5395,
      "step": 92
    },
    {
      "epoch": 0.18864097363083165,
      "grad_norm": 0.29812899789541797,
      "learning_rate": 6.283783783783784e-06,
      "loss": 0.525,
      "step": 93
    },
    {
      "epoch": 0.19066937119675456,
      "grad_norm": 0.3188868688702798,
      "learning_rate": 6.351351351351351e-06,
      "loss": 0.5284,
      "step": 94
    },
    {
      "epoch": 0.1926977687626775,
      "grad_norm": 0.33171926710697336,
      "learning_rate": 6.41891891891892e-06,
      "loss": 0.5544,
      "step": 95
    },
    {
      "epoch": 0.1947261663286004,
      "grad_norm": 0.358607980267626,
      "learning_rate": 6.486486486486487e-06,
      "loss": 0.5431,
      "step": 96
    },
    {
      "epoch": 0.19675456389452334,
      "grad_norm": 0.3330844932986448,
      "learning_rate": 6.554054054054054e-06,
      "loss": 0.5107,
      "step": 97
    },
    {
      "epoch": 0.19878296146044624,
      "grad_norm": 0.3787236422307977,
      "learning_rate": 6.621621621621622e-06,
      "loss": 0.5635,
      "step": 98
    },
    {
      "epoch": 0.20081135902636918,
      "grad_norm": 0.28291699816796007,
      "learning_rate": 6.689189189189191e-06,
      "loss": 0.5279,
      "step": 99
    },
    {
      "epoch": 0.2028397565922921,
      "grad_norm": 0.2767491566660039,
      "learning_rate": 6.7567567567567575e-06,
      "loss": 0.539,
      "step": 100
    },
    {
      "epoch": 0.20486815415821502,
      "grad_norm": 0.2708166188378995,
      "learning_rate": 6.824324324324325e-06,
      "loss": 0.5228,
      "step": 101
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.3395786370687359,
      "learning_rate": 6.891891891891892e-06,
      "loss": 0.5185,
      "step": 102
    },
    {
      "epoch": 0.20892494929006086,
      "grad_norm": 0.2770275989943144,
      "learning_rate": 6.95945945945946e-06,
      "loss": 0.5231,
      "step": 103
    },
    {
      "epoch": 0.21095334685598377,
      "grad_norm": 0.3360320193199171,
      "learning_rate": 7.027027027027028e-06,
      "loss": 0.5511,
      "step": 104
    },
    {
      "epoch": 0.2129817444219067,
      "grad_norm": 0.3463604119580233,
      "learning_rate": 7.0945945945945946e-06,
      "loss": 0.5357,
      "step": 105
    },
    {
      "epoch": 0.2150101419878296,
      "grad_norm": 0.3293967229589142,
      "learning_rate": 7.162162162162163e-06,
      "loss": 0.5341,
      "step": 106
    },
    {
      "epoch": 0.21703853955375255,
      "grad_norm": 0.31204847131089447,
      "learning_rate": 7.229729729729731e-06,
      "loss": 0.5308,
      "step": 107
    },
    {
      "epoch": 0.21906693711967545,
      "grad_norm": 0.3479546590865063,
      "learning_rate": 7.297297297297298e-06,
      "loss": 0.5406,
      "step": 108
    },
    {
      "epoch": 0.2210953346855984,
      "grad_norm": 0.33107517490759414,
      "learning_rate": 7.3648648648648655e-06,
      "loss": 0.5248,
      "step": 109
    },
    {
      "epoch": 0.2231237322515213,
      "grad_norm": 0.4041748184025717,
      "learning_rate": 7.4324324324324324e-06,
      "loss": 0.5448,
      "step": 110
    },
    {
      "epoch": 0.22515212981744423,
      "grad_norm": 0.3643789430651663,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.528,
      "step": 111
    },
    {
      "epoch": 0.22718052738336714,
      "grad_norm": 0.4058917062588064,
      "learning_rate": 7.567567567567569e-06,
      "loss": 0.5283,
      "step": 112
    },
    {
      "epoch": 0.22920892494929007,
      "grad_norm": 0.32412794964740516,
      "learning_rate": 7.635135135135135e-06,
      "loss": 0.5197,
      "step": 113
    },
    {
      "epoch": 0.23123732251521298,
      "grad_norm": 0.3866574840073195,
      "learning_rate": 7.702702702702704e-06,
      "loss": 0.5253,
      "step": 114
    },
    {
      "epoch": 0.2332657200811359,
      "grad_norm": 0.3169463573575409,
      "learning_rate": 7.77027027027027e-06,
      "loss": 0.524,
      "step": 115
    },
    {
      "epoch": 0.23529411764705882,
      "grad_norm": 0.37440076468451117,
      "learning_rate": 7.837837837837838e-06,
      "loss": 0.5486,
      "step": 116
    },
    {
      "epoch": 0.23732251521298176,
      "grad_norm": 0.33742696412918194,
      "learning_rate": 7.905405405405406e-06,
      "loss": 0.5421,
      "step": 117
    },
    {
      "epoch": 0.23935091277890466,
      "grad_norm": 0.34236694182399147,
      "learning_rate": 7.972972972972974e-06,
      "loss": 0.5254,
      "step": 118
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.3055948544797748,
      "learning_rate": 8.040540540540541e-06,
      "loss": 0.5266,
      "step": 119
    },
    {
      "epoch": 0.2434077079107505,
      "grad_norm": 0.30986859778185455,
      "learning_rate": 8.108108108108109e-06,
      "loss": 0.5372,
      "step": 120
    },
    {
      "epoch": 0.24543610547667344,
      "grad_norm": 0.3202062070674759,
      "learning_rate": 8.175675675675677e-06,
      "loss": 0.5128,
      "step": 121
    },
    {
      "epoch": 0.24746450304259635,
      "grad_norm": 0.39112456229768466,
      "learning_rate": 8.243243243243245e-06,
      "loss": 0.5303,
      "step": 122
    },
    {
      "epoch": 0.24949290060851928,
      "grad_norm": 0.31301688622173,
      "learning_rate": 8.31081081081081e-06,
      "loss": 0.5223,
      "step": 123
    },
    {
      "epoch": 0.2515212981744422,
      "grad_norm": 0.3605097643198615,
      "learning_rate": 8.378378378378378e-06,
      "loss": 0.5264,
      "step": 124
    },
    {
      "epoch": 0.2535496957403651,
      "grad_norm": 0.3893589004570892,
      "learning_rate": 8.445945945945948e-06,
      "loss": 0.5214,
      "step": 125
    },
    {
      "epoch": 0.25557809330628806,
      "grad_norm": 0.34786338226846947,
      "learning_rate": 8.513513513513514e-06,
      "loss": 0.5261,
      "step": 126
    },
    {
      "epoch": 0.25760649087221094,
      "grad_norm": 0.42542277604805984,
      "learning_rate": 8.581081081081082e-06,
      "loss": 0.5176,
      "step": 127
    },
    {
      "epoch": 0.25963488843813387,
      "grad_norm": 0.33235615315292794,
      "learning_rate": 8.64864864864865e-06,
      "loss": 0.5656,
      "step": 128
    },
    {
      "epoch": 0.2616632860040568,
      "grad_norm": 0.4059099927419361,
      "learning_rate": 8.716216216216217e-06,
      "loss": 0.5224,
      "step": 129
    },
    {
      "epoch": 0.26369168356997974,
      "grad_norm": 0.3410073363235178,
      "learning_rate": 8.783783783783785e-06,
      "loss": 0.5245,
      "step": 130
    },
    {
      "epoch": 0.2657200811359026,
      "grad_norm": 0.32553833251925696,
      "learning_rate": 8.851351351351351e-06,
      "loss": 0.5263,
      "step": 131
    },
    {
      "epoch": 0.26774847870182555,
      "grad_norm": 0.346567049342416,
      "learning_rate": 8.91891891891892e-06,
      "loss": 0.5376,
      "step": 132
    },
    {
      "epoch": 0.2697768762677485,
      "grad_norm": 0.33204642106147125,
      "learning_rate": 8.986486486486488e-06,
      "loss": 0.5188,
      "step": 133
    },
    {
      "epoch": 0.2718052738336714,
      "grad_norm": 0.30392710355555247,
      "learning_rate": 9.054054054054054e-06,
      "loss": 0.5068,
      "step": 134
    },
    {
      "epoch": 0.2738336713995943,
      "grad_norm": 0.38680428580613013,
      "learning_rate": 9.121621621621622e-06,
      "loss": 0.5016,
      "step": 135
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.39790152371671095,
      "learning_rate": 9.189189189189191e-06,
      "loss": 0.4988,
      "step": 136
    },
    {
      "epoch": 0.2778904665314402,
      "grad_norm": 0.3861597364032549,
      "learning_rate": 9.256756756756757e-06,
      "loss": 0.5101,
      "step": 137
    },
    {
      "epoch": 0.2799188640973631,
      "grad_norm": 0.3762795670048572,
      "learning_rate": 9.324324324324325e-06,
      "loss": 0.5054,
      "step": 138
    },
    {
      "epoch": 0.281947261663286,
      "grad_norm": 0.4187241647582684,
      "learning_rate": 9.391891891891893e-06,
      "loss": 0.5077,
      "step": 139
    },
    {
      "epoch": 0.2839756592292089,
      "grad_norm": 0.3515395768482751,
      "learning_rate": 9.45945945945946e-06,
      "loss": 0.539,
      "step": 140
    },
    {
      "epoch": 0.28600405679513186,
      "grad_norm": 0.43106611787952276,
      "learning_rate": 9.527027027027028e-06,
      "loss": 0.5062,
      "step": 141
    },
    {
      "epoch": 0.2880324543610548,
      "grad_norm": 0.3687940831567729,
      "learning_rate": 9.594594594594594e-06,
      "loss": 0.5066,
      "step": 142
    },
    {
      "epoch": 0.29006085192697767,
      "grad_norm": 0.38011336724301387,
      "learning_rate": 9.662162162162164e-06,
      "loss": 0.5178,
      "step": 143
    },
    {
      "epoch": 0.2920892494929006,
      "grad_norm": 0.31429200047974387,
      "learning_rate": 9.729729729729732e-06,
      "loss": 0.496,
      "step": 144
    },
    {
      "epoch": 0.29411764705882354,
      "grad_norm": 0.4201125034330777,
      "learning_rate": 9.797297297297298e-06,
      "loss": 0.5159,
      "step": 145
    },
    {
      "epoch": 0.2961460446247465,
      "grad_norm": 0.35209666572194664,
      "learning_rate": 9.864864864864865e-06,
      "loss": 0.5246,
      "step": 146
    },
    {
      "epoch": 0.29817444219066935,
      "grad_norm": 0.4576442414176714,
      "learning_rate": 9.932432432432433e-06,
      "loss": 0.5226,
      "step": 147
    },
    {
      "epoch": 0.3002028397565923,
      "grad_norm": 0.3697394116947825,
      "learning_rate": 1e-05,
      "loss": 0.5114,
      "step": 148
    },
    {
      "epoch": 0.3022312373225152,
      "grad_norm": 0.3765709762154172,
      "learning_rate": 9.999986072170506e-06,
      "loss": 0.5,
      "step": 149
    },
    {
      "epoch": 0.30425963488843816,
      "grad_norm": 0.38218504248369867,
      "learning_rate": 9.999944288759616e-06,
      "loss": 0.5184,
      "step": 150
    },
    {
      "epoch": 0.30628803245436104,
      "grad_norm": 0.39963846872805997,
      "learning_rate": 9.99987465000011e-06,
      "loss": 0.5302,
      "step": 151
    },
    {
      "epoch": 0.30831643002028397,
      "grad_norm": 0.31081393562431,
      "learning_rate": 9.999777156279958e-06,
      "loss": 0.519,
      "step": 152
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.4336147363463148,
      "learning_rate": 9.999651808142305e-06,
      "loss": 0.5234,
      "step": 153
    },
    {
      "epoch": 0.31237322515212984,
      "grad_norm": 0.343164098595223,
      "learning_rate": 9.999498606285486e-06,
      "loss": 0.5058,
      "step": 154
    },
    {
      "epoch": 0.3144016227180527,
      "grad_norm": 0.33502197602028255,
      "learning_rate": 9.999317551563011e-06,
      "loss": 0.5082,
      "step": 155
    },
    {
      "epoch": 0.31643002028397565,
      "grad_norm": 0.36831078366585984,
      "learning_rate": 9.999108644983555e-06,
      "loss": 0.5109,
      "step": 156
    },
    {
      "epoch": 0.3184584178498986,
      "grad_norm": 0.4696349505995303,
      "learning_rate": 9.998871887710965e-06,
      "loss": 0.5157,
      "step": 157
    },
    {
      "epoch": 0.3204868154158215,
      "grad_norm": 0.36018562761461775,
      "learning_rate": 9.998607281064248e-06,
      "loss": 0.5052,
      "step": 158
    },
    {
      "epoch": 0.3225152129817444,
      "grad_norm": 0.37213417685180616,
      "learning_rate": 9.998314826517564e-06,
      "loss": 0.52,
      "step": 159
    },
    {
      "epoch": 0.32454361054766734,
      "grad_norm": 0.34676431493821724,
      "learning_rate": 9.997994525700213e-06,
      "loss": 0.5042,
      "step": 160
    },
    {
      "epoch": 0.3265720081135903,
      "grad_norm": 0.31490716915110867,
      "learning_rate": 9.997646380396633e-06,
      "loss": 0.5312,
      "step": 161
    },
    {
      "epoch": 0.3286004056795132,
      "grad_norm": 0.33122176948156223,
      "learning_rate": 9.997270392546389e-06,
      "loss": 0.5177,
      "step": 162
    },
    {
      "epoch": 0.3306288032454361,
      "grad_norm": 0.3710199952434176,
      "learning_rate": 9.996866564244158e-06,
      "loss": 0.5107,
      "step": 163
    },
    {
      "epoch": 0.332657200811359,
      "grad_norm": 0.36144686134319154,
      "learning_rate": 9.996434897739722e-06,
      "loss": 0.5153,
      "step": 164
    },
    {
      "epoch": 0.33468559837728196,
      "grad_norm": 0.3592796047148368,
      "learning_rate": 9.995975395437952e-06,
      "loss": 0.4971,
      "step": 165
    },
    {
      "epoch": 0.3367139959432049,
      "grad_norm": 0.37543179661001796,
      "learning_rate": 9.995488059898792e-06,
      "loss": 0.5143,
      "step": 166
    },
    {
      "epoch": 0.33874239350912777,
      "grad_norm": 0.4148180424160791,
      "learning_rate": 9.994972893837259e-06,
      "loss": 0.5152,
      "step": 167
    },
    {
      "epoch": 0.3407707910750507,
      "grad_norm": 0.4093410100490591,
      "learning_rate": 9.994429900123405e-06,
      "loss": 0.5155,
      "step": 168
    },
    {
      "epoch": 0.34279918864097364,
      "grad_norm": 0.3589614344290518,
      "learning_rate": 9.993859081782322e-06,
      "loss": 0.5149,
      "step": 169
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.4100892176510221,
      "learning_rate": 9.993260441994116e-06,
      "loss": 0.4943,
      "step": 170
    },
    {
      "epoch": 0.34685598377281945,
      "grad_norm": 0.4084745130650046,
      "learning_rate": 9.992633984093886e-06,
      "loss": 0.5113,
      "step": 171
    },
    {
      "epoch": 0.3488843813387424,
      "grad_norm": 0.38033143059952806,
      "learning_rate": 9.99197971157171e-06,
      "loss": 0.5245,
      "step": 172
    },
    {
      "epoch": 0.3509127789046653,
      "grad_norm": 0.3820850140125576,
      "learning_rate": 9.991297628072632e-06,
      "loss": 0.5017,
      "step": 173
    },
    {
      "epoch": 0.35294117647058826,
      "grad_norm": 0.38152261934794357,
      "learning_rate": 9.990587737396623e-06,
      "loss": 0.5272,
      "step": 174
    },
    {
      "epoch": 0.35496957403651114,
      "grad_norm": 0.35331438135794485,
      "learning_rate": 9.98985004349858e-06,
      "loss": 0.5001,
      "step": 175
    },
    {
      "epoch": 0.35699797160243407,
      "grad_norm": 0.35071673381922097,
      "learning_rate": 9.989084550488293e-06,
      "loss": 0.506,
      "step": 176
    },
    {
      "epoch": 0.359026369168357,
      "grad_norm": 0.32166143741020575,
      "learning_rate": 9.988291262630425e-06,
      "loss": 0.5255,
      "step": 177
    },
    {
      "epoch": 0.36105476673427994,
      "grad_norm": 0.39977760638284704,
      "learning_rate": 9.987470184344485e-06,
      "loss": 0.5131,
      "step": 178
    },
    {
      "epoch": 0.3630831643002028,
      "grad_norm": 0.34518755205769946,
      "learning_rate": 9.986621320204813e-06,
      "loss": 0.5141,
      "step": 179
    },
    {
      "epoch": 0.36511156186612576,
      "grad_norm": 0.35341498577325103,
      "learning_rate": 9.985744674940537e-06,
      "loss": 0.5159,
      "step": 180
    },
    {
      "epoch": 0.3671399594320487,
      "grad_norm": 0.3210873140093819,
      "learning_rate": 9.984840253435569e-06,
      "loss": 0.5193,
      "step": 181
    },
    {
      "epoch": 0.3691683569979716,
      "grad_norm": 0.3030852641377623,
      "learning_rate": 9.983908060728554e-06,
      "loss": 0.4865,
      "step": 182
    },
    {
      "epoch": 0.3711967545638945,
      "grad_norm": 0.37377320134538206,
      "learning_rate": 9.982948102012866e-06,
      "loss": 0.5124,
      "step": 183
    },
    {
      "epoch": 0.37322515212981744,
      "grad_norm": 0.35041626046937496,
      "learning_rate": 9.98196038263656e-06,
      "loss": 0.4907,
      "step": 184
    },
    {
      "epoch": 0.3752535496957404,
      "grad_norm": 0.343410804581156,
      "learning_rate": 9.98094490810235e-06,
      "loss": 0.5197,
      "step": 185
    },
    {
      "epoch": 0.3772819472616633,
      "grad_norm": 0.3893400541922996,
      "learning_rate": 9.97990168406758e-06,
      "loss": 0.5031,
      "step": 186
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.31101383145183914,
      "learning_rate": 9.978830716344185e-06,
      "loss": 0.4991,
      "step": 187
    },
    {
      "epoch": 0.3813387423935091,
      "grad_norm": 0.3979767881790893,
      "learning_rate": 9.977732010898672e-06,
      "loss": 0.5097,
      "step": 188
    },
    {
      "epoch": 0.38336713995943206,
      "grad_norm": 0.3256277045157694,
      "learning_rate": 9.976605573852071e-06,
      "loss": 0.4913,
      "step": 189
    },
    {
      "epoch": 0.385395537525355,
      "grad_norm": 0.3633154072495414,
      "learning_rate": 9.975451411479912e-06,
      "loss": 0.4823,
      "step": 190
    },
    {
      "epoch": 0.38742393509127787,
      "grad_norm": 0.3631291452577119,
      "learning_rate": 9.974269530212185e-06,
      "loss": 0.4886,
      "step": 191
    },
    {
      "epoch": 0.3894523326572008,
      "grad_norm": 0.36799707517395047,
      "learning_rate": 9.973059936633308e-06,
      "loss": 0.4874,
      "step": 192
    },
    {
      "epoch": 0.39148073022312374,
      "grad_norm": 0.3639784747461556,
      "learning_rate": 9.971822637482085e-06,
      "loss": 0.4953,
      "step": 193
    },
    {
      "epoch": 0.3935091277890467,
      "grad_norm": 0.35700643811300387,
      "learning_rate": 9.970557639651672e-06,
      "loss": 0.5135,
      "step": 194
    },
    {
      "epoch": 0.39553752535496955,
      "grad_norm": 0.3792662728368027,
      "learning_rate": 9.969264950189539e-06,
      "loss": 0.5107,
      "step": 195
    },
    {
      "epoch": 0.3975659229208925,
      "grad_norm": 0.41832578182080166,
      "learning_rate": 9.96794457629743e-06,
      "loss": 0.5033,
      "step": 196
    },
    {
      "epoch": 0.3995943204868154,
      "grad_norm": 0.35200410815850464,
      "learning_rate": 9.966596525331324e-06,
      "loss": 0.4992,
      "step": 197
    },
    {
      "epoch": 0.40162271805273836,
      "grad_norm": 0.4085442403732693,
      "learning_rate": 9.965220804801386e-06,
      "loss": 0.4956,
      "step": 198
    },
    {
      "epoch": 0.40365111561866124,
      "grad_norm": 0.4221797000917262,
      "learning_rate": 9.96381742237194e-06,
      "loss": 0.4891,
      "step": 199
    },
    {
      "epoch": 0.4056795131845842,
      "grad_norm": 0.41878926968947683,
      "learning_rate": 9.962386385861414e-06,
      "loss": 0.5074,
      "step": 200
    },
    {
      "epoch": 0.4077079107505071,
      "grad_norm": 0.44063650867950754,
      "learning_rate": 9.960927703242298e-06,
      "loss": 0.5067,
      "step": 201
    },
    {
      "epoch": 0.40973630831643004,
      "grad_norm": 0.34823008873667444,
      "learning_rate": 9.959441382641109e-06,
      "loss": 0.4794,
      "step": 202
    },
    {
      "epoch": 0.4117647058823529,
      "grad_norm": 0.435596452545757,
      "learning_rate": 9.957927432338332e-06,
      "loss": 0.5003,
      "step": 203
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.34623803780412615,
      "learning_rate": 9.956385860768387e-06,
      "loss": 0.5173,
      "step": 204
    },
    {
      "epoch": 0.4158215010141988,
      "grad_norm": 0.39949711294476825,
      "learning_rate": 9.954816676519569e-06,
      "loss": 0.5122,
      "step": 205
    },
    {
      "epoch": 0.4178498985801217,
      "grad_norm": 0.4473020033532547,
      "learning_rate": 9.953219888334011e-06,
      "loss": 0.4771,
      "step": 206
    },
    {
      "epoch": 0.4198782961460446,
      "grad_norm": 0.38034953788840503,
      "learning_rate": 9.951595505107633e-06,
      "loss": 0.4939,
      "step": 207
    },
    {
      "epoch": 0.42190669371196754,
      "grad_norm": 0.45028456179210236,
      "learning_rate": 9.949943535890085e-06,
      "loss": 0.5004,
      "step": 208
    },
    {
      "epoch": 0.4239350912778905,
      "grad_norm": 0.36656743991021323,
      "learning_rate": 9.948263989884708e-06,
      "loss": 0.492,
      "step": 209
    },
    {
      "epoch": 0.4259634888438134,
      "grad_norm": 0.4006523167241784,
      "learning_rate": 9.94655687644847e-06,
      "loss": 0.4866,
      "step": 210
    },
    {
      "epoch": 0.4279918864097363,
      "grad_norm": 0.4169485531825728,
      "learning_rate": 9.944822205091929e-06,
      "loss": 0.4921,
      "step": 211
    },
    {
      "epoch": 0.4300202839756592,
      "grad_norm": 0.3818417953879158,
      "learning_rate": 9.943059985479167e-06,
      "loss": 0.4784,
      "step": 212
    },
    {
      "epoch": 0.43204868154158216,
      "grad_norm": 0.42963205578053243,
      "learning_rate": 9.94127022742774e-06,
      "loss": 0.5168,
      "step": 213
    },
    {
      "epoch": 0.4340770791075051,
      "grad_norm": 0.3708658566804409,
      "learning_rate": 9.939452940908627e-06,
      "loss": 0.508,
      "step": 214
    },
    {
      "epoch": 0.43610547667342797,
      "grad_norm": 0.35947853607387537,
      "learning_rate": 9.937608136046171e-06,
      "loss": 0.5038,
      "step": 215
    },
    {
      "epoch": 0.4381338742393509,
      "grad_norm": 0.40313515508252973,
      "learning_rate": 9.935735823118022e-06,
      "loss": 0.4996,
      "step": 216
    },
    {
      "epoch": 0.44016227180527384,
      "grad_norm": 0.3378203905538428,
      "learning_rate": 9.933836012555083e-06,
      "loss": 0.4875,
      "step": 217
    },
    {
      "epoch": 0.4421906693711968,
      "grad_norm": 0.39194017103631246,
      "learning_rate": 9.93190871494145e-06,
      "loss": 0.4889,
      "step": 218
    },
    {
      "epoch": 0.44421906693711966,
      "grad_norm": 0.3401918774132478,
      "learning_rate": 9.929953941014349e-06,
      "loss": 0.4729,
      "step": 219
    },
    {
      "epoch": 0.4462474645030426,
      "grad_norm": 0.4188232215717246,
      "learning_rate": 9.927971701664085e-06,
      "loss": 0.4997,
      "step": 220
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.37944461183932876,
      "learning_rate": 9.925962007933975e-06,
      "loss": 0.4926,
      "step": 221
    },
    {
      "epoch": 0.45030425963488846,
      "grad_norm": 0.4287990388904512,
      "learning_rate": 9.923924871020289e-06,
      "loss": 0.4977,
      "step": 222
    },
    {
      "epoch": 0.45233265720081134,
      "grad_norm": 0.3759217331093869,
      "learning_rate": 9.921860302272184e-06,
      "loss": 0.4879,
      "step": 223
    },
    {
      "epoch": 0.4543610547667343,
      "grad_norm": 0.5050643501498002,
      "learning_rate": 9.919768313191642e-06,
      "loss": 0.4786,
      "step": 224
    },
    {
      "epoch": 0.4563894523326572,
      "grad_norm": 0.3659936264920244,
      "learning_rate": 9.917648915433413e-06,
      "loss": 0.4957,
      "step": 225
    },
    {
      "epoch": 0.45841784989858014,
      "grad_norm": 0.4679847531337614,
      "learning_rate": 9.91550212080494e-06,
      "loss": 0.4829,
      "step": 226
    },
    {
      "epoch": 0.460446247464503,
      "grad_norm": 0.4012821444264157,
      "learning_rate": 9.9133279412663e-06,
      "loss": 0.496,
      "step": 227
    },
    {
      "epoch": 0.46247464503042596,
      "grad_norm": 0.41253850295718336,
      "learning_rate": 9.911126388930133e-06,
      "loss": 0.493,
      "step": 228
    },
    {
      "epoch": 0.4645030425963489,
      "grad_norm": 0.3971799713457988,
      "learning_rate": 9.908897476061576e-06,
      "loss": 0.4853,
      "step": 229
    },
    {
      "epoch": 0.4665314401622718,
      "grad_norm": 0.3565304571721629,
      "learning_rate": 9.906641215078196e-06,
      "loss": 0.4668,
      "step": 230
    },
    {
      "epoch": 0.4685598377281947,
      "grad_norm": 0.34728336856958186,
      "learning_rate": 9.904357618549925e-06,
      "loss": 0.4916,
      "step": 231
    },
    {
      "epoch": 0.47058823529411764,
      "grad_norm": 0.36441687763287006,
      "learning_rate": 9.902046699198976e-06,
      "loss": 0.4764,
      "step": 232
    },
    {
      "epoch": 0.4726166328600406,
      "grad_norm": 0.3687117116546322,
      "learning_rate": 9.899708469899786e-06,
      "loss": 0.4977,
      "step": 233
    },
    {
      "epoch": 0.4746450304259635,
      "grad_norm": 0.38093654063365173,
      "learning_rate": 9.897342943678938e-06,
      "loss": 0.5159,
      "step": 234
    },
    {
      "epoch": 0.4766734279918864,
      "grad_norm": 0.38573944607020483,
      "learning_rate": 9.894950133715094e-06,
      "loss": 0.4866,
      "step": 235
    },
    {
      "epoch": 0.4787018255578093,
      "grad_norm": 0.34467692226989927,
      "learning_rate": 9.892530053338909e-06,
      "loss": 0.4925,
      "step": 236
    },
    {
      "epoch": 0.48073022312373226,
      "grad_norm": 0.32981422021320117,
      "learning_rate": 9.89008271603297e-06,
      "loss": 0.5022,
      "step": 237
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.38016791046775644,
      "learning_rate": 9.88760813543172e-06,
      "loss": 0.5038,
      "step": 238
    },
    {
      "epoch": 0.4847870182555781,
      "grad_norm": 0.36740617995581926,
      "learning_rate": 9.885106325321371e-06,
      "loss": 0.4939,
      "step": 239
    },
    {
      "epoch": 0.486815415821501,
      "grad_norm": 0.33186819330580253,
      "learning_rate": 9.882577299639836e-06,
      "loss": 0.4815,
      "step": 240
    },
    {
      "epoch": 0.48884381338742394,
      "grad_norm": 0.39437011962253965,
      "learning_rate": 9.880021072476651e-06,
      "loss": 0.5005,
      "step": 241
    },
    {
      "epoch": 0.4908722109533469,
      "grad_norm": 0.3257383892325669,
      "learning_rate": 9.877437658072894e-06,
      "loss": 0.5105,
      "step": 242
    },
    {
      "epoch": 0.49290060851926976,
      "grad_norm": 0.33286684488911417,
      "learning_rate": 9.874827070821112e-06,
      "loss": 0.4787,
      "step": 243
    },
    {
      "epoch": 0.4949290060851927,
      "grad_norm": 0.34016308402278844,
      "learning_rate": 9.872189325265222e-06,
      "loss": 0.4877,
      "step": 244
    },
    {
      "epoch": 0.4969574036511156,
      "grad_norm": 0.3037663225329713,
      "learning_rate": 9.869524436100458e-06,
      "loss": 0.4853,
      "step": 245
    },
    {
      "epoch": 0.49898580121703856,
      "grad_norm": 0.3417510337154031,
      "learning_rate": 9.866832418173267e-06,
      "loss": 0.4634,
      "step": 246
    },
    {
      "epoch": 0.5010141987829615,
      "grad_norm": 0.35681090721544806,
      "learning_rate": 9.864113286481237e-06,
      "loss": 0.4925,
      "step": 247
    },
    {
      "epoch": 0.5030425963488844,
      "grad_norm": 0.32706696735026364,
      "learning_rate": 9.861367056173007e-06,
      "loss": 0.5049,
      "step": 248
    },
    {
      "epoch": 0.5050709939148073,
      "grad_norm": 0.3435809368684948,
      "learning_rate": 9.85859374254819e-06,
      "loss": 0.5003,
      "step": 249
    },
    {
      "epoch": 0.5070993914807302,
      "grad_norm": 0.3691569585077096,
      "learning_rate": 9.855793361057283e-06,
      "loss": 0.4764,
      "step": 250
    },
    {
      "epoch": 0.5091277890466531,
      "grad_norm": 0.34164487657630266,
      "learning_rate": 9.852965927301573e-06,
      "loss": 0.4895,
      "step": 251
    },
    {
      "epoch": 0.5111561866125761,
      "grad_norm": 0.32483845791121657,
      "learning_rate": 9.850111457033075e-06,
      "loss": 0.4936,
      "step": 252
    },
    {
      "epoch": 0.513184584178499,
      "grad_norm": 0.3734923583960948,
      "learning_rate": 9.847229966154415e-06,
      "loss": 0.4947,
      "step": 253
    },
    {
      "epoch": 0.5152129817444219,
      "grad_norm": 0.3012101378497654,
      "learning_rate": 9.84432147071876e-06,
      "loss": 0.4834,
      "step": 254
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.37946386612286853,
      "learning_rate": 9.841385986929716e-06,
      "loss": 0.481,
      "step": 255
    },
    {
      "epoch": 0.5192697768762677,
      "grad_norm": 0.4088867807970573,
      "learning_rate": 9.838423531141258e-06,
      "loss": 0.5018,
      "step": 256
    },
    {
      "epoch": 0.5212981744421906,
      "grad_norm": 0.3855786319691473,
      "learning_rate": 9.835434119857612e-06,
      "loss": 0.4964,
      "step": 257
    },
    {
      "epoch": 0.5233265720081136,
      "grad_norm": 0.39817072521514224,
      "learning_rate": 9.832417769733185e-06,
      "loss": 0.4742,
      "step": 258
    },
    {
      "epoch": 0.5253549695740365,
      "grad_norm": 0.3675418848101985,
      "learning_rate": 9.829374497572461e-06,
      "loss": 0.5015,
      "step": 259
    },
    {
      "epoch": 0.5273833671399595,
      "grad_norm": 0.3609443785117819,
      "learning_rate": 9.826304320329909e-06,
      "loss": 0.488,
      "step": 260
    },
    {
      "epoch": 0.5294117647058824,
      "grad_norm": 0.4131257698385137,
      "learning_rate": 9.823207255109891e-06,
      "loss": 0.4929,
      "step": 261
    },
    {
      "epoch": 0.5314401622718052,
      "grad_norm": 0.338670148644499,
      "learning_rate": 9.820083319166568e-06,
      "loss": 0.4735,
      "step": 262
    },
    {
      "epoch": 0.5334685598377282,
      "grad_norm": 0.32339292911160555,
      "learning_rate": 9.816932529903795e-06,
      "loss": 0.4668,
      "step": 263
    },
    {
      "epoch": 0.5354969574036511,
      "grad_norm": 0.33722691156487716,
      "learning_rate": 9.813754904875038e-06,
      "loss": 0.4939,
      "step": 264
    },
    {
      "epoch": 0.537525354969574,
      "grad_norm": 0.3670674339010904,
      "learning_rate": 9.810550461783261e-06,
      "loss": 0.502,
      "step": 265
    },
    {
      "epoch": 0.539553752535497,
      "grad_norm": 0.3054420774343527,
      "learning_rate": 9.807319218480845e-06,
      "loss": 0.4878,
      "step": 266
    },
    {
      "epoch": 0.5415821501014199,
      "grad_norm": 0.3851163312879934,
      "learning_rate": 9.804061192969465e-06,
      "loss": 0.4705,
      "step": 267
    },
    {
      "epoch": 0.5436105476673428,
      "grad_norm": 0.35904419381517816,
      "learning_rate": 9.800776403400014e-06,
      "loss": 0.4759,
      "step": 268
    },
    {
      "epoch": 0.5456389452332657,
      "grad_norm": 0.34076740892749335,
      "learning_rate": 9.797464868072489e-06,
      "loss": 0.5036,
      "step": 269
    },
    {
      "epoch": 0.5476673427991886,
      "grad_norm": 0.4012121126983399,
      "learning_rate": 9.794126605435885e-06,
      "loss": 0.4892,
      "step": 270
    },
    {
      "epoch": 0.5496957403651116,
      "grad_norm": 0.34055376580504937,
      "learning_rate": 9.790761634088108e-06,
      "loss": 0.4713,
      "step": 271
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.3669824597494662,
      "learning_rate": 9.787369972775858e-06,
      "loss": 0.4901,
      "step": 272
    },
    {
      "epoch": 0.5537525354969574,
      "grad_norm": 0.35367247190993767,
      "learning_rate": 9.78395164039452e-06,
      "loss": 0.487,
      "step": 273
    },
    {
      "epoch": 0.5557809330628803,
      "grad_norm": 0.3450604981947872,
      "learning_rate": 9.780506655988082e-06,
      "loss": 0.4594,
      "step": 274
    },
    {
      "epoch": 0.5578093306288032,
      "grad_norm": 0.34072658428072583,
      "learning_rate": 9.777035038749002e-06,
      "loss": 0.4867,
      "step": 275
    },
    {
      "epoch": 0.5598377281947262,
      "grad_norm": 0.34158850069920627,
      "learning_rate": 9.773536808018116e-06,
      "loss": 0.4786,
      "step": 276
    },
    {
      "epoch": 0.5618661257606491,
      "grad_norm": 0.3398118340115372,
      "learning_rate": 9.77001198328453e-06,
      "loss": 0.4805,
      "step": 277
    },
    {
      "epoch": 0.563894523326572,
      "grad_norm": 0.3552458308193095,
      "learning_rate": 9.766460584185511e-06,
      "loss": 0.4983,
      "step": 278
    },
    {
      "epoch": 0.565922920892495,
      "grad_norm": 0.39883785821735546,
      "learning_rate": 9.762882630506366e-06,
      "loss": 0.4771,
      "step": 279
    },
    {
      "epoch": 0.5679513184584178,
      "grad_norm": 0.373407115542334,
      "learning_rate": 9.759278142180348e-06,
      "loss": 0.4883,
      "step": 280
    },
    {
      "epoch": 0.5699797160243407,
      "grad_norm": 0.3929087962792425,
      "learning_rate": 9.75564713928854e-06,
      "loss": 0.4722,
      "step": 281
    },
    {
      "epoch": 0.5720081135902637,
      "grad_norm": 0.39481414979718144,
      "learning_rate": 9.751989642059731e-06,
      "loss": 0.4653,
      "step": 282
    },
    {
      "epoch": 0.5740365111561866,
      "grad_norm": 0.4116135314379905,
      "learning_rate": 9.748305670870326e-06,
      "loss": 0.4915,
      "step": 283
    },
    {
      "epoch": 0.5760649087221096,
      "grad_norm": 0.4084960130510332,
      "learning_rate": 9.74459524624421e-06,
      "loss": 0.4871,
      "step": 284
    },
    {
      "epoch": 0.5780933062880325,
      "grad_norm": 0.4590305346791398,
      "learning_rate": 9.740858388852652e-06,
      "loss": 0.4721,
      "step": 285
    },
    {
      "epoch": 0.5801217038539553,
      "grad_norm": 0.3281298290787826,
      "learning_rate": 9.737095119514172e-06,
      "loss": 0.5015,
      "step": 286
    },
    {
      "epoch": 0.5821501014198783,
      "grad_norm": 0.4370636466827182,
      "learning_rate": 9.733305459194444e-06,
      "loss": 0.4956,
      "step": 287
    },
    {
      "epoch": 0.5841784989858012,
      "grad_norm": 0.33000674205436814,
      "learning_rate": 9.729489429006162e-06,
      "loss": 0.4946,
      "step": 288
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.3284987729715477,
      "learning_rate": 9.725647050208936e-06,
      "loss": 0.4914,
      "step": 289
    },
    {
      "epoch": 0.5882352941176471,
      "grad_norm": 0.35733191029495326,
      "learning_rate": 9.72177834420916e-06,
      "loss": 0.4953,
      "step": 290
    },
    {
      "epoch": 0.59026369168357,
      "grad_norm": 0.3368062729038099,
      "learning_rate": 9.717883332559911e-06,
      "loss": 0.4721,
      "step": 291
    },
    {
      "epoch": 0.592292089249493,
      "grad_norm": 0.3585883136950177,
      "learning_rate": 9.713962036960806e-06,
      "loss": 0.4676,
      "step": 292
    },
    {
      "epoch": 0.5943204868154158,
      "grad_norm": 0.396548610321751,
      "learning_rate": 9.710014479257906e-06,
      "loss": 0.4796,
      "step": 293
    },
    {
      "epoch": 0.5963488843813387,
      "grad_norm": 0.3438109300818228,
      "learning_rate": 9.70604068144357e-06,
      "loss": 0.4822,
      "step": 294
    },
    {
      "epoch": 0.5983772819472617,
      "grad_norm": 0.3661259664889753,
      "learning_rate": 9.702040665656353e-06,
      "loss": 0.4847,
      "step": 295
    },
    {
      "epoch": 0.6004056795131846,
      "grad_norm": 0.35304897648976535,
      "learning_rate": 9.698014454180868e-06,
      "loss": 0.4842,
      "step": 296
    },
    {
      "epoch": 0.6024340770791075,
      "grad_norm": 0.3987826677896339,
      "learning_rate": 9.693962069447669e-06,
      "loss": 0.4778,
      "step": 297
    },
    {
      "epoch": 0.6044624746450304,
      "grad_norm": 0.3204270008027351,
      "learning_rate": 9.689883534033128e-06,
      "loss": 0.4835,
      "step": 298
    },
    {
      "epoch": 0.6064908722109533,
      "grad_norm": 0.3837853416739516,
      "learning_rate": 9.685778870659301e-06,
      "loss": 0.4889,
      "step": 299
    },
    {
      "epoch": 0.6085192697768763,
      "grad_norm": 0.3858497743703557,
      "learning_rate": 9.681648102193812e-06,
      "loss": 0.4747,
      "step": 300
    },
    {
      "epoch": 0.6105476673427992,
      "grad_norm": 0.42589241859546845,
      "learning_rate": 9.677491251649711e-06,
      "loss": 0.4936,
      "step": 301
    },
    {
      "epoch": 0.6125760649087221,
      "grad_norm": 0.45802182806696295,
      "learning_rate": 9.673308342185366e-06,
      "loss": 0.4825,
      "step": 302
    },
    {
      "epoch": 0.6146044624746451,
      "grad_norm": 0.4039819790042931,
      "learning_rate": 9.669099397104314e-06,
      "loss": 0.4773,
      "step": 303
    },
    {
      "epoch": 0.6166328600405679,
      "grad_norm": 0.3502862886973665,
      "learning_rate": 9.664864439855144e-06,
      "loss": 0.4839,
      "step": 304
    },
    {
      "epoch": 0.6186612576064908,
      "grad_norm": 0.4339980449268265,
      "learning_rate": 9.660603494031358e-06,
      "loss": 0.4758,
      "step": 305
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.36937332294945413,
      "learning_rate": 9.65631658337125e-06,
      "loss": 0.4826,
      "step": 306
    },
    {
      "epoch": 0.6227180527383367,
      "grad_norm": 0.3733681714958432,
      "learning_rate": 9.652003731757763e-06,
      "loss": 0.4966,
      "step": 307
    },
    {
      "epoch": 0.6247464503042597,
      "grad_norm": 0.42396722311560303,
      "learning_rate": 9.647664963218363e-06,
      "loss": 0.4766,
      "step": 308
    },
    {
      "epoch": 0.6267748478701826,
      "grad_norm": 0.36973804514543585,
      "learning_rate": 9.643300301924902e-06,
      "loss": 0.4714,
      "step": 309
    },
    {
      "epoch": 0.6288032454361054,
      "grad_norm": 0.4687665362894364,
      "learning_rate": 9.638909772193478e-06,
      "loss": 0.4791,
      "step": 310
    },
    {
      "epoch": 0.6308316430020284,
      "grad_norm": 0.36246197735201097,
      "learning_rate": 9.634493398484319e-06,
      "loss": 0.4688,
      "step": 311
    },
    {
      "epoch": 0.6328600405679513,
      "grad_norm": 0.3671117941858119,
      "learning_rate": 9.630051205401618e-06,
      "loss": 0.4554,
      "step": 312
    },
    {
      "epoch": 0.6348884381338742,
      "grad_norm": 0.3291787272808045,
      "learning_rate": 9.625583217693419e-06,
      "loss": 0.474,
      "step": 313
    },
    {
      "epoch": 0.6369168356997972,
      "grad_norm": 0.3743523142339175,
      "learning_rate": 9.621089460251474e-06,
      "loss": 0.4865,
      "step": 314
    },
    {
      "epoch": 0.6389452332657201,
      "grad_norm": 0.3550256529312365,
      "learning_rate": 9.616569958111097e-06,
      "loss": 0.4635,
      "step": 315
    },
    {
      "epoch": 0.640973630831643,
      "grad_norm": 0.33845141664102163,
      "learning_rate": 9.612024736451027e-06,
      "loss": 0.4772,
      "step": 316
    },
    {
      "epoch": 0.6430020283975659,
      "grad_norm": 0.3268827414286522,
      "learning_rate": 9.607453820593297e-06,
      "loss": 0.4866,
      "step": 317
    },
    {
      "epoch": 0.6450304259634888,
      "grad_norm": 0.28490605466592356,
      "learning_rate": 9.602857236003078e-06,
      "loss": 0.4805,
      "step": 318
    },
    {
      "epoch": 0.6470588235294118,
      "grad_norm": 0.36071035932042006,
      "learning_rate": 9.598235008288551e-06,
      "loss": 0.4603,
      "step": 319
    },
    {
      "epoch": 0.6490872210953347,
      "grad_norm": 0.31328927581437144,
      "learning_rate": 9.593587163200755e-06,
      "loss": 0.4871,
      "step": 320
    },
    {
      "epoch": 0.6511156186612576,
      "grad_norm": 0.4046088780282955,
      "learning_rate": 9.58891372663345e-06,
      "loss": 0.5087,
      "step": 321
    },
    {
      "epoch": 0.6531440162271805,
      "grad_norm": 0.34114217585470036,
      "learning_rate": 9.584214724622962e-06,
      "loss": 0.4796,
      "step": 322
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.36870439463588756,
      "learning_rate": 9.579490183348052e-06,
      "loss": 0.4884,
      "step": 323
    },
    {
      "epoch": 0.6572008113590264,
      "grad_norm": 0.3695987838066393,
      "learning_rate": 9.574740129129767e-06,
      "loss": 0.4904,
      "step": 324
    },
    {
      "epoch": 0.6592292089249493,
      "grad_norm": 0.3453614270327719,
      "learning_rate": 9.56996458843128e-06,
      "loss": 0.4847,
      "step": 325
    },
    {
      "epoch": 0.6612576064908722,
      "grad_norm": 0.41762658720642526,
      "learning_rate": 9.565163587857757e-06,
      "loss": 0.4728,
      "step": 326
    },
    {
      "epoch": 0.6632860040567952,
      "grad_norm": 0.35649046615544755,
      "learning_rate": 9.56033715415621e-06,
      "loss": 0.4576,
      "step": 327
    },
    {
      "epoch": 0.665314401622718,
      "grad_norm": 0.37110699861315066,
      "learning_rate": 9.555485314215333e-06,
      "loss": 0.4831,
      "step": 328
    },
    {
      "epoch": 0.6673427991886409,
      "grad_norm": 0.38029204422153157,
      "learning_rate": 9.550608095065367e-06,
      "loss": 0.4948,
      "step": 329
    },
    {
      "epoch": 0.6693711967545639,
      "grad_norm": 0.40132015653293013,
      "learning_rate": 9.545705523877944e-06,
      "loss": 0.4801,
      "step": 330
    },
    {
      "epoch": 0.6713995943204868,
      "grad_norm": 0.3705637032069402,
      "learning_rate": 9.540777627965933e-06,
      "loss": 0.4838,
      "step": 331
    },
    {
      "epoch": 0.6734279918864098,
      "grad_norm": 0.4440806820135944,
      "learning_rate": 9.535824434783291e-06,
      "loss": 0.4837,
      "step": 332
    },
    {
      "epoch": 0.6754563894523327,
      "grad_norm": 0.49538574855273615,
      "learning_rate": 9.53084597192491e-06,
      "loss": 0.4925,
      "step": 333
    },
    {
      "epoch": 0.6774847870182555,
      "grad_norm": 0.3820689138558799,
      "learning_rate": 9.525842267126464e-06,
      "loss": 0.4676,
      "step": 334
    },
    {
      "epoch": 0.6795131845841785,
      "grad_norm": 0.379544021434289,
      "learning_rate": 9.520813348264252e-06,
      "loss": 0.4947,
      "step": 335
    },
    {
      "epoch": 0.6815415821501014,
      "grad_norm": 0.35041303541767355,
      "learning_rate": 9.515759243355043e-06,
      "loss": 0.4873,
      "step": 336
    },
    {
      "epoch": 0.6835699797160243,
      "grad_norm": 0.3247312804910836,
      "learning_rate": 9.510679980555922e-06,
      "loss": 0.4739,
      "step": 337
    },
    {
      "epoch": 0.6855983772819473,
      "grad_norm": 0.3554397849658914,
      "learning_rate": 9.505575588164128e-06,
      "loss": 0.466,
      "step": 338
    },
    {
      "epoch": 0.6876267748478702,
      "grad_norm": 0.3374676220062917,
      "learning_rate": 9.500446094616911e-06,
      "loss": 0.4921,
      "step": 339
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.353812530253492,
      "learning_rate": 9.49529152849135e-06,
      "loss": 0.4903,
      "step": 340
    },
    {
      "epoch": 0.691683569979716,
      "grad_norm": 0.37831808795363,
      "learning_rate": 9.490111918504213e-06,
      "loss": 0.4933,
      "step": 341
    },
    {
      "epoch": 0.6937119675456389,
      "grad_norm": 0.36091652115855677,
      "learning_rate": 9.484907293511792e-06,
      "loss": 0.4766,
      "step": 342
    },
    {
      "epoch": 0.6957403651115619,
      "grad_norm": 0.32879826907143306,
      "learning_rate": 9.479677682509737e-06,
      "loss": 0.4741,
      "step": 343
    },
    {
      "epoch": 0.6977687626774848,
      "grad_norm": 0.4400382460889941,
      "learning_rate": 9.4744231146329e-06,
      "loss": 0.4838,
      "step": 344
    },
    {
      "epoch": 0.6997971602434077,
      "grad_norm": 0.3592437772537452,
      "learning_rate": 9.469143619155172e-06,
      "loss": 0.4499,
      "step": 345
    },
    {
      "epoch": 0.7018255578093306,
      "grad_norm": 0.4070444817939314,
      "learning_rate": 9.46383922548932e-06,
      "loss": 0.4791,
      "step": 346
    },
    {
      "epoch": 0.7038539553752535,
      "grad_norm": 0.4204606015374317,
      "learning_rate": 9.458509963186815e-06,
      "loss": 0.4891,
      "step": 347
    },
    {
      "epoch": 0.7058823529411765,
      "grad_norm": 0.39255481056934965,
      "learning_rate": 9.453155861937684e-06,
      "loss": 0.4909,
      "step": 348
    },
    {
      "epoch": 0.7079107505070994,
      "grad_norm": 0.40891186810657887,
      "learning_rate": 9.44777695157033e-06,
      "loss": 0.4772,
      "step": 349
    },
    {
      "epoch": 0.7099391480730223,
      "grad_norm": 0.3581783129668206,
      "learning_rate": 9.442373262051373e-06,
      "loss": 0.4697,
      "step": 350
    },
    {
      "epoch": 0.7119675456389453,
      "grad_norm": 0.386258785592575,
      "learning_rate": 9.436944823485475e-06,
      "loss": 0.4869,
      "step": 351
    },
    {
      "epoch": 0.7139959432048681,
      "grad_norm": 0.3783233295145611,
      "learning_rate": 9.431491666115187e-06,
      "loss": 0.4872,
      "step": 352
    },
    {
      "epoch": 0.716024340770791,
      "grad_norm": 0.38246274466550334,
      "learning_rate": 9.426013820320764e-06,
      "loss": 0.4811,
      "step": 353
    },
    {
      "epoch": 0.718052738336714,
      "grad_norm": 0.3374997468642555,
      "learning_rate": 9.420511316620008e-06,
      "loss": 0.4704,
      "step": 354
    },
    {
      "epoch": 0.7200811359026369,
      "grad_norm": 0.37233932825646016,
      "learning_rate": 9.414984185668097e-06,
      "loss": 0.4979,
      "step": 355
    },
    {
      "epoch": 0.7221095334685599,
      "grad_norm": 0.30360277334490376,
      "learning_rate": 9.409432458257399e-06,
      "loss": 0.4716,
      "step": 356
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.3274282368330117,
      "learning_rate": 9.403856165317322e-06,
      "loss": 0.4744,
      "step": 357
    },
    {
      "epoch": 0.7261663286004056,
      "grad_norm": 0.41131180097732173,
      "learning_rate": 9.398255337914131e-06,
      "loss": 0.5087,
      "step": 358
    },
    {
      "epoch": 0.7281947261663286,
      "grad_norm": 0.3521696717488722,
      "learning_rate": 9.392630007250769e-06,
      "loss": 0.4815,
      "step": 359
    },
    {
      "epoch": 0.7302231237322515,
      "grad_norm": 0.4531806879708264,
      "learning_rate": 9.3869802046667e-06,
      "loss": 0.5075,
      "step": 360
    },
    {
      "epoch": 0.7322515212981744,
      "grad_norm": 0.3719447870457112,
      "learning_rate": 9.381305961637713e-06,
      "loss": 0.4806,
      "step": 361
    },
    {
      "epoch": 0.7342799188640974,
      "grad_norm": 0.40534054868712566,
      "learning_rate": 9.37560730977577e-06,
      "loss": 0.5035,
      "step": 362
    },
    {
      "epoch": 0.7363083164300203,
      "grad_norm": 0.44390465866932305,
      "learning_rate": 9.369884280828806e-06,
      "loss": 0.4653,
      "step": 363
    },
    {
      "epoch": 0.7383367139959433,
      "grad_norm": 0.3575349972454524,
      "learning_rate": 9.364136906680573e-06,
      "loss": 0.4743,
      "step": 364
    },
    {
      "epoch": 0.7403651115618661,
      "grad_norm": 0.47906616709007405,
      "learning_rate": 9.358365219350448e-06,
      "loss": 0.4831,
      "step": 365
    },
    {
      "epoch": 0.742393509127789,
      "grad_norm": 0.3674706092032296,
      "learning_rate": 9.352569250993265e-06,
      "loss": 0.4863,
      "step": 366
    },
    {
      "epoch": 0.744421906693712,
      "grad_norm": 0.44344156527741385,
      "learning_rate": 9.346749033899121e-06,
      "loss": 0.4874,
      "step": 367
    },
    {
      "epoch": 0.7464503042596349,
      "grad_norm": 0.40855954447747156,
      "learning_rate": 9.34090460049322e-06,
      "loss": 0.4776,
      "step": 368
    },
    {
      "epoch": 0.7484787018255578,
      "grad_norm": 0.4184258966124469,
      "learning_rate": 9.335035983335667e-06,
      "loss": 0.4771,
      "step": 369
    },
    {
      "epoch": 0.7505070993914807,
      "grad_norm": 0.43315784268619706,
      "learning_rate": 9.329143215121301e-06,
      "loss": 0.4575,
      "step": 370
    },
    {
      "epoch": 0.7525354969574036,
      "grad_norm": 0.39010283007483293,
      "learning_rate": 9.323226328679512e-06,
      "loss": 0.4588,
      "step": 371
    },
    {
      "epoch": 0.7545638945233266,
      "grad_norm": 0.4281956198968921,
      "learning_rate": 9.317285356974055e-06,
      "loss": 0.471,
      "step": 372
    },
    {
      "epoch": 0.7565922920892495,
      "grad_norm": 0.3658271397556928,
      "learning_rate": 9.311320333102864e-06,
      "loss": 0.4583,
      "step": 373
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.4861648287378186,
      "learning_rate": 9.305331290297876e-06,
      "loss": 0.4762,
      "step": 374
    },
    {
      "epoch": 0.7606490872210954,
      "grad_norm": 0.31226279294007625,
      "learning_rate": 9.299318261924834e-06,
      "loss": 0.4646,
      "step": 375
    },
    {
      "epoch": 0.7626774847870182,
      "grad_norm": 0.4171552815770829,
      "learning_rate": 9.293281281483115e-06,
      "loss": 0.478,
      "step": 376
    },
    {
      "epoch": 0.7647058823529411,
      "grad_norm": 0.3249381166027118,
      "learning_rate": 9.287220382605532e-06,
      "loss": 0.4712,
      "step": 377
    },
    {
      "epoch": 0.7667342799188641,
      "grad_norm": 0.39609995108161844,
      "learning_rate": 9.28113559905815e-06,
      "loss": 0.4714,
      "step": 378
    },
    {
      "epoch": 0.768762677484787,
      "grad_norm": 0.2990518080582066,
      "learning_rate": 9.275026964740101e-06,
      "loss": 0.4631,
      "step": 379
    },
    {
      "epoch": 0.77079107505071,
      "grad_norm": 0.41894992411883947,
      "learning_rate": 9.26889451368339e-06,
      "loss": 0.4923,
      "step": 380
    },
    {
      "epoch": 0.7728194726166329,
      "grad_norm": 0.41440595715889506,
      "learning_rate": 9.262738280052715e-06,
      "loss": 0.4831,
      "step": 381
    },
    {
      "epoch": 0.7748478701825557,
      "grad_norm": 0.33061552357055557,
      "learning_rate": 9.25655829814526e-06,
      "loss": 0.4898,
      "step": 382
    },
    {
      "epoch": 0.7768762677484787,
      "grad_norm": 0.40401420060630927,
      "learning_rate": 9.250354602390523e-06,
      "loss": 0.4734,
      "step": 383
    },
    {
      "epoch": 0.7789046653144016,
      "grad_norm": 0.3575538610127102,
      "learning_rate": 9.244127227350107e-06,
      "loss": 0.469,
      "step": 384
    },
    {
      "epoch": 0.7809330628803245,
      "grad_norm": 0.4488285805446171,
      "learning_rate": 9.237876207717538e-06,
      "loss": 0.4877,
      "step": 385
    },
    {
      "epoch": 0.7829614604462475,
      "grad_norm": 0.43336532963324603,
      "learning_rate": 9.231601578318075e-06,
      "loss": 0.5045,
      "step": 386
    },
    {
      "epoch": 0.7849898580121704,
      "grad_norm": 0.42378828352301,
      "learning_rate": 9.225303374108503e-06,
      "loss": 0.4789,
      "step": 387
    },
    {
      "epoch": 0.7870182555780934,
      "grad_norm": 0.5671804269099896,
      "learning_rate": 9.218981630176947e-06,
      "loss": 0.4882,
      "step": 388
    },
    {
      "epoch": 0.7890466531440162,
      "grad_norm": 0.3745219045528459,
      "learning_rate": 9.212636381742676e-06,
      "loss": 0.4833,
      "step": 389
    },
    {
      "epoch": 0.7910750507099391,
      "grad_norm": 0.4263517390295865,
      "learning_rate": 9.206267664155906e-06,
      "loss": 0.4615,
      "step": 390
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.40712417063351813,
      "learning_rate": 9.199875512897602e-06,
      "loss": 0.4858,
      "step": 391
    },
    {
      "epoch": 0.795131845841785,
      "grad_norm": 0.4375263581287648,
      "learning_rate": 9.19345996357928e-06,
      "loss": 0.4791,
      "step": 392
    },
    {
      "epoch": 0.7971602434077079,
      "grad_norm": 0.39972411460498686,
      "learning_rate": 9.187021051942814e-06,
      "loss": 0.484,
      "step": 393
    },
    {
      "epoch": 0.7991886409736308,
      "grad_norm": 0.414081569693378,
      "learning_rate": 9.180558813860225e-06,
      "loss": 0.4953,
      "step": 394
    },
    {
      "epoch": 0.8012170385395537,
      "grad_norm": 0.32910699460960563,
      "learning_rate": 9.174073285333498e-06,
      "loss": 0.4782,
      "step": 395
    },
    {
      "epoch": 0.8032454361054767,
      "grad_norm": 0.5317915004257037,
      "learning_rate": 9.167564502494361e-06,
      "loss": 0.4713,
      "step": 396
    },
    {
      "epoch": 0.8052738336713996,
      "grad_norm": 0.32120539348853866,
      "learning_rate": 9.161032501604106e-06,
      "loss": 0.4855,
      "step": 397
    },
    {
      "epoch": 0.8073022312373225,
      "grad_norm": 0.5804798913684851,
      "learning_rate": 9.154477319053368e-06,
      "loss": 0.4877,
      "step": 398
    },
    {
      "epoch": 0.8093306288032455,
      "grad_norm": 0.38342729720648305,
      "learning_rate": 9.147898991361936e-06,
      "loss": 0.4774,
      "step": 399
    },
    {
      "epoch": 0.8113590263691683,
      "grad_norm": 0.48153574663887855,
      "learning_rate": 9.141297555178537e-06,
      "loss": 0.4695,
      "step": 400
    },
    {
      "epoch": 0.8133874239350912,
      "grad_norm": 0.3956106727114391,
      "learning_rate": 9.134673047280644e-06,
      "loss": 0.4855,
      "step": 401
    },
    {
      "epoch": 0.8154158215010142,
      "grad_norm": 0.42603223435731363,
      "learning_rate": 9.128025504574264e-06,
      "loss": 0.4779,
      "step": 402
    },
    {
      "epoch": 0.8174442190669371,
      "grad_norm": 0.36937633960410426,
      "learning_rate": 9.121354964093732e-06,
      "loss": 0.4615,
      "step": 403
    },
    {
      "epoch": 0.8194726166328601,
      "grad_norm": 0.37784775023534506,
      "learning_rate": 9.11466146300151e-06,
      "loss": 0.4677,
      "step": 404
    },
    {
      "epoch": 0.821501014198783,
      "grad_norm": 0.34820953864875487,
      "learning_rate": 9.107945038587974e-06,
      "loss": 0.4623,
      "step": 405
    },
    {
      "epoch": 0.8235294117647058,
      "grad_norm": 0.36467498873133636,
      "learning_rate": 9.10120572827121e-06,
      "loss": 0.474,
      "step": 406
    },
    {
      "epoch": 0.8255578093306288,
      "grad_norm": 0.3507650480626536,
      "learning_rate": 9.094443569596802e-06,
      "loss": 0.451,
      "step": 407
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.38028010104697485,
      "learning_rate": 9.08765860023763e-06,
      "loss": 0.4812,
      "step": 408
    },
    {
      "epoch": 0.8296146044624746,
      "grad_norm": 0.34747559963528935,
      "learning_rate": 9.08085085799365e-06,
      "loss": 0.4639,
      "step": 409
    },
    {
      "epoch": 0.8316430020283976,
      "grad_norm": 0.36130493873841935,
      "learning_rate": 9.074020380791694e-06,
      "loss": 0.4862,
      "step": 410
    },
    {
      "epoch": 0.8336713995943205,
      "grad_norm": 0.3388234464782985,
      "learning_rate": 9.067167206685248e-06,
      "loss": 0.4631,
      "step": 411
    },
    {
      "epoch": 0.8356997971602435,
      "grad_norm": 0.3725311521232289,
      "learning_rate": 9.060291373854252e-06,
      "loss": 0.48,
      "step": 412
    },
    {
      "epoch": 0.8377281947261663,
      "grad_norm": 0.3719959900954804,
      "learning_rate": 9.05339292060487e-06,
      "loss": 0.4654,
      "step": 413
    },
    {
      "epoch": 0.8397565922920892,
      "grad_norm": 0.418562098470519,
      "learning_rate": 9.046471885369302e-06,
      "loss": 0.4917,
      "step": 414
    },
    {
      "epoch": 0.8417849898580122,
      "grad_norm": 0.41826521414429696,
      "learning_rate": 9.039528306705543e-06,
      "loss": 0.4778,
      "step": 415
    },
    {
      "epoch": 0.8438133874239351,
      "grad_norm": 0.32882682096468013,
      "learning_rate": 9.032562223297187e-06,
      "loss": 0.468,
      "step": 416
    },
    {
      "epoch": 0.845841784989858,
      "grad_norm": 0.3928572386962237,
      "learning_rate": 9.025573673953201e-06,
      "loss": 0.4815,
      "step": 417
    },
    {
      "epoch": 0.847870182555781,
      "grad_norm": 0.3827888592981188,
      "learning_rate": 9.018562697607716e-06,
      "loss": 0.47,
      "step": 418
    },
    {
      "epoch": 0.8498985801217038,
      "grad_norm": 0.35804441350467364,
      "learning_rate": 9.011529333319804e-06,
      "loss": 0.4787,
      "step": 419
    },
    {
      "epoch": 0.8519269776876268,
      "grad_norm": 0.36884510058085956,
      "learning_rate": 9.004473620273263e-06,
      "loss": 0.4729,
      "step": 420
    },
    {
      "epoch": 0.8539553752535497,
      "grad_norm": 0.3773373236604394,
      "learning_rate": 8.997395597776404e-06,
      "loss": 0.4842,
      "step": 421
    },
    {
      "epoch": 0.8559837728194726,
      "grad_norm": 0.32218671439585006,
      "learning_rate": 8.990295305261823e-06,
      "loss": 0.4881,
      "step": 422
    },
    {
      "epoch": 0.8580121703853956,
      "grad_norm": 0.3850325896005668,
      "learning_rate": 8.98317278228618e-06,
      "loss": 0.4777,
      "step": 423
    },
    {
      "epoch": 0.8600405679513184,
      "grad_norm": 0.34858981639658393,
      "learning_rate": 8.976028068529997e-06,
      "loss": 0.4715,
      "step": 424
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.31712733487245864,
      "learning_rate": 8.96886120379741e-06,
      "loss": 0.4685,
      "step": 425
    },
    {
      "epoch": 0.8640973630831643,
      "grad_norm": 0.4522740409961723,
      "learning_rate": 8.96167222801597e-06,
      "loss": 0.4809,
      "step": 426
    },
    {
      "epoch": 0.8661257606490872,
      "grad_norm": 0.3828415174500192,
      "learning_rate": 8.954461181236406e-06,
      "loss": 0.476,
      "step": 427
    },
    {
      "epoch": 0.8681541582150102,
      "grad_norm": 0.4168321870968123,
      "learning_rate": 8.947228103632414e-06,
      "loss": 0.4928,
      "step": 428
    },
    {
      "epoch": 0.8701825557809331,
      "grad_norm": 0.3582939914060879,
      "learning_rate": 8.939973035500418e-06,
      "loss": 0.4988,
      "step": 429
    },
    {
      "epoch": 0.8722109533468559,
      "grad_norm": 0.40455973347260266,
      "learning_rate": 8.932696017259363e-06,
      "loss": 0.4729,
      "step": 430
    },
    {
      "epoch": 0.8742393509127789,
      "grad_norm": 0.34859935094072886,
      "learning_rate": 8.925397089450473e-06,
      "loss": 0.4777,
      "step": 431
    },
    {
      "epoch": 0.8762677484787018,
      "grad_norm": 0.38364247132756707,
      "learning_rate": 8.91807629273704e-06,
      "loss": 0.4673,
      "step": 432
    },
    {
      "epoch": 0.8782961460446247,
      "grad_norm": 0.3498446407762224,
      "learning_rate": 8.910733667904186e-06,
      "loss": 0.4845,
      "step": 433
    },
    {
      "epoch": 0.8803245436105477,
      "grad_norm": 0.33470764499096894,
      "learning_rate": 8.90336925585864e-06,
      "loss": 0.4535,
      "step": 434
    },
    {
      "epoch": 0.8823529411764706,
      "grad_norm": 0.3415458535579387,
      "learning_rate": 8.895983097628515e-06,
      "loss": 0.4594,
      "step": 435
    },
    {
      "epoch": 0.8843813387423936,
      "grad_norm": 0.355119732317347,
      "learning_rate": 8.888575234363071e-06,
      "loss": 0.4762,
      "step": 436
    },
    {
      "epoch": 0.8864097363083164,
      "grad_norm": 0.3311037642326978,
      "learning_rate": 8.88114570733249e-06,
      "loss": 0.4635,
      "step": 437
    },
    {
      "epoch": 0.8884381338742393,
      "grad_norm": 0.4014942099662035,
      "learning_rate": 8.873694557927648e-06,
      "loss": 0.486,
      "step": 438
    },
    {
      "epoch": 0.8904665314401623,
      "grad_norm": 0.37840489593667037,
      "learning_rate": 8.866221827659876e-06,
      "loss": 0.4803,
      "step": 439
    },
    {
      "epoch": 0.8924949290060852,
      "grad_norm": 0.3665259203928634,
      "learning_rate": 8.858727558160744e-06,
      "loss": 0.477,
      "step": 440
    },
    {
      "epoch": 0.8945233265720081,
      "grad_norm": 0.3570557420013508,
      "learning_rate": 8.851211791181813e-06,
      "loss": 0.4775,
      "step": 441
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.4478440070600395,
      "learning_rate": 8.843674568594412e-06,
      "loss": 0.4719,
      "step": 442
    },
    {
      "epoch": 0.8985801217038539,
      "grad_norm": 0.31071843081014094,
      "learning_rate": 8.8361159323894e-06,
      "loss": 0.4676,
      "step": 443
    },
    {
      "epoch": 0.9006085192697769,
      "grad_norm": 0.3495828731583443,
      "learning_rate": 8.828535924676938e-06,
      "loss": 0.4875,
      "step": 444
    },
    {
      "epoch": 0.9026369168356998,
      "grad_norm": 0.3806421380009587,
      "learning_rate": 8.820934587686247e-06,
      "loss": 0.4919,
      "step": 445
    },
    {
      "epoch": 0.9046653144016227,
      "grad_norm": 0.31451220765636634,
      "learning_rate": 8.813311963765376e-06,
      "loss": 0.4785,
      "step": 446
    },
    {
      "epoch": 0.9066937119675457,
      "grad_norm": 0.34193700437742514,
      "learning_rate": 8.805668095380969e-06,
      "loss": 0.4678,
      "step": 447
    },
    {
      "epoch": 0.9087221095334685,
      "grad_norm": 0.33938175201773485,
      "learning_rate": 8.798003025118021e-06,
      "loss": 0.4622,
      "step": 448
    },
    {
      "epoch": 0.9107505070993914,
      "grad_norm": 0.3832530706451466,
      "learning_rate": 8.790316795679654e-06,
      "loss": 0.4826,
      "step": 449
    },
    {
      "epoch": 0.9127789046653144,
      "grad_norm": 0.36994417434256494,
      "learning_rate": 8.782609449886862e-06,
      "loss": 0.475,
      "step": 450
    },
    {
      "epoch": 0.9148073022312373,
      "grad_norm": 0.3205983243101067,
      "learning_rate": 8.774881030678284e-06,
      "loss": 0.4501,
      "step": 451
    },
    {
      "epoch": 0.9168356997971603,
      "grad_norm": 0.3265508117603187,
      "learning_rate": 8.767131581109961e-06,
      "loss": 0.4699,
      "step": 452
    },
    {
      "epoch": 0.9188640973630832,
      "grad_norm": 0.3005428948585711,
      "learning_rate": 8.759361144355103e-06,
      "loss": 0.481,
      "step": 453
    },
    {
      "epoch": 0.920892494929006,
      "grad_norm": 0.35143392505181914,
      "learning_rate": 8.751569763703833e-06,
      "loss": 0.4863,
      "step": 454
    },
    {
      "epoch": 0.922920892494929,
      "grad_norm": 0.33454543804716796,
      "learning_rate": 8.74375748256296e-06,
      "loss": 0.466,
      "step": 455
    },
    {
      "epoch": 0.9249492900608519,
      "grad_norm": 0.34088858798273186,
      "learning_rate": 8.735924344455732e-06,
      "loss": 0.4928,
      "step": 456
    },
    {
      "epoch": 0.9269776876267748,
      "grad_norm": 0.3886123483687258,
      "learning_rate": 8.728070393021595e-06,
      "loss": 0.4712,
      "step": 457
    },
    {
      "epoch": 0.9290060851926978,
      "grad_norm": 0.27230558421191287,
      "learning_rate": 8.720195672015945e-06,
      "loss": 0.4712,
      "step": 458
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.3848515104149214,
      "learning_rate": 8.712300225309894e-06,
      "loss": 0.4739,
      "step": 459
    },
    {
      "epoch": 0.9330628803245437,
      "grad_norm": 0.34163172197690894,
      "learning_rate": 8.704384096890014e-06,
      "loss": 0.4757,
      "step": 460
    },
    {
      "epoch": 0.9350912778904665,
      "grad_norm": 0.2876902662299937,
      "learning_rate": 8.6964473308581e-06,
      "loss": 0.4762,
      "step": 461
    },
    {
      "epoch": 0.9371196754563894,
      "grad_norm": 0.3258445484230377,
      "learning_rate": 8.688489971430923e-06,
      "loss": 0.4672,
      "step": 462
    },
    {
      "epoch": 0.9391480730223124,
      "grad_norm": 0.3426820773646926,
      "learning_rate": 8.680512062939976e-06,
      "loss": 0.4863,
      "step": 463
    },
    {
      "epoch": 0.9411764705882353,
      "grad_norm": 0.35732396160649327,
      "learning_rate": 8.672513649831247e-06,
      "loss": 0.4587,
      "step": 464
    },
    {
      "epoch": 0.9432048681541582,
      "grad_norm": 0.36337673055134967,
      "learning_rate": 8.664494776664942e-06,
      "loss": 0.4606,
      "step": 465
    },
    {
      "epoch": 0.9452332657200812,
      "grad_norm": 0.2850468979593698,
      "learning_rate": 8.656455488115265e-06,
      "loss": 0.4499,
      "step": 466
    },
    {
      "epoch": 0.947261663286004,
      "grad_norm": 0.3851091309961093,
      "learning_rate": 8.64839582897015e-06,
      "loss": 0.4742,
      "step": 467
    },
    {
      "epoch": 0.949290060851927,
      "grad_norm": 0.3412203427326395,
      "learning_rate": 8.640315844131021e-06,
      "loss": 0.4697,
      "step": 468
    },
    {
      "epoch": 0.9513184584178499,
      "grad_norm": 0.3556929714740312,
      "learning_rate": 8.63221557861254e-06,
      "loss": 0.4404,
      "step": 469
    },
    {
      "epoch": 0.9533468559837728,
      "grad_norm": 0.33410315410782027,
      "learning_rate": 8.62409507754235e-06,
      "loss": 0.4649,
      "step": 470
    },
    {
      "epoch": 0.9553752535496958,
      "grad_norm": 0.3533792474625136,
      "learning_rate": 8.615954386160836e-06,
      "loss": 0.468,
      "step": 471
    },
    {
      "epoch": 0.9574036511156186,
      "grad_norm": 0.32152926821234556,
      "learning_rate": 8.607793549820862e-06,
      "loss": 0.4695,
      "step": 472
    },
    {
      "epoch": 0.9594320486815415,
      "grad_norm": 0.3441389108706895,
      "learning_rate": 8.599612613987522e-06,
      "loss": 0.469,
      "step": 473
    },
    {
      "epoch": 0.9614604462474645,
      "grad_norm": 0.33248310528714725,
      "learning_rate": 8.591411624237887e-06,
      "loss": 0.4909,
      "step": 474
    },
    {
      "epoch": 0.9634888438133874,
      "grad_norm": 0.3553610627628786,
      "learning_rate": 8.583190626260754e-06,
      "loss": 0.4735,
      "step": 475
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.2993291706388498,
      "learning_rate": 8.574949665856385e-06,
      "loss": 0.469,
      "step": 476
    },
    {
      "epoch": 0.9675456389452333,
      "grad_norm": 0.3089402383436147,
      "learning_rate": 8.566688788936254e-06,
      "loss": 0.4522,
      "step": 477
    },
    {
      "epoch": 0.9695740365111561,
      "grad_norm": 0.2984491578428065,
      "learning_rate": 8.558408041522801e-06,
      "loss": 0.4825,
      "step": 478
    },
    {
      "epoch": 0.9716024340770791,
      "grad_norm": 0.34288526967062255,
      "learning_rate": 8.550107469749159e-06,
      "loss": 0.4934,
      "step": 479
    },
    {
      "epoch": 0.973630831643002,
      "grad_norm": 0.3194271467055375,
      "learning_rate": 8.541787119858904e-06,
      "loss": 0.4792,
      "step": 480
    },
    {
      "epoch": 0.9756592292089249,
      "grad_norm": 0.32810912334885517,
      "learning_rate": 8.533447038205805e-06,
      "loss": 0.4823,
      "step": 481
    },
    {
      "epoch": 0.9776876267748479,
      "grad_norm": 0.3281478390736348,
      "learning_rate": 8.525087271253553e-06,
      "loss": 0.473,
      "step": 482
    },
    {
      "epoch": 0.9797160243407708,
      "grad_norm": 0.35627260956184137,
      "learning_rate": 8.516707865575515e-06,
      "loss": 0.461,
      "step": 483
    },
    {
      "epoch": 0.9817444219066938,
      "grad_norm": 0.30588165834719905,
      "learning_rate": 8.508308867854465e-06,
      "loss": 0.4741,
      "step": 484
    },
    {
      "epoch": 0.9837728194726166,
      "grad_norm": 0.30964402103670674,
      "learning_rate": 8.499890324882323e-06,
      "loss": 0.4712,
      "step": 485
    },
    {
      "epoch": 0.9858012170385395,
      "grad_norm": 0.32532942439407236,
      "learning_rate": 8.491452283559904e-06,
      "loss": 0.472,
      "step": 486
    },
    {
      "epoch": 0.9878296146044625,
      "grad_norm": 0.33335789251278725,
      "learning_rate": 8.482994790896645e-06,
      "loss": 0.4778,
      "step": 487
    },
    {
      "epoch": 0.9898580121703854,
      "grad_norm": 0.33216113746100284,
      "learning_rate": 8.474517894010356e-06,
      "loss": 0.4657,
      "step": 488
    },
    {
      "epoch": 0.9918864097363083,
      "grad_norm": 0.3483450169096007,
      "learning_rate": 8.466021640126946e-06,
      "loss": 0.4796,
      "step": 489
    },
    {
      "epoch": 0.9939148073022313,
      "grad_norm": 0.3813974654439515,
      "learning_rate": 8.457506076580164e-06,
      "loss": 0.4894,
      "step": 490
    },
    {
      "epoch": 0.9959432048681541,
      "grad_norm": 0.34139252365627665,
      "learning_rate": 8.448971250811337e-06,
      "loss": 0.4583,
      "step": 491
    },
    {
      "epoch": 0.9979716024340771,
      "grad_norm": 0.3449667508968355,
      "learning_rate": 8.440417210369104e-06,
      "loss": 0.4627,
      "step": 492
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3163691831408595,
      "learning_rate": 8.431844002909153e-06,
      "loss": 0.4536,
      "step": 493
    },
    {
      "epoch": 1.002028397565923,
      "grad_norm": 0.3374714068816851,
      "learning_rate": 8.423251676193951e-06,
      "loss": 0.4587,
      "step": 494
    },
    {
      "epoch": 1.0040567951318458,
      "grad_norm": 0.3250592189134637,
      "learning_rate": 8.414640278092485e-06,
      "loss": 0.4595,
      "step": 495
    },
    {
      "epoch": 1.0060851926977687,
      "grad_norm": 0.32527112871108943,
      "learning_rate": 8.406009856579986e-06,
      "loss": 0.4559,
      "step": 496
    },
    {
      "epoch": 1.0081135902636917,
      "grad_norm": 0.31327427046584677,
      "learning_rate": 8.397360459737673e-06,
      "loss": 0.4551,
      "step": 497
    },
    {
      "epoch": 1.0101419878296145,
      "grad_norm": 0.30915690983728517,
      "learning_rate": 8.388692135752473e-06,
      "loss": 0.4398,
      "step": 498
    },
    {
      "epoch": 1.0121703853955375,
      "grad_norm": 0.36018821082513264,
      "learning_rate": 8.38000493291676e-06,
      "loss": 0.4493,
      "step": 499
    },
    {
      "epoch": 1.0141987829614605,
      "grad_norm": 0.34534524015531415,
      "learning_rate": 8.371298899628091e-06,
      "loss": 0.4549,
      "step": 500
    },
    {
      "epoch": 1.0162271805273835,
      "grad_norm": 0.3518668940859276,
      "learning_rate": 8.362574084388921e-06,
      "loss": 0.4794,
      "step": 501
    },
    {
      "epoch": 1.0182555780933062,
      "grad_norm": 0.3155483233174309,
      "learning_rate": 8.353830535806347e-06,
      "loss": 0.446,
      "step": 502
    },
    {
      "epoch": 1.0202839756592292,
      "grad_norm": 0.37123725714984285,
      "learning_rate": 8.34506830259183e-06,
      "loss": 0.4656,
      "step": 503
    },
    {
      "epoch": 1.0223123732251522,
      "grad_norm": 0.36756484013587537,
      "learning_rate": 8.336287433560925e-06,
      "loss": 0.4596,
      "step": 504
    },
    {
      "epoch": 1.024340770791075,
      "grad_norm": 0.3139683709939055,
      "learning_rate": 8.327487977633013e-06,
      "loss": 0.4517,
      "step": 505
    },
    {
      "epoch": 1.026369168356998,
      "grad_norm": 0.35381349826485553,
      "learning_rate": 8.318669983831022e-06,
      "loss": 0.4543,
      "step": 506
    },
    {
      "epoch": 1.028397565922921,
      "grad_norm": 0.3909537893523297,
      "learning_rate": 8.309833501281159e-06,
      "loss": 0.4406,
      "step": 507
    },
    {
      "epoch": 1.0304259634888437,
      "grad_norm": 0.3146805462626415,
      "learning_rate": 8.300978579212629e-06,
      "loss": 0.4565,
      "step": 508
    },
    {
      "epoch": 1.0324543610547667,
      "grad_norm": 0.3981701531448272,
      "learning_rate": 8.292105266957372e-06,
      "loss": 0.4378,
      "step": 509
    },
    {
      "epoch": 1.0344827586206897,
      "grad_norm": 0.32674561265809116,
      "learning_rate": 8.283213613949781e-06,
      "loss": 0.442,
      "step": 510
    },
    {
      "epoch": 1.0365111561866125,
      "grad_norm": 0.3259602201298059,
      "learning_rate": 8.274303669726427e-06,
      "loss": 0.4527,
      "step": 511
    },
    {
      "epoch": 1.0385395537525355,
      "grad_norm": 0.3612714009496788,
      "learning_rate": 8.265375483925781e-06,
      "loss": 0.4687,
      "step": 512
    },
    {
      "epoch": 1.0405679513184585,
      "grad_norm": 0.31948990182618214,
      "learning_rate": 8.256429106287944e-06,
      "loss": 0.4367,
      "step": 513
    },
    {
      "epoch": 1.0425963488843812,
      "grad_norm": 0.3239848513112578,
      "learning_rate": 8.247464586654367e-06,
      "loss": 0.4389,
      "step": 514
    },
    {
      "epoch": 1.0446247464503042,
      "grad_norm": 0.3496050489410677,
      "learning_rate": 8.238481974967567e-06,
      "loss": 0.4452,
      "step": 515
    },
    {
      "epoch": 1.0466531440162272,
      "grad_norm": 0.31787930417325605,
      "learning_rate": 8.229481321270862e-06,
      "loss": 0.4496,
      "step": 516
    },
    {
      "epoch": 1.04868154158215,
      "grad_norm": 0.3185049459350567,
      "learning_rate": 8.220462675708075e-06,
      "loss": 0.4501,
      "step": 517
    },
    {
      "epoch": 1.050709939148073,
      "grad_norm": 0.3170291181025711,
      "learning_rate": 8.211426088523273e-06,
      "loss": 0.4569,
      "step": 518
    },
    {
      "epoch": 1.052738336713996,
      "grad_norm": 0.3023867469741922,
      "learning_rate": 8.202371610060471e-06,
      "loss": 0.4536,
      "step": 519
    },
    {
      "epoch": 1.054766734279919,
      "grad_norm": 0.3123595365903278,
      "learning_rate": 8.193299290763364e-06,
      "loss": 0.4631,
      "step": 520
    },
    {
      "epoch": 1.0567951318458417,
      "grad_norm": 0.281525074269082,
      "learning_rate": 8.184209181175038e-06,
      "loss": 0.4326,
      "step": 521
    },
    {
      "epoch": 1.0588235294117647,
      "grad_norm": 0.34746961980595914,
      "learning_rate": 8.175101331937692e-06,
      "loss": 0.4593,
      "step": 522
    },
    {
      "epoch": 1.0608519269776877,
      "grad_norm": 0.2869130375359574,
      "learning_rate": 8.165975793792355e-06,
      "loss": 0.46,
      "step": 523
    },
    {
      "epoch": 1.0628803245436105,
      "grad_norm": 0.358952257452001,
      "learning_rate": 8.1568326175786e-06,
      "loss": 0.4422,
      "step": 524
    },
    {
      "epoch": 1.0649087221095335,
      "grad_norm": 0.29788983004888003,
      "learning_rate": 8.14767185423427e-06,
      "loss": 0.4624,
      "step": 525
    },
    {
      "epoch": 1.0669371196754565,
      "grad_norm": 0.31194750107377467,
      "learning_rate": 8.138493554795184e-06,
      "loss": 0.4634,
      "step": 526
    },
    {
      "epoch": 1.0689655172413792,
      "grad_norm": 0.32969446615721026,
      "learning_rate": 8.129297770394855e-06,
      "loss": 0.4554,
      "step": 527
    },
    {
      "epoch": 1.0709939148073022,
      "grad_norm": 0.35116798266330285,
      "learning_rate": 8.120084552264213e-06,
      "loss": 0.4228,
      "step": 528
    },
    {
      "epoch": 1.0730223123732252,
      "grad_norm": 0.2848873632191636,
      "learning_rate": 8.11085395173131e-06,
      "loss": 0.4394,
      "step": 529
    },
    {
      "epoch": 1.075050709939148,
      "grad_norm": 0.38112259419897654,
      "learning_rate": 8.101606020221038e-06,
      "loss": 0.4457,
      "step": 530
    },
    {
      "epoch": 1.077079107505071,
      "grad_norm": 0.3917101035493545,
      "learning_rate": 8.092340809254844e-06,
      "loss": 0.4462,
      "step": 531
    },
    {
      "epoch": 1.079107505070994,
      "grad_norm": 0.3092994672414441,
      "learning_rate": 8.083058370450436e-06,
      "loss": 0.4496,
      "step": 532
    },
    {
      "epoch": 1.081135902636917,
      "grad_norm": 0.38304748198641775,
      "learning_rate": 8.073758755521506e-06,
      "loss": 0.451,
      "step": 533
    },
    {
      "epoch": 1.0831643002028397,
      "grad_norm": 0.33530738429072043,
      "learning_rate": 8.064442016277434e-06,
      "loss": 0.4664,
      "step": 534
    },
    {
      "epoch": 1.0851926977687627,
      "grad_norm": 0.333819128640187,
      "learning_rate": 8.055108204623001e-06,
      "loss": 0.4351,
      "step": 535
    },
    {
      "epoch": 1.0872210953346857,
      "grad_norm": 0.3233319365075351,
      "learning_rate": 8.045757372558103e-06,
      "loss": 0.4383,
      "step": 536
    },
    {
      "epoch": 1.0892494929006085,
      "grad_norm": 0.3552704540388316,
      "learning_rate": 8.03638957217746e-06,
      "loss": 0.4592,
      "step": 537
    },
    {
      "epoch": 1.0912778904665315,
      "grad_norm": 0.3554799322026267,
      "learning_rate": 8.02700485567032e-06,
      "loss": 0.4459,
      "step": 538
    },
    {
      "epoch": 1.0933062880324544,
      "grad_norm": 0.37465504216528767,
      "learning_rate": 8.017603275320176e-06,
      "loss": 0.4568,
      "step": 539
    },
    {
      "epoch": 1.0953346855983772,
      "grad_norm": 0.2998465511135317,
      "learning_rate": 8.008184883504473e-06,
      "loss": 0.4447,
      "step": 540
    },
    {
      "epoch": 1.0973630831643002,
      "grad_norm": 0.4357422307333169,
      "learning_rate": 7.998749732694308e-06,
      "loss": 0.4431,
      "step": 541
    },
    {
      "epoch": 1.0993914807302232,
      "grad_norm": 0.3084521629360656,
      "learning_rate": 7.989297875454157e-06,
      "loss": 0.442,
      "step": 542
    },
    {
      "epoch": 1.101419878296146,
      "grad_norm": 0.3732372897137307,
      "learning_rate": 7.979829364441555e-06,
      "loss": 0.443,
      "step": 543
    },
    {
      "epoch": 1.103448275862069,
      "grad_norm": 0.32540563961557306,
      "learning_rate": 7.970344252406832e-06,
      "loss": 0.4641,
      "step": 544
    },
    {
      "epoch": 1.105476673427992,
      "grad_norm": 0.3004948943445507,
      "learning_rate": 7.960842592192792e-06,
      "loss": 0.443,
      "step": 545
    },
    {
      "epoch": 1.1075050709939147,
      "grad_norm": 0.3710842815404189,
      "learning_rate": 7.951324436734438e-06,
      "loss": 0.4482,
      "step": 546
    },
    {
      "epoch": 1.1095334685598377,
      "grad_norm": 0.32396918172412376,
      "learning_rate": 7.94178983905867e-06,
      "loss": 0.4515,
      "step": 547
    },
    {
      "epoch": 1.1115618661257607,
      "grad_norm": 0.34728241362862566,
      "learning_rate": 7.932238852283985e-06,
      "loss": 0.4495,
      "step": 548
    },
    {
      "epoch": 1.1135902636916835,
      "grad_norm": 0.326933592058978,
      "learning_rate": 7.922671529620192e-06,
      "loss": 0.464,
      "step": 549
    },
    {
      "epoch": 1.1156186612576064,
      "grad_norm": 0.406622662105528,
      "learning_rate": 7.913087924368104e-06,
      "loss": 0.4703,
      "step": 550
    },
    {
      "epoch": 1.1176470588235294,
      "grad_norm": 0.3074237535761807,
      "learning_rate": 7.903488089919253e-06,
      "loss": 0.4449,
      "step": 551
    },
    {
      "epoch": 1.1196754563894524,
      "grad_norm": 0.3197869099067984,
      "learning_rate": 7.893872079755578e-06,
      "loss": 0.4194,
      "step": 552
    },
    {
      "epoch": 1.1217038539553752,
      "grad_norm": 0.3223925468496244,
      "learning_rate": 7.88423994744914e-06,
      "loss": 0.4231,
      "step": 553
    },
    {
      "epoch": 1.1237322515212982,
      "grad_norm": 0.35127370725996887,
      "learning_rate": 7.874591746661818e-06,
      "loss": 0.4621,
      "step": 554
    },
    {
      "epoch": 1.1257606490872212,
      "grad_norm": 0.3395130262578899,
      "learning_rate": 7.864927531145012e-06,
      "loss": 0.456,
      "step": 555
    },
    {
      "epoch": 1.127789046653144,
      "grad_norm": 0.33504742412133054,
      "learning_rate": 7.855247354739337e-06,
      "loss": 0.4588,
      "step": 556
    },
    {
      "epoch": 1.129817444219067,
      "grad_norm": 0.3323680978061306,
      "learning_rate": 7.845551271374333e-06,
      "loss": 0.4507,
      "step": 557
    },
    {
      "epoch": 1.13184584178499,
      "grad_norm": 0.30404253156187977,
      "learning_rate": 7.835839335068159e-06,
      "loss": 0.4387,
      "step": 558
    },
    {
      "epoch": 1.1338742393509127,
      "grad_norm": 0.31637280060502415,
      "learning_rate": 7.82611159992729e-06,
      "loss": 0.4478,
      "step": 559
    },
    {
      "epoch": 1.1359026369168357,
      "grad_norm": 0.30989254570237046,
      "learning_rate": 7.816368120146226e-06,
      "loss": 0.4546,
      "step": 560
    },
    {
      "epoch": 1.1379310344827587,
      "grad_norm": 0.3029273156671296,
      "learning_rate": 7.80660895000717e-06,
      "loss": 0.4466,
      "step": 561
    },
    {
      "epoch": 1.1399594320486814,
      "grad_norm": 0.2971037070501276,
      "learning_rate": 7.79683414387975e-06,
      "loss": 0.4491,
      "step": 562
    },
    {
      "epoch": 1.1419878296146044,
      "grad_norm": 0.25976168028958047,
      "learning_rate": 7.787043756220698e-06,
      "loss": 0.4437,
      "step": 563
    },
    {
      "epoch": 1.1440162271805274,
      "grad_norm": 0.341826130206304,
      "learning_rate": 7.77723784157355e-06,
      "loss": 0.4603,
      "step": 564
    },
    {
      "epoch": 1.1460446247464504,
      "grad_norm": 0.3051005807595728,
      "learning_rate": 7.767416454568358e-06,
      "loss": 0.4699,
      "step": 565
    },
    {
      "epoch": 1.1480730223123732,
      "grad_norm": 0.3045171740451653,
      "learning_rate": 7.757579649921354e-06,
      "loss": 0.4463,
      "step": 566
    },
    {
      "epoch": 1.1501014198782962,
      "grad_norm": 0.34694655699110927,
      "learning_rate": 7.747727482434679e-06,
      "loss": 0.4476,
      "step": 567
    },
    {
      "epoch": 1.1521298174442192,
      "grad_norm": 0.3085655531186579,
      "learning_rate": 7.737860006996053e-06,
      "loss": 0.4576,
      "step": 568
    },
    {
      "epoch": 1.154158215010142,
      "grad_norm": 0.283246910873189,
      "learning_rate": 7.727977278578484e-06,
      "loss": 0.446,
      "step": 569
    },
    {
      "epoch": 1.156186612576065,
      "grad_norm": 0.3394381025130293,
      "learning_rate": 7.718079352239956e-06,
      "loss": 0.4634,
      "step": 570
    },
    {
      "epoch": 1.158215010141988,
      "grad_norm": 0.2618839038367694,
      "learning_rate": 7.708166283123118e-06,
      "loss": 0.429,
      "step": 571
    },
    {
      "epoch": 1.1602434077079107,
      "grad_norm": 0.4065969274081818,
      "learning_rate": 7.698238126454986e-06,
      "loss": 0.4455,
      "step": 572
    },
    {
      "epoch": 1.1622718052738337,
      "grad_norm": 0.299578217568346,
      "learning_rate": 7.68829493754663e-06,
      "loss": 0.442,
      "step": 573
    },
    {
      "epoch": 1.1643002028397567,
      "grad_norm": 0.3573561790149975,
      "learning_rate": 7.678336771792866e-06,
      "loss": 0.4372,
      "step": 574
    },
    {
      "epoch": 1.1663286004056794,
      "grad_norm": 0.3129281081939506,
      "learning_rate": 7.668363684671947e-06,
      "loss": 0.4469,
      "step": 575
    },
    {
      "epoch": 1.1683569979716024,
      "grad_norm": 0.32266427209694387,
      "learning_rate": 7.658375731745256e-06,
      "loss": 0.4462,
      "step": 576
    },
    {
      "epoch": 1.1703853955375254,
      "grad_norm": 0.3275392501333595,
      "learning_rate": 7.648372968656995e-06,
      "loss": 0.4533,
      "step": 577
    },
    {
      "epoch": 1.1724137931034484,
      "grad_norm": 0.3403315010600426,
      "learning_rate": 7.638355451133874e-06,
      "loss": 0.4672,
      "step": 578
    },
    {
      "epoch": 1.1744421906693712,
      "grad_norm": 0.3044315086748403,
      "learning_rate": 7.628323234984806e-06,
      "loss": 0.4282,
      "step": 579
    },
    {
      "epoch": 1.1764705882352942,
      "grad_norm": 0.31863206725632853,
      "learning_rate": 7.618276376100589e-06,
      "loss": 0.4353,
      "step": 580
    },
    {
      "epoch": 1.178498985801217,
      "grad_norm": 0.262899824867327,
      "learning_rate": 7.608214930453597e-06,
      "loss": 0.4415,
      "step": 581
    },
    {
      "epoch": 1.18052738336714,
      "grad_norm": 0.29383303651908455,
      "learning_rate": 7.5981389540974696e-06,
      "loss": 0.4677,
      "step": 582
    },
    {
      "epoch": 1.182555780933063,
      "grad_norm": 0.3024285950945986,
      "learning_rate": 7.588048503166801e-06,
      "loss": 0.4631,
      "step": 583
    },
    {
      "epoch": 1.184584178498986,
      "grad_norm": 0.29499671912837255,
      "learning_rate": 7.577943633876821e-06,
      "loss": 0.4284,
      "step": 584
    },
    {
      "epoch": 1.1866125760649087,
      "grad_norm": 0.31901263173416944,
      "learning_rate": 7.5678244025230894e-06,
      "loss": 0.4588,
      "step": 585
    },
    {
      "epoch": 1.1886409736308317,
      "grad_norm": 0.31169109026743486,
      "learning_rate": 7.557690865481177e-06,
      "loss": 0.4705,
      "step": 586
    },
    {
      "epoch": 1.1906693711967546,
      "grad_norm": 0.3549340053818905,
      "learning_rate": 7.547543079206355e-06,
      "loss": 0.4591,
      "step": 587
    },
    {
      "epoch": 1.1926977687626774,
      "grad_norm": 0.308293542902282,
      "learning_rate": 7.5373811002332785e-06,
      "loss": 0.4511,
      "step": 588
    },
    {
      "epoch": 1.1947261663286004,
      "grad_norm": 0.30525177831547823,
      "learning_rate": 7.5272049851756716e-06,
      "loss": 0.4616,
      "step": 589
    },
    {
      "epoch": 1.1967545638945234,
      "grad_norm": 0.2899630178629302,
      "learning_rate": 7.517014790726011e-06,
      "loss": 0.4656,
      "step": 590
    },
    {
      "epoch": 1.1987829614604462,
      "grad_norm": 0.29465370442173316,
      "learning_rate": 7.506810573655215e-06,
      "loss": 0.4376,
      "step": 591
    },
    {
      "epoch": 1.2008113590263692,
      "grad_norm": 0.2731355426465377,
      "learning_rate": 7.496592390812319e-06,
      "loss": 0.4568,
      "step": 592
    },
    {
      "epoch": 1.2028397565922921,
      "grad_norm": 0.2798003290876117,
      "learning_rate": 7.486360299124169e-06,
      "loss": 0.4327,
      "step": 593
    },
    {
      "epoch": 1.204868154158215,
      "grad_norm": 0.3252288854962471,
      "learning_rate": 7.476114355595096e-06,
      "loss": 0.4543,
      "step": 594
    },
    {
      "epoch": 1.206896551724138,
      "grad_norm": 0.2913169269982524,
      "learning_rate": 7.4658546173066005e-06,
      "loss": 0.4409,
      "step": 595
    },
    {
      "epoch": 1.208924949290061,
      "grad_norm": 0.30817710724212716,
      "learning_rate": 7.455581141417038e-06,
      "loss": 0.4457,
      "step": 596
    },
    {
      "epoch": 1.2109533468559839,
      "grad_norm": 0.2987818496451199,
      "learning_rate": 7.445293985161296e-06,
      "loss": 0.4428,
      "step": 597
    },
    {
      "epoch": 1.2129817444219066,
      "grad_norm": 0.29310604940939566,
      "learning_rate": 7.4349932058504785e-06,
      "loss": 0.4614,
      "step": 598
    },
    {
      "epoch": 1.2150101419878296,
      "grad_norm": 0.2814641108862498,
      "learning_rate": 7.424678860871584e-06,
      "loss": 0.4501,
      "step": 599
    },
    {
      "epoch": 1.2170385395537526,
      "grad_norm": 0.3442220935773512,
      "learning_rate": 7.414351007687188e-06,
      "loss": 0.4506,
      "step": 600
    },
    {
      "epoch": 1.2190669371196754,
      "grad_norm": 0.2646007947426903,
      "learning_rate": 7.404009703835121e-06,
      "loss": 0.4329,
      "step": 601
    },
    {
      "epoch": 1.2210953346855984,
      "grad_norm": 0.3257940854833134,
      "learning_rate": 7.393655006928151e-06,
      "loss": 0.4549,
      "step": 602
    },
    {
      "epoch": 1.2231237322515214,
      "grad_norm": 0.2627071250064789,
      "learning_rate": 7.383286974653659e-06,
      "loss": 0.4405,
      "step": 603
    },
    {
      "epoch": 1.2251521298174441,
      "grad_norm": 0.3201649725142573,
      "learning_rate": 7.37290566477332e-06,
      "loss": 0.4562,
      "step": 604
    },
    {
      "epoch": 1.2271805273833671,
      "grad_norm": 0.28400858173877647,
      "learning_rate": 7.362511135122779e-06,
      "loss": 0.4477,
      "step": 605
    },
    {
      "epoch": 1.2292089249492901,
      "grad_norm": 0.29417555376842314,
      "learning_rate": 7.352103443611327e-06,
      "loss": 0.4469,
      "step": 606
    },
    {
      "epoch": 1.231237322515213,
      "grad_norm": 0.4021039660569394,
      "learning_rate": 7.341682648221591e-06,
      "loss": 0.4501,
      "step": 607
    },
    {
      "epoch": 1.2332657200811359,
      "grad_norm": 0.32537504640560644,
      "learning_rate": 7.33124880700919e-06,
      "loss": 0.4539,
      "step": 608
    },
    {
      "epoch": 1.2352941176470589,
      "grad_norm": 0.4717850402936866,
      "learning_rate": 7.320801978102434e-06,
      "loss": 0.4611,
      "step": 609
    },
    {
      "epoch": 1.2373225152129819,
      "grad_norm": 0.30164174239294717,
      "learning_rate": 7.310342219701981e-06,
      "loss": 0.4779,
      "step": 610
    },
    {
      "epoch": 1.2393509127789046,
      "grad_norm": 0.3857187841932404,
      "learning_rate": 7.299869590080524e-06,
      "loss": 0.4419,
      "step": 611
    },
    {
      "epoch": 1.2413793103448276,
      "grad_norm": 0.32479667510211785,
      "learning_rate": 7.289384147582461e-06,
      "loss": 0.4572,
      "step": 612
    },
    {
      "epoch": 1.2434077079107504,
      "grad_norm": 0.3259346153605337,
      "learning_rate": 7.278885950623578e-06,
      "loss": 0.4596,
      "step": 613
    },
    {
      "epoch": 1.2454361054766734,
      "grad_norm": 0.3823156460771375,
      "learning_rate": 7.268375057690711e-06,
      "loss": 0.4562,
      "step": 614
    },
    {
      "epoch": 1.2474645030425964,
      "grad_norm": 0.30939941828665596,
      "learning_rate": 7.257851527341429e-06,
      "loss": 0.4357,
      "step": 615
    },
    {
      "epoch": 1.2494929006085194,
      "grad_norm": 0.3617024081745379,
      "learning_rate": 7.247315418203709e-06,
      "loss": 0.4587,
      "step": 616
    },
    {
      "epoch": 1.2515212981744421,
      "grad_norm": 0.3338980041717798,
      "learning_rate": 7.236766788975603e-06,
      "loss": 0.4597,
      "step": 617
    },
    {
      "epoch": 1.2535496957403651,
      "grad_norm": 0.3185068995078411,
      "learning_rate": 7.2262056984249154e-06,
      "loss": 0.4387,
      "step": 618
    },
    {
      "epoch": 1.2555780933062881,
      "grad_norm": 0.3669233588991798,
      "learning_rate": 7.215632205388872e-06,
      "loss": 0.4585,
      "step": 619
    },
    {
      "epoch": 1.2576064908722109,
      "grad_norm": 0.2748092716409909,
      "learning_rate": 7.205046368773796e-06,
      "loss": 0.4522,
      "step": 620
    },
    {
      "epoch": 1.2596348884381339,
      "grad_norm": 0.35688440932489,
      "learning_rate": 7.19444824755478e-06,
      "loss": 0.4432,
      "step": 621
    },
    {
      "epoch": 1.2616632860040569,
      "grad_norm": 0.29488322149004803,
      "learning_rate": 7.183837900775352e-06,
      "loss": 0.4447,
      "step": 622
    },
    {
      "epoch": 1.2636916835699799,
      "grad_norm": 0.3120899810955614,
      "learning_rate": 7.173215387547155e-06,
      "loss": 0.4407,
      "step": 623
    },
    {
      "epoch": 1.2657200811359026,
      "grad_norm": 0.3033949336444941,
      "learning_rate": 7.162580767049607e-06,
      "loss": 0.447,
      "step": 624
    },
    {
      "epoch": 1.2677484787018256,
      "grad_norm": 0.29180434673042976,
      "learning_rate": 7.151934098529583e-06,
      "loss": 0.4337,
      "step": 625
    },
    {
      "epoch": 1.2697768762677484,
      "grad_norm": 0.30049194319484307,
      "learning_rate": 7.141275441301077e-06,
      "loss": 0.4476,
      "step": 626
    },
    {
      "epoch": 1.2718052738336714,
      "grad_norm": 0.26538986528671504,
      "learning_rate": 7.130604854744871e-06,
      "loss": 0.464,
      "step": 627
    },
    {
      "epoch": 1.2738336713995944,
      "grad_norm": 0.26105894220621617,
      "learning_rate": 7.11992239830821e-06,
      "loss": 0.456,
      "step": 628
    },
    {
      "epoch": 1.2758620689655173,
      "grad_norm": 0.3355119564761078,
      "learning_rate": 7.109228131504465e-06,
      "loss": 0.4455,
      "step": 629
    },
    {
      "epoch": 1.2778904665314401,
      "grad_norm": 0.2573400222701869,
      "learning_rate": 7.098522113912809e-06,
      "loss": 0.4357,
      "step": 630
    },
    {
      "epoch": 1.279918864097363,
      "grad_norm": 0.31995026014531863,
      "learning_rate": 7.087804405177876e-06,
      "loss": 0.4753,
      "step": 631
    },
    {
      "epoch": 1.2819472616632859,
      "grad_norm": 0.30599673134235267,
      "learning_rate": 7.0770750650094335e-06,
      "loss": 0.4407,
      "step": 632
    },
    {
      "epoch": 1.2839756592292089,
      "grad_norm": 0.27803007674614694,
      "learning_rate": 7.066334153182049e-06,
      "loss": 0.4464,
      "step": 633
    },
    {
      "epoch": 1.2860040567951319,
      "grad_norm": 0.3373170432477396,
      "learning_rate": 7.055581729534758e-06,
      "loss": 0.4603,
      "step": 634
    },
    {
      "epoch": 1.2880324543610548,
      "grad_norm": 0.32116609214155517,
      "learning_rate": 7.044817853970732e-06,
      "loss": 0.4462,
      "step": 635
    },
    {
      "epoch": 1.2900608519269776,
      "grad_norm": 0.3390135562883033,
      "learning_rate": 7.034042586456939e-06,
      "loss": 0.4286,
      "step": 636
    },
    {
      "epoch": 1.2920892494929006,
      "grad_norm": 0.3243579778477668,
      "learning_rate": 7.023255987023813e-06,
      "loss": 0.4353,
      "step": 637
    },
    {
      "epoch": 1.2941176470588236,
      "grad_norm": 0.30116939509917295,
      "learning_rate": 7.012458115764923e-06,
      "loss": 0.4419,
      "step": 638
    },
    {
      "epoch": 1.2961460446247464,
      "grad_norm": 0.3377132933838073,
      "learning_rate": 7.001649032836631e-06,
      "loss": 0.4466,
      "step": 639
    },
    {
      "epoch": 1.2981744421906694,
      "grad_norm": 0.306529069396835,
      "learning_rate": 6.990828798457765e-06,
      "loss": 0.4616,
      "step": 640
    },
    {
      "epoch": 1.3002028397565923,
      "grad_norm": 0.3551033749201116,
      "learning_rate": 6.9799974729092765e-06,
      "loss": 0.4402,
      "step": 641
    },
    {
      "epoch": 1.3022312373225153,
      "grad_norm": 0.34015277187927573,
      "learning_rate": 6.969155116533908e-06,
      "loss": 0.478,
      "step": 642
    },
    {
      "epoch": 1.304259634888438,
      "grad_norm": 0.36142069872396976,
      "learning_rate": 6.958301789735853e-06,
      "loss": 0.4509,
      "step": 643
    },
    {
      "epoch": 1.306288032454361,
      "grad_norm": 0.31477741434795214,
      "learning_rate": 6.947437552980426e-06,
      "loss": 0.455,
      "step": 644
    },
    {
      "epoch": 1.3083164300202839,
      "grad_norm": 0.3163093777055954,
      "learning_rate": 6.936562466793724e-06,
      "loss": 0.4406,
      "step": 645
    },
    {
      "epoch": 1.3103448275862069,
      "grad_norm": 0.30521159490653677,
      "learning_rate": 6.9256765917622855e-06,
      "loss": 0.4472,
      "step": 646
    },
    {
      "epoch": 1.3123732251521298,
      "grad_norm": 0.27382084756569774,
      "learning_rate": 6.914779988532755e-06,
      "loss": 0.4498,
      "step": 647
    },
    {
      "epoch": 1.3144016227180528,
      "grad_norm": 0.30259932512765997,
      "learning_rate": 6.903872717811542e-06,
      "loss": 0.4425,
      "step": 648
    },
    {
      "epoch": 1.3164300202839756,
      "grad_norm": 0.2662904085650847,
      "learning_rate": 6.892954840364493e-06,
      "loss": 0.4462,
      "step": 649
    },
    {
      "epoch": 1.3184584178498986,
      "grad_norm": 0.29720291426561124,
      "learning_rate": 6.882026417016542e-06,
      "loss": 0.4578,
      "step": 650
    },
    {
      "epoch": 1.3204868154158216,
      "grad_norm": 0.28774365212730685,
      "learning_rate": 6.871087508651373e-06,
      "loss": 0.4363,
      "step": 651
    },
    {
      "epoch": 1.3225152129817443,
      "grad_norm": 0.2658072680370139,
      "learning_rate": 6.860138176211088e-06,
      "loss": 0.4493,
      "step": 652
    },
    {
      "epoch": 1.3245436105476673,
      "grad_norm": 0.3021642089228888,
      "learning_rate": 6.8491784806958616e-06,
      "loss": 0.449,
      "step": 653
    },
    {
      "epoch": 1.3265720081135903,
      "grad_norm": 0.2782748435222328,
      "learning_rate": 6.838208483163601e-06,
      "loss": 0.4623,
      "step": 654
    },
    {
      "epoch": 1.3286004056795133,
      "grad_norm": 0.2944132098575465,
      "learning_rate": 6.827228244729609e-06,
      "loss": 0.4469,
      "step": 655
    },
    {
      "epoch": 1.330628803245436,
      "grad_norm": 0.3090969086512804,
      "learning_rate": 6.816237826566241e-06,
      "loss": 0.4427,
      "step": 656
    },
    {
      "epoch": 1.332657200811359,
      "grad_norm": 0.28131335830059506,
      "learning_rate": 6.805237289902565e-06,
      "loss": 0.4235,
      "step": 657
    },
    {
      "epoch": 1.3346855983772818,
      "grad_norm": 0.3245614980440333,
      "learning_rate": 6.79422669602402e-06,
      "loss": 0.4363,
      "step": 658
    },
    {
      "epoch": 1.3367139959432048,
      "grad_norm": 0.31088659057190304,
      "learning_rate": 6.783206106272076e-06,
      "loss": 0.4712,
      "step": 659
    },
    {
      "epoch": 1.3387423935091278,
      "grad_norm": 0.2870272417894291,
      "learning_rate": 6.772175582043891e-06,
      "loss": 0.442,
      "step": 660
    },
    {
      "epoch": 1.3407707910750508,
      "grad_norm": 0.32938830213356474,
      "learning_rate": 6.761135184791969e-06,
      "loss": 0.4662,
      "step": 661
    },
    {
      "epoch": 1.3427991886409736,
      "grad_norm": 0.2935744723818878,
      "learning_rate": 6.750084976023821e-06,
      "loss": 0.4361,
      "step": 662
    },
    {
      "epoch": 1.3448275862068966,
      "grad_norm": 0.3298917454802601,
      "learning_rate": 6.7390250173016104e-06,
      "loss": 0.4411,
      "step": 663
    },
    {
      "epoch": 1.3468559837728193,
      "grad_norm": 0.3257305561663728,
      "learning_rate": 6.7279553702418325e-06,
      "loss": 0.4452,
      "step": 664
    },
    {
      "epoch": 1.3488843813387423,
      "grad_norm": 0.3401945432096944,
      "learning_rate": 6.716876096514944e-06,
      "loss": 0.4566,
      "step": 665
    },
    {
      "epoch": 1.3509127789046653,
      "grad_norm": 0.2919061950722634,
      "learning_rate": 6.705787257845041e-06,
      "loss": 0.4467,
      "step": 666
    },
    {
      "epoch": 1.3529411764705883,
      "grad_norm": 0.3317068607311084,
      "learning_rate": 6.694688916009505e-06,
      "loss": 0.4529,
      "step": 667
    },
    {
      "epoch": 1.354969574036511,
      "grad_norm": 0.280737647347209,
      "learning_rate": 6.683581132838661e-06,
      "loss": 0.4307,
      "step": 668
    },
    {
      "epoch": 1.356997971602434,
      "grad_norm": 0.32349883129595725,
      "learning_rate": 6.672463970215436e-06,
      "loss": 0.4451,
      "step": 669
    },
    {
      "epoch": 1.359026369168357,
      "grad_norm": 0.28838030563148137,
      "learning_rate": 6.661337490075004e-06,
      "loss": 0.4528,
      "step": 670
    },
    {
      "epoch": 1.3610547667342798,
      "grad_norm": 0.3208611198091076,
      "learning_rate": 6.650201754404455e-06,
      "loss": 0.4575,
      "step": 671
    },
    {
      "epoch": 1.3630831643002028,
      "grad_norm": 0.2788542763523886,
      "learning_rate": 6.639056825242438e-06,
      "loss": 0.4561,
      "step": 672
    },
    {
      "epoch": 1.3651115618661258,
      "grad_norm": 0.3688786012761057,
      "learning_rate": 6.627902764678824e-06,
      "loss": 0.4427,
      "step": 673
    },
    {
      "epoch": 1.3671399594320488,
      "grad_norm": 0.2678439397363886,
      "learning_rate": 6.616739634854352e-06,
      "loss": 0.4384,
      "step": 674
    },
    {
      "epoch": 1.3691683569979716,
      "grad_norm": 0.353987227298648,
      "learning_rate": 6.605567497960295e-06,
      "loss": 0.4467,
      "step": 675
    },
    {
      "epoch": 1.3711967545638946,
      "grad_norm": 0.27944119318234467,
      "learning_rate": 6.594386416238095e-06,
      "loss": 0.4534,
      "step": 676
    },
    {
      "epoch": 1.3732251521298173,
      "grad_norm": 0.29709386574547847,
      "learning_rate": 6.583196451979031e-06,
      "loss": 0.4317,
      "step": 677
    },
    {
      "epoch": 1.3752535496957403,
      "grad_norm": 0.3079011258264504,
      "learning_rate": 6.571997667523873e-06,
      "loss": 0.4298,
      "step": 678
    },
    {
      "epoch": 1.3772819472616633,
      "grad_norm": 0.31852886134327507,
      "learning_rate": 6.560790125262524e-06,
      "loss": 0.4427,
      "step": 679
    },
    {
      "epoch": 1.3793103448275863,
      "grad_norm": 0.27517137030158306,
      "learning_rate": 6.549573887633677e-06,
      "loss": 0.4387,
      "step": 680
    },
    {
      "epoch": 1.381338742393509,
      "grad_norm": 0.2876297008210803,
      "learning_rate": 6.538349017124472e-06,
      "loss": 0.4442,
      "step": 681
    },
    {
      "epoch": 1.383367139959432,
      "grad_norm": 0.30210847068567515,
      "learning_rate": 6.52711557627014e-06,
      "loss": 0.4618,
      "step": 682
    },
    {
      "epoch": 1.385395537525355,
      "grad_norm": 0.3216589440455159,
      "learning_rate": 6.515873627653663e-06,
      "loss": 0.4298,
      "step": 683
    },
    {
      "epoch": 1.3874239350912778,
      "grad_norm": 0.29173893946731644,
      "learning_rate": 6.504623233905418e-06,
      "loss": 0.4297,
      "step": 684
    },
    {
      "epoch": 1.3894523326572008,
      "grad_norm": 0.3438928210475039,
      "learning_rate": 6.493364457702831e-06,
      "loss": 0.4437,
      "step": 685
    },
    {
      "epoch": 1.3914807302231238,
      "grad_norm": 0.29010466786846073,
      "learning_rate": 6.482097361770027e-06,
      "loss": 0.4474,
      "step": 686
    },
    {
      "epoch": 1.3935091277890468,
      "grad_norm": 0.2787123577361506,
      "learning_rate": 6.470822008877482e-06,
      "loss": 0.4443,
      "step": 687
    },
    {
      "epoch": 1.3955375253549696,
      "grad_norm": 0.3480451845685342,
      "learning_rate": 6.4595384618416765e-06,
      "loss": 0.4484,
      "step": 688
    },
    {
      "epoch": 1.3975659229208925,
      "grad_norm": 0.2585590391511514,
      "learning_rate": 6.448246783524734e-06,
      "loss": 0.4405,
      "step": 689
    },
    {
      "epoch": 1.3995943204868153,
      "grad_norm": 0.38264985784992245,
      "learning_rate": 6.436947036834087e-06,
      "loss": 0.4316,
      "step": 690
    },
    {
      "epoch": 1.4016227180527383,
      "grad_norm": 0.274630695433327,
      "learning_rate": 6.42563928472211e-06,
      "loss": 0.4318,
      "step": 691
    },
    {
      "epoch": 1.4036511156186613,
      "grad_norm": 0.28135739274384874,
      "learning_rate": 6.4143235901857805e-06,
      "loss": 0.4599,
      "step": 692
    },
    {
      "epoch": 1.4056795131845843,
      "grad_norm": 0.3126200191603351,
      "learning_rate": 6.403000016266326e-06,
      "loss": 0.4709,
      "step": 693
    },
    {
      "epoch": 1.407707910750507,
      "grad_norm": 0.25336263560137373,
      "learning_rate": 6.39166862604887e-06,
      "loss": 0.4349,
      "step": 694
    },
    {
      "epoch": 1.40973630831643,
      "grad_norm": 0.33531566569300314,
      "learning_rate": 6.380329482662078e-06,
      "loss": 0.46,
      "step": 695
    },
    {
      "epoch": 1.4117647058823528,
      "grad_norm": 0.2819146030526314,
      "learning_rate": 6.368982649277813e-06,
      "loss": 0.4302,
      "step": 696
    },
    {
      "epoch": 1.4137931034482758,
      "grad_norm": 0.31005820062900513,
      "learning_rate": 6.35762818911078e-06,
      "loss": 0.4396,
      "step": 697
    },
    {
      "epoch": 1.4158215010141988,
      "grad_norm": 0.2978461105418995,
      "learning_rate": 6.346266165418173e-06,
      "loss": 0.456,
      "step": 698
    },
    {
      "epoch": 1.4178498985801218,
      "grad_norm": 0.3681222971302691,
      "learning_rate": 6.334896641499324e-06,
      "loss": 0.4575,
      "step": 699
    },
    {
      "epoch": 1.4198782961460445,
      "grad_norm": 0.30325038905168517,
      "learning_rate": 6.323519680695349e-06,
      "loss": 0.4405,
      "step": 700
    },
    {
      "epoch": 1.4219066937119675,
      "grad_norm": 0.3555248676876023,
      "learning_rate": 6.312135346388793e-06,
      "loss": 0.4594,
      "step": 701
    },
    {
      "epoch": 1.4239350912778905,
      "grad_norm": 0.3036483692685623,
      "learning_rate": 6.300743702003289e-06,
      "loss": 0.4298,
      "step": 702
    },
    {
      "epoch": 1.4259634888438133,
      "grad_norm": 0.3571564658430927,
      "learning_rate": 6.289344811003184e-06,
      "loss": 0.4613,
      "step": 703
    },
    {
      "epoch": 1.4279918864097363,
      "grad_norm": 0.3032584366561266,
      "learning_rate": 6.277938736893204e-06,
      "loss": 0.4521,
      "step": 704
    },
    {
      "epoch": 1.4300202839756593,
      "grad_norm": 0.3699337261578937,
      "learning_rate": 6.2665255432180916e-06,
      "loss": 0.4496,
      "step": 705
    },
    {
      "epoch": 1.4320486815415823,
      "grad_norm": 0.2848952674388909,
      "learning_rate": 6.255105293562252e-06,
      "loss": 0.4562,
      "step": 706
    },
    {
      "epoch": 1.434077079107505,
      "grad_norm": 0.3331006771983738,
      "learning_rate": 6.2436780515494035e-06,
      "loss": 0.4432,
      "step": 707
    },
    {
      "epoch": 1.436105476673428,
      "grad_norm": 0.39890585676797424,
      "learning_rate": 6.232243880842215e-06,
      "loss": 0.4478,
      "step": 708
    },
    {
      "epoch": 1.4381338742393508,
      "grad_norm": 0.34023018914290565,
      "learning_rate": 6.2208028451419575e-06,
      "loss": 0.4471,
      "step": 709
    },
    {
      "epoch": 1.4401622718052738,
      "grad_norm": 0.41278822462161036,
      "learning_rate": 6.209355008188153e-06,
      "loss": 0.4401,
      "step": 710
    },
    {
      "epoch": 1.4421906693711968,
      "grad_norm": 0.38780312492585706,
      "learning_rate": 6.197900433758205e-06,
      "loss": 0.4262,
      "step": 711
    },
    {
      "epoch": 1.4442190669371198,
      "grad_norm": 0.36490380582820076,
      "learning_rate": 6.1864391856670595e-06,
      "loss": 0.4379,
      "step": 712
    },
    {
      "epoch": 1.4462474645030425,
      "grad_norm": 0.30687350229800453,
      "learning_rate": 6.174971327766842e-06,
      "loss": 0.4512,
      "step": 713
    },
    {
      "epoch": 1.4482758620689655,
      "grad_norm": 0.41417789533708355,
      "learning_rate": 6.1634969239464995e-06,
      "loss": 0.458,
      "step": 714
    },
    {
      "epoch": 1.4503042596348885,
      "grad_norm": 0.2961628946007071,
      "learning_rate": 6.1520160381314465e-06,
      "loss": 0.4433,
      "step": 715
    },
    {
      "epoch": 1.4523326572008113,
      "grad_norm": 0.3547441080939827,
      "learning_rate": 6.14052873428321e-06,
      "loss": 0.4402,
      "step": 716
    },
    {
      "epoch": 1.4543610547667343,
      "grad_norm": 0.2750410382955491,
      "learning_rate": 6.129035076399077e-06,
      "loss": 0.4336,
      "step": 717
    },
    {
      "epoch": 1.4563894523326573,
      "grad_norm": 0.30582658130544055,
      "learning_rate": 6.117535128511729e-06,
      "loss": 0.4533,
      "step": 718
    },
    {
      "epoch": 1.4584178498985803,
      "grad_norm": 0.2754806207796925,
      "learning_rate": 6.106028954688892e-06,
      "loss": 0.4387,
      "step": 719
    },
    {
      "epoch": 1.460446247464503,
      "grad_norm": 0.31919666990646595,
      "learning_rate": 6.094516619032975e-06,
      "loss": 0.4624,
      "step": 720
    },
    {
      "epoch": 1.462474645030426,
      "grad_norm": 0.294033152308826,
      "learning_rate": 6.082998185680718e-06,
      "loss": 0.455,
      "step": 721
    },
    {
      "epoch": 1.4645030425963488,
      "grad_norm": 0.3148075268834361,
      "learning_rate": 6.071473718802833e-06,
      "loss": 0.4467,
      "step": 722
    },
    {
      "epoch": 1.4665314401622718,
      "grad_norm": 0.29063693080789504,
      "learning_rate": 6.059943282603642e-06,
      "loss": 0.4411,
      "step": 723
    },
    {
      "epoch": 1.4685598377281948,
      "grad_norm": 0.29231684570728167,
      "learning_rate": 6.048406941320726e-06,
      "loss": 0.4328,
      "step": 724
    },
    {
      "epoch": 1.4705882352941178,
      "grad_norm": 0.34435562598892094,
      "learning_rate": 6.03686475922456e-06,
      "loss": 0.4579,
      "step": 725
    },
    {
      "epoch": 1.4726166328600405,
      "grad_norm": 0.266897749396002,
      "learning_rate": 6.025316800618165e-06,
      "loss": 0.4422,
      "step": 726
    },
    {
      "epoch": 1.4746450304259635,
      "grad_norm": 0.31104282300820635,
      "learning_rate": 6.013763129836739e-06,
      "loss": 0.436,
      "step": 727
    },
    {
      "epoch": 1.4766734279918863,
      "grad_norm": 0.30767782587257003,
      "learning_rate": 6.002203811247306e-06,
      "loss": 0.4537,
      "step": 728
    },
    {
      "epoch": 1.4787018255578093,
      "grad_norm": 0.27922998767059637,
      "learning_rate": 5.990638909248352e-06,
      "loss": 0.4414,
      "step": 729
    },
    {
      "epoch": 1.4807302231237323,
      "grad_norm": 0.25681265675396625,
      "learning_rate": 5.979068488269468e-06,
      "loss": 0.4368,
      "step": 730
    },
    {
      "epoch": 1.4827586206896552,
      "grad_norm": 0.3195826798211906,
      "learning_rate": 5.967492612770999e-06,
      "loss": 0.435,
      "step": 731
    },
    {
      "epoch": 1.484787018255578,
      "grad_norm": 0.24852126873778013,
      "learning_rate": 5.95591134724367e-06,
      "loss": 0.4456,
      "step": 732
    },
    {
      "epoch": 1.486815415821501,
      "grad_norm": 0.29966062327221693,
      "learning_rate": 5.944324756208238e-06,
      "loss": 0.4388,
      "step": 733
    },
    {
      "epoch": 1.488843813387424,
      "grad_norm": 0.26072809112244677,
      "learning_rate": 5.93273290421513e-06,
      "loss": 0.4437,
      "step": 734
    },
    {
      "epoch": 1.4908722109533468,
      "grad_norm": 0.2840414742246271,
      "learning_rate": 5.92113585584408e-06,
      "loss": 0.4434,
      "step": 735
    },
    {
      "epoch": 1.4929006085192698,
      "grad_norm": 0.25261617163283784,
      "learning_rate": 5.909533675703774e-06,
      "loss": 0.4665,
      "step": 736
    },
    {
      "epoch": 1.4949290060851927,
      "grad_norm": 0.28374297064092696,
      "learning_rate": 5.897926428431485e-06,
      "loss": 0.4577,
      "step": 737
    },
    {
      "epoch": 1.4969574036511157,
      "grad_norm": 0.2702077163191035,
      "learning_rate": 5.88631417869272e-06,
      "loss": 0.4372,
      "step": 738
    },
    {
      "epoch": 1.4989858012170385,
      "grad_norm": 0.2561044064395688,
      "learning_rate": 5.87469699118085e-06,
      "loss": 0.4565,
      "step": 739
    },
    {
      "epoch": 1.5010141987829615,
      "grad_norm": 0.2585256950234163,
      "learning_rate": 5.863074930616757e-06,
      "loss": 0.4546,
      "step": 740
    },
    {
      "epoch": 1.5030425963488843,
      "grad_norm": 0.24529602462170857,
      "learning_rate": 5.851448061748477e-06,
      "loss": 0.4393,
      "step": 741
    },
    {
      "epoch": 1.5050709939148073,
      "grad_norm": 0.2853660166151909,
      "learning_rate": 5.839816449350824e-06,
      "loss": 0.4377,
      "step": 742
    },
    {
      "epoch": 1.5070993914807302,
      "grad_norm": 0.26313362857462974,
      "learning_rate": 5.828180158225047e-06,
      "loss": 0.4658,
      "step": 743
    },
    {
      "epoch": 1.5091277890466532,
      "grad_norm": 0.2671683826506484,
      "learning_rate": 5.8165392531984565e-06,
      "loss": 0.4485,
      "step": 744
    },
    {
      "epoch": 1.5111561866125762,
      "grad_norm": 0.3303189512425596,
      "learning_rate": 5.804893799124068e-06,
      "loss": 0.4511,
      "step": 745
    },
    {
      "epoch": 1.513184584178499,
      "grad_norm": 0.2694234262561527,
      "learning_rate": 5.79324386088024e-06,
      "loss": 0.4324,
      "step": 746
    },
    {
      "epoch": 1.5152129817444218,
      "grad_norm": 0.27222862159477407,
      "learning_rate": 5.7815895033703164e-06,
      "loss": 0.4509,
      "step": 747
    },
    {
      "epoch": 1.5172413793103448,
      "grad_norm": 0.28515752654227866,
      "learning_rate": 5.769930791522256e-06,
      "loss": 0.43,
      "step": 748
    },
    {
      "epoch": 1.5192697768762677,
      "grad_norm": 0.2556648741554994,
      "learning_rate": 5.758267790288282e-06,
      "loss": 0.4385,
      "step": 749
    },
    {
      "epoch": 1.5212981744421907,
      "grad_norm": 0.2965906239794784,
      "learning_rate": 5.74660056464451e-06,
      "loss": 0.4571,
      "step": 750
    },
    {
      "epoch": 1.5233265720081137,
      "grad_norm": 0.2465869524327756,
      "learning_rate": 5.734929179590593e-06,
      "loss": 0.432,
      "step": 751
    },
    {
      "epoch": 1.5253549695740365,
      "grad_norm": 0.27433203466402767,
      "learning_rate": 5.723253700149354e-06,
      "loss": 0.4596,
      "step": 752
    },
    {
      "epoch": 1.5273833671399595,
      "grad_norm": 0.2865264611300629,
      "learning_rate": 5.711574191366427e-06,
      "loss": 0.4413,
      "step": 753
    },
    {
      "epoch": 1.5294117647058822,
      "grad_norm": 0.26737961718581055,
      "learning_rate": 5.699890718309896e-06,
      "loss": 0.4408,
      "step": 754
    },
    {
      "epoch": 1.5314401622718052,
      "grad_norm": 0.25026964263272367,
      "learning_rate": 5.6882033460699294e-06,
      "loss": 0.4381,
      "step": 755
    },
    {
      "epoch": 1.5334685598377282,
      "grad_norm": 0.2710806006102463,
      "learning_rate": 5.676512139758419e-06,
      "loss": 0.45,
      "step": 756
    },
    {
      "epoch": 1.5354969574036512,
      "grad_norm": 0.27246374958564185,
      "learning_rate": 5.664817164508614e-06,
      "loss": 0.4466,
      "step": 757
    },
    {
      "epoch": 1.537525354969574,
      "grad_norm": 0.2752523999615091,
      "learning_rate": 5.653118485474765e-06,
      "loss": 0.4535,
      "step": 758
    },
    {
      "epoch": 1.539553752535497,
      "grad_norm": 0.3043761368258922,
      "learning_rate": 5.641416167831752e-06,
      "loss": 0.4615,
      "step": 759
    },
    {
      "epoch": 1.5415821501014197,
      "grad_norm": 0.27084656931596335,
      "learning_rate": 5.629710276774734e-06,
      "loss": 0.4377,
      "step": 760
    },
    {
      "epoch": 1.5436105476673427,
      "grad_norm": 0.28988435855658273,
      "learning_rate": 5.618000877518767e-06,
      "loss": 0.4464,
      "step": 761
    },
    {
      "epoch": 1.5456389452332657,
      "grad_norm": 0.2841043179712608,
      "learning_rate": 5.606288035298461e-06,
      "loss": 0.4489,
      "step": 762
    },
    {
      "epoch": 1.5476673427991887,
      "grad_norm": 0.2719539410396961,
      "learning_rate": 5.594571815367602e-06,
      "loss": 0.4509,
      "step": 763
    },
    {
      "epoch": 1.5496957403651117,
      "grad_norm": 0.27658656503377627,
      "learning_rate": 5.5828522829987965e-06,
      "loss": 0.4426,
      "step": 764
    },
    {
      "epoch": 1.5517241379310345,
      "grad_norm": 0.27554007669007075,
      "learning_rate": 5.5711295034831034e-06,
      "loss": 0.4404,
      "step": 765
    },
    {
      "epoch": 1.5537525354969572,
      "grad_norm": 0.27251111872599143,
      "learning_rate": 5.559403542129674e-06,
      "loss": 0.4366,
      "step": 766
    },
    {
      "epoch": 1.5557809330628802,
      "grad_norm": 0.31198222605818093,
      "learning_rate": 5.547674464265384e-06,
      "loss": 0.4552,
      "step": 767
    },
    {
      "epoch": 1.5578093306288032,
      "grad_norm": 0.28516438852818937,
      "learning_rate": 5.535942335234469e-06,
      "loss": 0.4488,
      "step": 768
    },
    {
      "epoch": 1.5598377281947262,
      "grad_norm": 0.3132304372313493,
      "learning_rate": 5.524207220398169e-06,
      "loss": 0.4455,
      "step": 769
    },
    {
      "epoch": 1.5618661257606492,
      "grad_norm": 0.2833204517007794,
      "learning_rate": 5.512469185134355e-06,
      "loss": 0.4334,
      "step": 770
    },
    {
      "epoch": 1.563894523326572,
      "grad_norm": 0.25624803725195067,
      "learning_rate": 5.500728294837168e-06,
      "loss": 0.4406,
      "step": 771
    },
    {
      "epoch": 1.565922920892495,
      "grad_norm": 0.29510863770362217,
      "learning_rate": 5.488984614916656e-06,
      "loss": 0.422,
      "step": 772
    },
    {
      "epoch": 1.5679513184584177,
      "grad_norm": 0.26970885991988036,
      "learning_rate": 5.477238210798406e-06,
      "loss": 0.4491,
      "step": 773
    },
    {
      "epoch": 1.5699797160243407,
      "grad_norm": 0.2860599780404054,
      "learning_rate": 5.4654891479231864e-06,
      "loss": 0.4416,
      "step": 774
    },
    {
      "epoch": 1.5720081135902637,
      "grad_norm": 0.2756465251353537,
      "learning_rate": 5.453737491746572e-06,
      "loss": 0.4585,
      "step": 775
    },
    {
      "epoch": 1.5740365111561867,
      "grad_norm": 0.2856877794235753,
      "learning_rate": 5.441983307738591e-06,
      "loss": 0.4407,
      "step": 776
    },
    {
      "epoch": 1.5760649087221097,
      "grad_norm": 0.2569178370835297,
      "learning_rate": 5.430226661383348e-06,
      "loss": 0.4296,
      "step": 777
    },
    {
      "epoch": 1.5780933062880325,
      "grad_norm": 0.288447308766631,
      "learning_rate": 5.418467618178671e-06,
      "loss": 0.4376,
      "step": 778
    },
    {
      "epoch": 1.5801217038539552,
      "grad_norm": 0.2709425407165248,
      "learning_rate": 5.406706243635742e-06,
      "loss": 0.4467,
      "step": 779
    },
    {
      "epoch": 1.5821501014198782,
      "grad_norm": 0.24991678567241207,
      "learning_rate": 5.394942603278727e-06,
      "loss": 0.4466,
      "step": 780
    },
    {
      "epoch": 1.5841784989858012,
      "grad_norm": 0.25896904168275636,
      "learning_rate": 5.383176762644416e-06,
      "loss": 0.4341,
      "step": 781
    },
    {
      "epoch": 1.5862068965517242,
      "grad_norm": 0.2744896865298867,
      "learning_rate": 5.3714087872818576e-06,
      "loss": 0.4355,
      "step": 782
    },
    {
      "epoch": 1.5882352941176472,
      "grad_norm": 0.24450566769809756,
      "learning_rate": 5.359638742751994e-06,
      "loss": 0.4351,
      "step": 783
    },
    {
      "epoch": 1.59026369168357,
      "grad_norm": 0.24680120441932368,
      "learning_rate": 5.347866694627296e-06,
      "loss": 0.4276,
      "step": 784
    },
    {
      "epoch": 1.592292089249493,
      "grad_norm": 0.2722482280625572,
      "learning_rate": 5.3360927084913925e-06,
      "loss": 0.4533,
      "step": 785
    },
    {
      "epoch": 1.5943204868154157,
      "grad_norm": 0.27492058573321115,
      "learning_rate": 5.324316849938715e-06,
      "loss": 0.4368,
      "step": 786
    },
    {
      "epoch": 1.5963488843813387,
      "grad_norm": 0.2773406847094387,
      "learning_rate": 5.312539184574123e-06,
      "loss": 0.4637,
      "step": 787
    },
    {
      "epoch": 1.5983772819472617,
      "grad_norm": 0.2775939269798677,
      "learning_rate": 5.30075977801254e-06,
      "loss": 0.4573,
      "step": 788
    },
    {
      "epoch": 1.6004056795131847,
      "grad_norm": 0.27060433059641426,
      "learning_rate": 5.288978695878596e-06,
      "loss": 0.4591,
      "step": 789
    },
    {
      "epoch": 1.6024340770791075,
      "grad_norm": 0.25320105986819097,
      "learning_rate": 5.27719600380625e-06,
      "loss": 0.4603,
      "step": 790
    },
    {
      "epoch": 1.6044624746450304,
      "grad_norm": 0.28297584926216734,
      "learning_rate": 5.265411767438432e-06,
      "loss": 0.4381,
      "step": 791
    },
    {
      "epoch": 1.6064908722109532,
      "grad_norm": 0.25356691612626636,
      "learning_rate": 5.253626052426677e-06,
      "loss": 0.4177,
      "step": 792
    },
    {
      "epoch": 1.6085192697768762,
      "grad_norm": 0.30170484372872197,
      "learning_rate": 5.241838924430757e-06,
      "loss": 0.4547,
      "step": 793
    },
    {
      "epoch": 1.6105476673427992,
      "grad_norm": 0.29277679376831134,
      "learning_rate": 5.230050449118315e-06,
      "loss": 0.4394,
      "step": 794
    },
    {
      "epoch": 1.6125760649087222,
      "grad_norm": 0.2585275994393407,
      "learning_rate": 5.2182606921645e-06,
      "loss": 0.4513,
      "step": 795
    },
    {
      "epoch": 1.6146044624746452,
      "grad_norm": 0.2736629047148514,
      "learning_rate": 5.206469719251604e-06,
      "loss": 0.4485,
      "step": 796
    },
    {
      "epoch": 1.616632860040568,
      "grad_norm": 0.2848946414237589,
      "learning_rate": 5.194677596068689e-06,
      "loss": 0.4358,
      "step": 797
    },
    {
      "epoch": 1.6186612576064907,
      "grad_norm": 0.2748265190105223,
      "learning_rate": 5.182884388311229e-06,
      "loss": 0.447,
      "step": 798
    },
    {
      "epoch": 1.6206896551724137,
      "grad_norm": 0.28806841294996743,
      "learning_rate": 5.171090161680736e-06,
      "loss": 0.4586,
      "step": 799
    },
    {
      "epoch": 1.6227180527383367,
      "grad_norm": 0.2745126607470088,
      "learning_rate": 5.159294981884405e-06,
      "loss": 0.4442,
      "step": 800
    },
    {
      "epoch": 1.6247464503042597,
      "grad_norm": 0.28982683653411395,
      "learning_rate": 5.1474989146347355e-06,
      "loss": 0.4477,
      "step": 801
    },
    {
      "epoch": 1.6267748478701827,
      "grad_norm": 0.2609489001828053,
      "learning_rate": 5.135702025649171e-06,
      "loss": 0.4459,
      "step": 802
    },
    {
      "epoch": 1.6288032454361054,
      "grad_norm": 0.31472946610606484,
      "learning_rate": 5.1239043806497365e-06,
      "loss": 0.4447,
      "step": 803
    },
    {
      "epoch": 1.6308316430020284,
      "grad_norm": 0.2696195889089416,
      "learning_rate": 5.112106045362668e-06,
      "loss": 0.4407,
      "step": 804
    },
    {
      "epoch": 1.6328600405679512,
      "grad_norm": 0.29464653916099465,
      "learning_rate": 5.100307085518046e-06,
      "loss": 0.4343,
      "step": 805
    },
    {
      "epoch": 1.6348884381338742,
      "grad_norm": 0.2741378488638008,
      "learning_rate": 5.088507566849428e-06,
      "loss": 0.4611,
      "step": 806
    },
    {
      "epoch": 1.6369168356997972,
      "grad_norm": 0.27259859710492806,
      "learning_rate": 5.076707555093491e-06,
      "loss": 0.4683,
      "step": 807
    },
    {
      "epoch": 1.6389452332657202,
      "grad_norm": 0.27928679437769527,
      "learning_rate": 5.064907115989655e-06,
      "loss": 0.4303,
      "step": 808
    },
    {
      "epoch": 1.6409736308316432,
      "grad_norm": 0.26951926997615777,
      "learning_rate": 5.053106315279721e-06,
      "loss": 0.44,
      "step": 809
    },
    {
      "epoch": 1.643002028397566,
      "grad_norm": 0.25380677400708185,
      "learning_rate": 5.041305218707506e-06,
      "loss": 0.4245,
      "step": 810
    },
    {
      "epoch": 1.6450304259634887,
      "grad_norm": 0.24297841831837266,
      "learning_rate": 5.029503892018472e-06,
      "loss": 0.4364,
      "step": 811
    },
    {
      "epoch": 1.6470588235294117,
      "grad_norm": 0.25753841002405314,
      "learning_rate": 5.017702400959369e-06,
      "loss": 0.4324,
      "step": 812
    },
    {
      "epoch": 1.6490872210953347,
      "grad_norm": 0.2583090367867325,
      "learning_rate": 5.005900811277856e-06,
      "loss": 0.4584,
      "step": 813
    },
    {
      "epoch": 1.6511156186612577,
      "grad_norm": 0.3016549177583762,
      "learning_rate": 4.994099188722147e-06,
      "loss": 0.4516,
      "step": 814
    },
    {
      "epoch": 1.6531440162271807,
      "grad_norm": 0.2600081771090172,
      "learning_rate": 4.982297599040633e-06,
      "loss": 0.447,
      "step": 815
    },
    {
      "epoch": 1.6551724137931034,
      "grad_norm": 0.30960752747802023,
      "learning_rate": 4.970496107981529e-06,
      "loss": 0.4639,
      "step": 816
    },
    {
      "epoch": 1.6572008113590264,
      "grad_norm": 0.3405393787396165,
      "learning_rate": 4.958694781292496e-06,
      "loss": 0.4805,
      "step": 817
    },
    {
      "epoch": 1.6592292089249492,
      "grad_norm": 0.2562150294912409,
      "learning_rate": 4.946893684720279e-06,
      "loss": 0.4655,
      "step": 818
    },
    {
      "epoch": 1.6612576064908722,
      "grad_norm": 0.2905843105834792,
      "learning_rate": 4.935092884010347e-06,
      "loss": 0.4281,
      "step": 819
    },
    {
      "epoch": 1.6632860040567952,
      "grad_norm": 0.26818847934808615,
      "learning_rate": 4.92329244490651e-06,
      "loss": 0.4414,
      "step": 820
    },
    {
      "epoch": 1.6653144016227182,
      "grad_norm": 0.2720367661413882,
      "learning_rate": 4.911492433150573e-06,
      "loss": 0.4327,
      "step": 821
    },
    {
      "epoch": 1.667342799188641,
      "grad_norm": 0.2595190202065188,
      "learning_rate": 4.899692914481956e-06,
      "loss": 0.4333,
      "step": 822
    },
    {
      "epoch": 1.669371196754564,
      "grad_norm": 0.26746973736737467,
      "learning_rate": 4.887893954637335e-06,
      "loss": 0.4418,
      "step": 823
    },
    {
      "epoch": 1.6713995943204867,
      "grad_norm": 0.23737777497338453,
      "learning_rate": 4.876095619350264e-06,
      "loss": 0.4431,
      "step": 824
    },
    {
      "epoch": 1.6734279918864097,
      "grad_norm": 0.2678011480695968,
      "learning_rate": 4.86429797435083e-06,
      "loss": 0.4538,
      "step": 825
    },
    {
      "epoch": 1.6754563894523327,
      "grad_norm": 0.24777399450065554,
      "learning_rate": 4.852501085365266e-06,
      "loss": 0.4431,
      "step": 826
    },
    {
      "epoch": 1.6774847870182557,
      "grad_norm": 0.2527329435025946,
      "learning_rate": 4.840705018115595e-06,
      "loss": 0.4245,
      "step": 827
    },
    {
      "epoch": 1.6795131845841786,
      "grad_norm": 0.26461284669955915,
      "learning_rate": 4.8289098383192645e-06,
      "loss": 0.4497,
      "step": 828
    },
    {
      "epoch": 1.6815415821501014,
      "grad_norm": 0.2712736821764884,
      "learning_rate": 4.8171156116887725e-06,
      "loss": 0.4615,
      "step": 829
    },
    {
      "epoch": 1.6835699797160242,
      "grad_norm": 0.25064411880874543,
      "learning_rate": 4.805322403931312e-06,
      "loss": 0.4406,
      "step": 830
    },
    {
      "epoch": 1.6855983772819472,
      "grad_norm": 0.24890966643470544,
      "learning_rate": 4.7935302807483965e-06,
      "loss": 0.441,
      "step": 831
    },
    {
      "epoch": 1.6876267748478702,
      "grad_norm": 0.2861310322476368,
      "learning_rate": 4.7817393078355e-06,
      "loss": 0.4357,
      "step": 832
    },
    {
      "epoch": 1.6896551724137931,
      "grad_norm": 0.48221881615755563,
      "learning_rate": 4.769949550881687e-06,
      "loss": 0.4576,
      "step": 833
    },
    {
      "epoch": 1.6916835699797161,
      "grad_norm": 0.27071032419039265,
      "learning_rate": 4.758161075569244e-06,
      "loss": 0.4404,
      "step": 834
    },
    {
      "epoch": 1.693711967545639,
      "grad_norm": 0.2818580374012473,
      "learning_rate": 4.746373947573325e-06,
      "loss": 0.4329,
      "step": 835
    },
    {
      "epoch": 1.695740365111562,
      "grad_norm": 0.2696542561017053,
      "learning_rate": 4.7345882325615685e-06,
      "loss": 0.4154,
      "step": 836
    },
    {
      "epoch": 1.6977687626774847,
      "grad_norm": 0.24818951240134995,
      "learning_rate": 4.722803996193753e-06,
      "loss": 0.4402,
      "step": 837
    },
    {
      "epoch": 1.6997971602434077,
      "grad_norm": 0.26206575598076126,
      "learning_rate": 4.711021304121406e-06,
      "loss": 0.4388,
      "step": 838
    },
    {
      "epoch": 1.7018255578093306,
      "grad_norm": 0.30102676626642033,
      "learning_rate": 4.699240221987461e-06,
      "loss": 0.4361,
      "step": 839
    },
    {
      "epoch": 1.7038539553752536,
      "grad_norm": 0.2622282292414772,
      "learning_rate": 4.687460815425878e-06,
      "loss": 0.4374,
      "step": 840
    },
    {
      "epoch": 1.7058823529411766,
      "grad_norm": 0.23774033659545574,
      "learning_rate": 4.6756831500612846e-06,
      "loss": 0.4405,
      "step": 841
    },
    {
      "epoch": 1.7079107505070994,
      "grad_norm": 0.28471322072963784,
      "learning_rate": 4.663907291508608e-06,
      "loss": 0.4449,
      "step": 842
    },
    {
      "epoch": 1.7099391480730222,
      "grad_norm": 0.3259106027804292,
      "learning_rate": 4.652133305372705e-06,
      "loss": 0.4548,
      "step": 843
    },
    {
      "epoch": 1.7119675456389452,
      "grad_norm": 0.2544042085594743,
      "learning_rate": 4.640361257248008e-06,
      "loss": 0.4573,
      "step": 844
    },
    {
      "epoch": 1.7139959432048681,
      "grad_norm": 0.2695637760683073,
      "learning_rate": 4.628591212718144e-06,
      "loss": 0.4516,
      "step": 845
    },
    {
      "epoch": 1.7160243407707911,
      "grad_norm": 0.3470155449266578,
      "learning_rate": 4.616823237355588e-06,
      "loss": 0.4472,
      "step": 846
    },
    {
      "epoch": 1.7180527383367141,
      "grad_norm": 0.28379974795560353,
      "learning_rate": 4.605057396721275e-06,
      "loss": 0.4434,
      "step": 847
    },
    {
      "epoch": 1.720081135902637,
      "grad_norm": 0.27801060608892547,
      "learning_rate": 4.593293756364258e-06,
      "loss": 0.4501,
      "step": 848
    },
    {
      "epoch": 1.7221095334685599,
      "grad_norm": 0.3305793852005204,
      "learning_rate": 4.58153238182133e-06,
      "loss": 0.4501,
      "step": 849
    },
    {
      "epoch": 1.7241379310344827,
      "grad_norm": 0.2864472267596549,
      "learning_rate": 4.569773338616653e-06,
      "loss": 0.452,
      "step": 850
    },
    {
      "epoch": 1.7261663286004056,
      "grad_norm": 0.30544779227666846,
      "learning_rate": 4.558016692261412e-06,
      "loss": 0.4288,
      "step": 851
    },
    {
      "epoch": 1.7281947261663286,
      "grad_norm": 0.2873549517350851,
      "learning_rate": 4.546262508253429e-06,
      "loss": 0.4536,
      "step": 852
    },
    {
      "epoch": 1.7302231237322516,
      "grad_norm": 0.26055781268258593,
      "learning_rate": 4.534510852076817e-06,
      "loss": 0.4405,
      "step": 853
    },
    {
      "epoch": 1.7322515212981744,
      "grad_norm": 0.29445000006529426,
      "learning_rate": 4.522761789201595e-06,
      "loss": 0.4549,
      "step": 854
    },
    {
      "epoch": 1.7342799188640974,
      "grad_norm": 0.28013012406796717,
      "learning_rate": 4.511015385083345e-06,
      "loss": 0.4413,
      "step": 855
    },
    {
      "epoch": 1.7363083164300201,
      "grad_norm": 0.27996861737353146,
      "learning_rate": 4.499271705162834e-06,
      "loss": 0.4249,
      "step": 856
    },
    {
      "epoch": 1.7383367139959431,
      "grad_norm": 0.27210797224463135,
      "learning_rate": 4.487530814865646e-06,
      "loss": 0.4338,
      "step": 857
    },
    {
      "epoch": 1.7403651115618661,
      "grad_norm": 0.285864721131408,
      "learning_rate": 4.475792779601834e-06,
      "loss": 0.4616,
      "step": 858
    },
    {
      "epoch": 1.7423935091277891,
      "grad_norm": 0.26256195917983793,
      "learning_rate": 4.464057664765532e-06,
      "loss": 0.4436,
      "step": 859
    },
    {
      "epoch": 1.744421906693712,
      "grad_norm": 0.2521164712542866,
      "learning_rate": 4.452325535734619e-06,
      "loss": 0.4576,
      "step": 860
    },
    {
      "epoch": 1.7464503042596349,
      "grad_norm": 0.23754944094239938,
      "learning_rate": 4.440596457870327e-06,
      "loss": 0.4338,
      "step": 861
    },
    {
      "epoch": 1.7484787018255576,
      "grad_norm": 0.2810107129475458,
      "learning_rate": 4.428870496516896e-06,
      "loss": 0.453,
      "step": 862
    },
    {
      "epoch": 1.7505070993914806,
      "grad_norm": 0.24826206322067965,
      "learning_rate": 4.417147717001205e-06,
      "loss": 0.4445,
      "step": 863
    },
    {
      "epoch": 1.7525354969574036,
      "grad_norm": 0.25090297694559904,
      "learning_rate": 4.405428184632399e-06,
      "loss": 0.4446,
      "step": 864
    },
    {
      "epoch": 1.7545638945233266,
      "grad_norm": 0.25750090625700384,
      "learning_rate": 4.393711964701541e-06,
      "loss": 0.4492,
      "step": 865
    },
    {
      "epoch": 1.7565922920892496,
      "grad_norm": 0.24856470784052367,
      "learning_rate": 4.381999122481234e-06,
      "loss": 0.4432,
      "step": 866
    },
    {
      "epoch": 1.7586206896551724,
      "grad_norm": 0.2474877792487339,
      "learning_rate": 4.37028972322527e-06,
      "loss": 0.4378,
      "step": 867
    },
    {
      "epoch": 1.7606490872210954,
      "grad_norm": 0.2560052133817747,
      "learning_rate": 4.3585838321682486e-06,
      "loss": 0.4281,
      "step": 868
    },
    {
      "epoch": 1.7626774847870181,
      "grad_norm": 0.24397983140040516,
      "learning_rate": 4.346881514525236e-06,
      "loss": 0.441,
      "step": 869
    },
    {
      "epoch": 1.7647058823529411,
      "grad_norm": 0.25401782351315266,
      "learning_rate": 4.335182835491387e-06,
      "loss": 0.447,
      "step": 870
    },
    {
      "epoch": 1.7667342799188641,
      "grad_norm": 0.25935922754449675,
      "learning_rate": 4.323487860241582e-06,
      "loss": 0.4345,
      "step": 871
    },
    {
      "epoch": 1.768762677484787,
      "grad_norm": 0.25882836573600515,
      "learning_rate": 4.311796653930072e-06,
      "loss": 0.4424,
      "step": 872
    },
    {
      "epoch": 1.77079107505071,
      "grad_norm": 0.2747971728608191,
      "learning_rate": 4.3001092816901055e-06,
      "loss": 0.4583,
      "step": 873
    },
    {
      "epoch": 1.7728194726166329,
      "grad_norm": 0.22259814995617197,
      "learning_rate": 4.2884258086335755e-06,
      "loss": 0.4285,
      "step": 874
    },
    {
      "epoch": 1.7748478701825556,
      "grad_norm": 0.22861890964769183,
      "learning_rate": 4.2767462998506485e-06,
      "loss": 0.4291,
      "step": 875
    },
    {
      "epoch": 1.7768762677484786,
      "grad_norm": 0.2653780078809505,
      "learning_rate": 4.2650708204094105e-06,
      "loss": 0.4486,
      "step": 876
    },
    {
      "epoch": 1.7789046653144016,
      "grad_norm": 0.25593056432376304,
      "learning_rate": 4.253399435355492e-06,
      "loss": 0.4518,
      "step": 877
    },
    {
      "epoch": 1.7809330628803246,
      "grad_norm": 0.25196106724936945,
      "learning_rate": 4.241732209711718e-06,
      "loss": 0.4465,
      "step": 878
    },
    {
      "epoch": 1.7829614604462476,
      "grad_norm": 0.2624752729334754,
      "learning_rate": 4.230069208477745e-06,
      "loss": 0.4428,
      "step": 879
    },
    {
      "epoch": 1.7849898580121704,
      "grad_norm": 0.25655653379432924,
      "learning_rate": 4.218410496629685e-06,
      "loss": 0.4386,
      "step": 880
    },
    {
      "epoch": 1.7870182555780934,
      "grad_norm": 0.23558117662377323,
      "learning_rate": 4.206756139119762e-06,
      "loss": 0.4399,
      "step": 881
    },
    {
      "epoch": 1.7890466531440161,
      "grad_norm": 0.26584028019181327,
      "learning_rate": 4.195106200875935e-06,
      "loss": 0.456,
      "step": 882
    },
    {
      "epoch": 1.791075050709939,
      "grad_norm": 0.23752016986752034,
      "learning_rate": 4.183460746801546e-06,
      "loss": 0.4419,
      "step": 883
    },
    {
      "epoch": 1.793103448275862,
      "grad_norm": 0.24556294242278556,
      "learning_rate": 4.171819841774954e-06,
      "loss": 0.4141,
      "step": 884
    },
    {
      "epoch": 1.795131845841785,
      "grad_norm": 0.2810345780029813,
      "learning_rate": 4.160183550649176e-06,
      "loss": 0.449,
      "step": 885
    },
    {
      "epoch": 1.7971602434077079,
      "grad_norm": 0.25432015666958635,
      "learning_rate": 4.148551938251525e-06,
      "loss": 0.4444,
      "step": 886
    },
    {
      "epoch": 1.7991886409736308,
      "grad_norm": 0.2930286757609919,
      "learning_rate": 4.136925069383243e-06,
      "loss": 0.4836,
      "step": 887
    },
    {
      "epoch": 1.8012170385395536,
      "grad_norm": 0.2365908436449825,
      "learning_rate": 4.125303008819153e-06,
      "loss": 0.4363,
      "step": 888
    },
    {
      "epoch": 1.8032454361054766,
      "grad_norm": 0.25874302178784825,
      "learning_rate": 4.113685821307282e-06,
      "loss": 0.4412,
      "step": 889
    },
    {
      "epoch": 1.8052738336713996,
      "grad_norm": 0.24003292606868973,
      "learning_rate": 4.102073571568517e-06,
      "loss": 0.4226,
      "step": 890
    },
    {
      "epoch": 1.8073022312373226,
      "grad_norm": 0.26798766045597644,
      "learning_rate": 4.090466324296228e-06,
      "loss": 0.431,
      "step": 891
    },
    {
      "epoch": 1.8093306288032456,
      "grad_norm": 0.28597188666259477,
      "learning_rate": 4.0788641441559205e-06,
      "loss": 0.4833,
      "step": 892
    },
    {
      "epoch": 1.8113590263691683,
      "grad_norm": 0.24100235244016638,
      "learning_rate": 4.067267095784871e-06,
      "loss": 0.4366,
      "step": 893
    },
    {
      "epoch": 1.8133874239350911,
      "grad_norm": 0.2569067244518368,
      "learning_rate": 4.0556752437917616e-06,
      "loss": 0.4319,
      "step": 894
    },
    {
      "epoch": 1.815415821501014,
      "grad_norm": 0.2388547979538123,
      "learning_rate": 4.044088652756332e-06,
      "loss": 0.4243,
      "step": 895
    },
    {
      "epoch": 1.817444219066937,
      "grad_norm": 0.27854571868908357,
      "learning_rate": 4.032507387229002e-06,
      "loss": 0.4383,
      "step": 896
    },
    {
      "epoch": 1.81947261663286,
      "grad_norm": 0.24748630103063796,
      "learning_rate": 4.020931511730533e-06,
      "loss": 0.456,
      "step": 897
    },
    {
      "epoch": 1.821501014198783,
      "grad_norm": 0.24957060243910706,
      "learning_rate": 4.00936109075165e-06,
      "loss": 0.443,
      "step": 898
    },
    {
      "epoch": 1.8235294117647058,
      "grad_norm": 0.29044430860152376,
      "learning_rate": 3.997796188752695e-06,
      "loss": 0.4441,
      "step": 899
    },
    {
      "epoch": 1.8255578093306288,
      "grad_norm": 0.2548648793414548,
      "learning_rate": 3.986236870163262e-06,
      "loss": 0.4602,
      "step": 900
    },
    {
      "epoch": 1.8275862068965516,
      "grad_norm": 0.25134502838895084,
      "learning_rate": 3.974683199381836e-06,
      "loss": 0.4397,
      "step": 901
    },
    {
      "epoch": 1.8296146044624746,
      "grad_norm": 0.22879029302459206,
      "learning_rate": 3.963135240775441e-06,
      "loss": 0.4408,
      "step": 902
    },
    {
      "epoch": 1.8316430020283976,
      "grad_norm": 0.25401513525252684,
      "learning_rate": 3.951593058679276e-06,
      "loss": 0.4346,
      "step": 903
    },
    {
      "epoch": 1.8336713995943206,
      "grad_norm": 0.2558941741854143,
      "learning_rate": 3.940056717396361e-06,
      "loss": 0.4342,
      "step": 904
    },
    {
      "epoch": 1.8356997971602436,
      "grad_norm": 0.24365337472443999,
      "learning_rate": 3.928526281197169e-06,
      "loss": 0.454,
      "step": 905
    },
    {
      "epoch": 1.8377281947261663,
      "grad_norm": 0.2891370324134481,
      "learning_rate": 3.917001814319282e-06,
      "loss": 0.4333,
      "step": 906
    },
    {
      "epoch": 1.839756592292089,
      "grad_norm": 0.2559993854442267,
      "learning_rate": 3.905483380967027e-06,
      "loss": 0.4441,
      "step": 907
    },
    {
      "epoch": 1.841784989858012,
      "grad_norm": 0.24946066213639467,
      "learning_rate": 3.893971045311109e-06,
      "loss": 0.4465,
      "step": 908
    },
    {
      "epoch": 1.843813387423935,
      "grad_norm": 0.2677097269203519,
      "learning_rate": 3.882464871488273e-06,
      "loss": 0.4318,
      "step": 909
    },
    {
      "epoch": 1.845841784989858,
      "grad_norm": 0.2565751319755398,
      "learning_rate": 3.870964923600924e-06,
      "loss": 0.411,
      "step": 910
    },
    {
      "epoch": 1.847870182555781,
      "grad_norm": 0.28799499875957,
      "learning_rate": 3.859471265716791e-06,
      "loss": 0.4377,
      "step": 911
    },
    {
      "epoch": 1.8498985801217038,
      "grad_norm": 0.26064877549583865,
      "learning_rate": 3.847983961868555e-06,
      "loss": 0.4681,
      "step": 912
    },
    {
      "epoch": 1.8519269776876268,
      "grad_norm": 0.2821858838796224,
      "learning_rate": 3.836503076053501e-06,
      "loss": 0.4219,
      "step": 913
    },
    {
      "epoch": 1.8539553752535496,
      "grad_norm": 0.2670465245393212,
      "learning_rate": 3.825028672233159e-06,
      "loss": 0.4345,
      "step": 914
    },
    {
      "epoch": 1.8559837728194726,
      "grad_norm": 0.28247032969750063,
      "learning_rate": 3.8135608143329404e-06,
      "loss": 0.4307,
      "step": 915
    },
    {
      "epoch": 1.8580121703853956,
      "grad_norm": 0.2604158396111888,
      "learning_rate": 3.802099566241797e-06,
      "loss": 0.4425,
      "step": 916
    },
    {
      "epoch": 1.8600405679513186,
      "grad_norm": 0.2709434756862799,
      "learning_rate": 3.7906449918118493e-06,
      "loss": 0.4432,
      "step": 917
    },
    {
      "epoch": 1.8620689655172413,
      "grad_norm": 0.2669007186309109,
      "learning_rate": 3.779197154858044e-06,
      "loss": 0.439,
      "step": 918
    },
    {
      "epoch": 1.8640973630831643,
      "grad_norm": 0.26276620363706,
      "learning_rate": 3.7677561191577873e-06,
      "loss": 0.4344,
      "step": 919
    },
    {
      "epoch": 1.866125760649087,
      "grad_norm": 0.2685064363461762,
      "learning_rate": 3.7563219484506e-06,
      "loss": 0.4388,
      "step": 920
    },
    {
      "epoch": 1.86815415821501,
      "grad_norm": 0.29471900328490375,
      "learning_rate": 3.7448947064377496e-06,
      "loss": 0.4382,
      "step": 921
    },
    {
      "epoch": 1.870182555780933,
      "grad_norm": 0.24097715120919105,
      "learning_rate": 3.733474456781908e-06,
      "loss": 0.4261,
      "step": 922
    },
    {
      "epoch": 1.872210953346856,
      "grad_norm": 0.2687061997128588,
      "learning_rate": 3.722061263106797e-06,
      "loss": 0.4243,
      "step": 923
    },
    {
      "epoch": 1.874239350912779,
      "grad_norm": 0.3029606288359608,
      "learning_rate": 3.7106551889968167e-06,
      "loss": 0.4414,
      "step": 924
    },
    {
      "epoch": 1.8762677484787018,
      "grad_norm": 0.2230365697770601,
      "learning_rate": 3.699256297996714e-06,
      "loss": 0.442,
      "step": 925
    },
    {
      "epoch": 1.8782961460446246,
      "grad_norm": 0.25351156293582766,
      "learning_rate": 3.6878646536112074e-06,
      "loss": 0.4298,
      "step": 926
    },
    {
      "epoch": 1.8803245436105476,
      "grad_norm": 0.27661342104752784,
      "learning_rate": 3.6764803193046538e-06,
      "loss": 0.4446,
      "step": 927
    },
    {
      "epoch": 1.8823529411764706,
      "grad_norm": 0.24925372748219415,
      "learning_rate": 3.6651033585006773e-06,
      "loss": 0.4483,
      "step": 928
    },
    {
      "epoch": 1.8843813387423936,
      "grad_norm": 0.26871305500661863,
      "learning_rate": 3.6537338345818273e-06,
      "loss": 0.4585,
      "step": 929
    },
    {
      "epoch": 1.8864097363083165,
      "grad_norm": 0.26785157954373623,
      "learning_rate": 3.642371810889222e-06,
      "loss": 0.4394,
      "step": 930
    },
    {
      "epoch": 1.8884381338742393,
      "grad_norm": 0.24046144328939262,
      "learning_rate": 3.6310173507221884e-06,
      "loss": 0.4437,
      "step": 931
    },
    {
      "epoch": 1.8904665314401623,
      "grad_norm": 0.26877047764643647,
      "learning_rate": 3.619670517337924e-06,
      "loss": 0.4362,
      "step": 932
    },
    {
      "epoch": 1.892494929006085,
      "grad_norm": 0.25568296686416064,
      "learning_rate": 3.6083313739511316e-06,
      "loss": 0.4373,
      "step": 933
    },
    {
      "epoch": 1.894523326572008,
      "grad_norm": 0.21722376509003644,
      "learning_rate": 3.5969999837336755e-06,
      "loss": 0.4297,
      "step": 934
    },
    {
      "epoch": 1.896551724137931,
      "grad_norm": 0.26002516601207715,
      "learning_rate": 3.5856764098142207e-06,
      "loss": 0.4325,
      "step": 935
    },
    {
      "epoch": 1.898580121703854,
      "grad_norm": 0.24269838922376674,
      "learning_rate": 3.5743607152778915e-06,
      "loss": 0.4321,
      "step": 936
    },
    {
      "epoch": 1.900608519269777,
      "grad_norm": 0.23798392351139502,
      "learning_rate": 3.563052963165915e-06,
      "loss": 0.4244,
      "step": 937
    },
    {
      "epoch": 1.9026369168356998,
      "grad_norm": 0.2540094718666685,
      "learning_rate": 3.5517532164752657e-06,
      "loss": 0.4308,
      "step": 938
    },
    {
      "epoch": 1.9046653144016226,
      "grad_norm": 0.25426471679617113,
      "learning_rate": 3.5404615381583264e-06,
      "loss": 0.4428,
      "step": 939
    },
    {
      "epoch": 1.9066937119675456,
      "grad_norm": 0.28179205544421515,
      "learning_rate": 3.529177991122519e-06,
      "loss": 0.4261,
      "step": 940
    },
    {
      "epoch": 1.9087221095334685,
      "grad_norm": 0.2634589511231162,
      "learning_rate": 3.5179026382299752e-06,
      "loss": 0.4377,
      "step": 941
    },
    {
      "epoch": 1.9107505070993915,
      "grad_norm": 0.22383061723303,
      "learning_rate": 3.506635542297171e-06,
      "loss": 0.4289,
      "step": 942
    },
    {
      "epoch": 1.9127789046653145,
      "grad_norm": 0.32038575840564487,
      "learning_rate": 3.4953767660945825e-06,
      "loss": 0.4446,
      "step": 943
    },
    {
      "epoch": 1.9148073022312373,
      "grad_norm": 0.27337512786201695,
      "learning_rate": 3.484126372346338e-06,
      "loss": 0.4467,
      "step": 944
    },
    {
      "epoch": 1.9168356997971603,
      "grad_norm": 0.23951401753946552,
      "learning_rate": 3.472884423729861e-06,
      "loss": 0.4317,
      "step": 945
    },
    {
      "epoch": 1.918864097363083,
      "grad_norm": 0.27542263838767267,
      "learning_rate": 3.4616509828755306e-06,
      "loss": 0.4555,
      "step": 946
    },
    {
      "epoch": 1.920892494929006,
      "grad_norm": 0.2681899554732679,
      "learning_rate": 3.4504261123663243e-06,
      "loss": 0.4461,
      "step": 947
    },
    {
      "epoch": 1.922920892494929,
      "grad_norm": 0.2585103556320343,
      "learning_rate": 3.439209874737479e-06,
      "loss": 0.4533,
      "step": 948
    },
    {
      "epoch": 1.924949290060852,
      "grad_norm": 0.2371136547205798,
      "learning_rate": 3.4280023324761287e-06,
      "loss": 0.4278,
      "step": 949
    },
    {
      "epoch": 1.9269776876267748,
      "grad_norm": 0.24296409607750263,
      "learning_rate": 3.4168035480209694e-06,
      "loss": 0.4327,
      "step": 950
    },
    {
      "epoch": 1.9290060851926978,
      "grad_norm": 0.26223646296021,
      "learning_rate": 3.4056135837619077e-06,
      "loss": 0.4667,
      "step": 951
    },
    {
      "epoch": 1.9310344827586206,
      "grad_norm": 0.2580495739040847,
      "learning_rate": 3.3944325020397064e-06,
      "loss": 0.4267,
      "step": 952
    },
    {
      "epoch": 1.9330628803245435,
      "grad_norm": 0.2660497650988227,
      "learning_rate": 3.3832603651456486e-06,
      "loss": 0.4427,
      "step": 953
    },
    {
      "epoch": 1.9350912778904665,
      "grad_norm": 0.2643176993040225,
      "learning_rate": 3.3720972353211776e-06,
      "loss": 0.4345,
      "step": 954
    },
    {
      "epoch": 1.9371196754563895,
      "grad_norm": 0.26707924931462784,
      "learning_rate": 3.360943174757564e-06,
      "loss": 0.4334,
      "step": 955
    },
    {
      "epoch": 1.9391480730223125,
      "grad_norm": 0.3073204122467851,
      "learning_rate": 3.3497982455955463e-06,
      "loss": 0.4421,
      "step": 956
    },
    {
      "epoch": 1.9411764705882353,
      "grad_norm": 0.2616512365930539,
      "learning_rate": 3.3386625099249957e-06,
      "loss": 0.4503,
      "step": 957
    },
    {
      "epoch": 1.943204868154158,
      "grad_norm": 0.2744276013043432,
      "learning_rate": 3.327536029784566e-06,
      "loss": 0.4312,
      "step": 958
    },
    {
      "epoch": 1.945233265720081,
      "grad_norm": 0.27126447902632944,
      "learning_rate": 3.3164188671613382e-06,
      "loss": 0.4421,
      "step": 959
    },
    {
      "epoch": 1.947261663286004,
      "grad_norm": 0.25654598396310246,
      "learning_rate": 3.3053110839904966e-06,
      "loss": 0.443,
      "step": 960
    },
    {
      "epoch": 1.949290060851927,
      "grad_norm": 0.23425112631442374,
      "learning_rate": 3.29421274215496e-06,
      "loss": 0.4407,
      "step": 961
    },
    {
      "epoch": 1.95131845841785,
      "grad_norm": 0.2510435054926272,
      "learning_rate": 3.2831239034850593e-06,
      "loss": 0.4319,
      "step": 962
    },
    {
      "epoch": 1.9533468559837728,
      "grad_norm": 0.2552800172832482,
      "learning_rate": 3.2720446297581696e-06,
      "loss": 0.4456,
      "step": 963
    },
    {
      "epoch": 1.9553752535496958,
      "grad_norm": 0.2585097112195621,
      "learning_rate": 3.2609749826983904e-06,
      "loss": 0.441,
      "step": 964
    },
    {
      "epoch": 1.9574036511156185,
      "grad_norm": 0.24437763321049127,
      "learning_rate": 3.2499150239761813e-06,
      "loss": 0.4309,
      "step": 965
    },
    {
      "epoch": 1.9594320486815415,
      "grad_norm": 0.23954437972533332,
      "learning_rate": 3.2388648152080304e-06,
      "loss": 0.4606,
      "step": 966
    },
    {
      "epoch": 1.9614604462474645,
      "grad_norm": 0.25410915324470895,
      "learning_rate": 3.2278244179561107e-06,
      "loss": 0.4602,
      "step": 967
    },
    {
      "epoch": 1.9634888438133875,
      "grad_norm": 0.2637006625969084,
      "learning_rate": 3.2167938937279253e-06,
      "loss": 0.4187,
      "step": 968
    },
    {
      "epoch": 1.9655172413793105,
      "grad_norm": 0.24380671434229967,
      "learning_rate": 3.205773303975982e-06,
      "loss": 0.4301,
      "step": 969
    },
    {
      "epoch": 1.9675456389452333,
      "grad_norm": 0.2598744029890059,
      "learning_rate": 3.1947627100974367e-06,
      "loss": 0.4333,
      "step": 970
    },
    {
      "epoch": 1.969574036511156,
      "grad_norm": 0.28987019955178217,
      "learning_rate": 3.1837621734337607e-06,
      "loss": 0.4327,
      "step": 971
    },
    {
      "epoch": 1.971602434077079,
      "grad_norm": 0.23956329212199567,
      "learning_rate": 3.1727717552703925e-06,
      "loss": 0.4438,
      "step": 972
    },
    {
      "epoch": 1.973630831643002,
      "grad_norm": 0.2667487808411275,
      "learning_rate": 3.1617915168363994e-06,
      "loss": 0.4558,
      "step": 973
    },
    {
      "epoch": 1.975659229208925,
      "grad_norm": 0.2750841532362856,
      "learning_rate": 3.1508215193041392e-06,
      "loss": 0.4479,
      "step": 974
    },
    {
      "epoch": 1.977687626774848,
      "grad_norm": 0.26912523671564775,
      "learning_rate": 3.1398618237889124e-06,
      "loss": 0.4328,
      "step": 975
    },
    {
      "epoch": 1.9797160243407708,
      "grad_norm": 0.2278888454897423,
      "learning_rate": 3.128912491348629e-06,
      "loss": 0.444,
      "step": 976
    },
    {
      "epoch": 1.9817444219066938,
      "grad_norm": 0.26558421097313767,
      "learning_rate": 3.11797358298346e-06,
      "loss": 0.4362,
      "step": 977
    },
    {
      "epoch": 1.9837728194726165,
      "grad_norm": 0.27523538108079326,
      "learning_rate": 3.1070451596355077e-06,
      "loss": 0.4424,
      "step": 978
    },
    {
      "epoch": 1.9858012170385395,
      "grad_norm": 0.25267381856830806,
      "learning_rate": 3.096127282188458e-06,
      "loss": 0.4492,
      "step": 979
    },
    {
      "epoch": 1.9878296146044625,
      "grad_norm": 0.27942724531403346,
      "learning_rate": 3.085220011467246e-06,
      "loss": 0.4542,
      "step": 980
    },
    {
      "epoch": 1.9898580121703855,
      "grad_norm": 0.24473898664507351,
      "learning_rate": 3.074323408237716e-06,
      "loss": 0.441,
      "step": 981
    },
    {
      "epoch": 1.9918864097363083,
      "grad_norm": 0.23237121774012892,
      "learning_rate": 3.063437533206276e-06,
      "loss": 0.4372,
      "step": 982
    },
    {
      "epoch": 1.9939148073022313,
      "grad_norm": 0.2625485845141687,
      "learning_rate": 3.0525624470195746e-06,
      "loss": 0.4449,
      "step": 983
    },
    {
      "epoch": 1.995943204868154,
      "grad_norm": 0.25638898696517093,
      "learning_rate": 3.041698210264149e-06,
      "loss": 0.4514,
      "step": 984
    },
    {
      "epoch": 1.997971602434077,
      "grad_norm": 0.2343260604736344,
      "learning_rate": 3.0308448834660953e-06,
      "loss": 0.4284,
      "step": 985
    },
    {
      "epoch": 2.0,
      "grad_norm": 0.24238153069532428,
      "learning_rate": 3.0200025270907247e-06,
      "loss": 0.4306,
      "step": 986
    },
    {
      "epoch": 2.002028397565923,
      "grad_norm": 0.28236849922172597,
      "learning_rate": 3.009171201542235e-06,
      "loss": 0.4116,
      "step": 987
    },
    {
      "epoch": 2.004056795131846,
      "grad_norm": 0.23747918247491034,
      "learning_rate": 2.9983509671633703e-06,
      "loss": 0.4156,
      "step": 988
    },
    {
      "epoch": 2.006085192697769,
      "grad_norm": 0.22817151306182032,
      "learning_rate": 2.987541884235078e-06,
      "loss": 0.42,
      "step": 989
    },
    {
      "epoch": 2.0081135902636915,
      "grad_norm": 0.23823779320108684,
      "learning_rate": 2.976744012976189e-06,
      "loss": 0.4008,
      "step": 990
    },
    {
      "epoch": 2.0101419878296145,
      "grad_norm": 0.26260611330289385,
      "learning_rate": 2.965957413543063e-06,
      "loss": 0.44,
      "step": 991
    },
    {
      "epoch": 2.0121703853955375,
      "grad_norm": 0.2304201091317437,
      "learning_rate": 2.955182146029271e-06,
      "loss": 0.4227,
      "step": 992
    },
    {
      "epoch": 2.0141987829614605,
      "grad_norm": 0.27500226310080295,
      "learning_rate": 2.944418270465243e-06,
      "loss": 0.4333,
      "step": 993
    },
    {
      "epoch": 2.0162271805273835,
      "grad_norm": 0.24619613736971224,
      "learning_rate": 2.9336658468179526e-06,
      "loss": 0.4539,
      "step": 994
    },
    {
      "epoch": 2.0182555780933065,
      "grad_norm": 0.258843884855112,
      "learning_rate": 2.9229249349905686e-06,
      "loss": 0.4129,
      "step": 995
    },
    {
      "epoch": 2.020283975659229,
      "grad_norm": 0.23370159143654207,
      "learning_rate": 2.9121955948221257e-06,
      "loss": 0.4245,
      "step": 996
    },
    {
      "epoch": 2.022312373225152,
      "grad_norm": 0.24400526409091178,
      "learning_rate": 2.9014778860871916e-06,
      "loss": 0.4055,
      "step": 997
    },
    {
      "epoch": 2.024340770791075,
      "grad_norm": 0.24128622623102924,
      "learning_rate": 2.8907718684955354e-06,
      "loss": 0.4177,
      "step": 998
    },
    {
      "epoch": 2.026369168356998,
      "grad_norm": 0.2409686196128367,
      "learning_rate": 2.880077601691793e-06,
      "loss": 0.4178,
      "step": 999
    },
    {
      "epoch": 2.028397565922921,
      "grad_norm": 0.232421036266123,
      "learning_rate": 2.869395145255131e-06,
      "loss": 0.4245,
      "step": 1000
    },
    {
      "epoch": 2.030425963488844,
      "grad_norm": 0.2375150005287111,
      "learning_rate": 2.8587245586989265e-06,
      "loss": 0.4187,
      "step": 1001
    },
    {
      "epoch": 2.032454361054767,
      "grad_norm": 0.2623079261309285,
      "learning_rate": 2.848065901470418e-06,
      "loss": 0.4265,
      "step": 1002
    },
    {
      "epoch": 2.0344827586206895,
      "grad_norm": 0.22887444913345925,
      "learning_rate": 2.8374192329503934e-06,
      "loss": 0.4285,
      "step": 1003
    },
    {
      "epoch": 2.0365111561866125,
      "grad_norm": 0.22847920404842564,
      "learning_rate": 2.8267846124528487e-06,
      "loss": 0.4128,
      "step": 1004
    },
    {
      "epoch": 2.0385395537525355,
      "grad_norm": 0.2259139996177232,
      "learning_rate": 2.8161620992246497e-06,
      "loss": 0.419,
      "step": 1005
    },
    {
      "epoch": 2.0405679513184585,
      "grad_norm": 0.21793160879951462,
      "learning_rate": 2.805551752445222e-06,
      "loss": 0.4142,
      "step": 1006
    },
    {
      "epoch": 2.0425963488843815,
      "grad_norm": 0.24151170154552778,
      "learning_rate": 2.7949536312262048e-06,
      "loss": 0.4149,
      "step": 1007
    },
    {
      "epoch": 2.0446247464503045,
      "grad_norm": 0.24960710531505148,
      "learning_rate": 2.7843677946111303e-06,
      "loss": 0.4392,
      "step": 1008
    },
    {
      "epoch": 2.046653144016227,
      "grad_norm": 0.22771321734907415,
      "learning_rate": 2.7737943015750862e-06,
      "loss": 0.4118,
      "step": 1009
    },
    {
      "epoch": 2.04868154158215,
      "grad_norm": 0.244924777618989,
      "learning_rate": 2.763233211024397e-06,
      "loss": 0.4299,
      "step": 1010
    },
    {
      "epoch": 2.050709939148073,
      "grad_norm": 0.25002862091563627,
      "learning_rate": 2.752684581796292e-06,
      "loss": 0.4238,
      "step": 1011
    },
    {
      "epoch": 2.052738336713996,
      "grad_norm": 0.24754782545266588,
      "learning_rate": 2.7421484726585715e-06,
      "loss": 0.418,
      "step": 1012
    },
    {
      "epoch": 2.054766734279919,
      "grad_norm": 0.2432593798643919,
      "learning_rate": 2.7316249423092923e-06,
      "loss": 0.4216,
      "step": 1013
    },
    {
      "epoch": 2.056795131845842,
      "grad_norm": 0.25505049371810373,
      "learning_rate": 2.721114049376424e-06,
      "loss": 0.4155,
      "step": 1014
    },
    {
      "epoch": 2.0588235294117645,
      "grad_norm": 0.25221902455894785,
      "learning_rate": 2.7106158524175396e-06,
      "loss": 0.4305,
      "step": 1015
    },
    {
      "epoch": 2.0608519269776875,
      "grad_norm": 0.2164589798164763,
      "learning_rate": 2.700130409919477e-06,
      "loss": 0.4325,
      "step": 1016
    },
    {
      "epoch": 2.0628803245436105,
      "grad_norm": 0.23631726791941768,
      "learning_rate": 2.689657780298019e-06,
      "loss": 0.4255,
      "step": 1017
    },
    {
      "epoch": 2.0649087221095335,
      "grad_norm": 0.21780390519604323,
      "learning_rate": 2.6791980218975665e-06,
      "loss": 0.436,
      "step": 1018
    },
    {
      "epoch": 2.0669371196754565,
      "grad_norm": 0.22104770567193388,
      "learning_rate": 2.6687511929908093e-06,
      "loss": 0.4197,
      "step": 1019
    },
    {
      "epoch": 2.0689655172413794,
      "grad_norm": 0.2578005435984509,
      "learning_rate": 2.6583173517784122e-06,
      "loss": 0.4249,
      "step": 1020
    },
    {
      "epoch": 2.0709939148073024,
      "grad_norm": 0.21563391996273504,
      "learning_rate": 2.6478965563886745e-06,
      "loss": 0.42,
      "step": 1021
    },
    {
      "epoch": 2.073022312373225,
      "grad_norm": 0.24044432168594262,
      "learning_rate": 2.637488864877226e-06,
      "loss": 0.4314,
      "step": 1022
    },
    {
      "epoch": 2.075050709939148,
      "grad_norm": 0.24364118882314256,
      "learning_rate": 2.627094335226682e-06,
      "loss": 0.431,
      "step": 1023
    },
    {
      "epoch": 2.077079107505071,
      "grad_norm": 0.21450973862507416,
      "learning_rate": 2.616713025346341e-06,
      "loss": 0.4294,
      "step": 1024
    },
    {
      "epoch": 2.079107505070994,
      "grad_norm": 0.2299521761466141,
      "learning_rate": 2.6063449930718487e-06,
      "loss": 0.44,
      "step": 1025
    },
    {
      "epoch": 2.081135902636917,
      "grad_norm": 0.22592527614203903,
      "learning_rate": 2.5959902961648785e-06,
      "loss": 0.4206,
      "step": 1026
    },
    {
      "epoch": 2.08316430020284,
      "grad_norm": 0.21132613179252066,
      "learning_rate": 2.5856489923128136e-06,
      "loss": 0.4143,
      "step": 1027
    },
    {
      "epoch": 2.0851926977687625,
      "grad_norm": 0.24327522120054643,
      "learning_rate": 2.5753211391284172e-06,
      "loss": 0.4356,
      "step": 1028
    },
    {
      "epoch": 2.0872210953346855,
      "grad_norm": 0.23487404470302647,
      "learning_rate": 2.5650067941495236e-06,
      "loss": 0.424,
      "step": 1029
    },
    {
      "epoch": 2.0892494929006085,
      "grad_norm": 0.23646893083303633,
      "learning_rate": 2.5547060148387052e-06,
      "loss": 0.4266,
      "step": 1030
    },
    {
      "epoch": 2.0912778904665315,
      "grad_norm": 0.2653114865907158,
      "learning_rate": 2.5444188585829634e-06,
      "loss": 0.4158,
      "step": 1031
    },
    {
      "epoch": 2.0933062880324544,
      "grad_norm": 0.24604220853503694,
      "learning_rate": 2.5341453826934016e-06,
      "loss": 0.4262,
      "step": 1032
    },
    {
      "epoch": 2.0953346855983774,
      "grad_norm": 0.22472938628477665,
      "learning_rate": 2.523885644404906e-06,
      "loss": 0.4227,
      "step": 1033
    },
    {
      "epoch": 2.0973630831643,
      "grad_norm": 0.2431041002961114,
      "learning_rate": 2.5136397008758316e-06,
      "loss": 0.3973,
      "step": 1034
    },
    {
      "epoch": 2.099391480730223,
      "grad_norm": 0.26895574277304035,
      "learning_rate": 2.5034076091876813e-06,
      "loss": 0.4097,
      "step": 1035
    },
    {
      "epoch": 2.101419878296146,
      "grad_norm": 0.24057090910234952,
      "learning_rate": 2.4931894263447874e-06,
      "loss": 0.43,
      "step": 1036
    },
    {
      "epoch": 2.103448275862069,
      "grad_norm": 0.21507717615740657,
      "learning_rate": 2.48298520927399e-06,
      "loss": 0.4418,
      "step": 1037
    },
    {
      "epoch": 2.105476673427992,
      "grad_norm": 0.2531269703705729,
      "learning_rate": 2.472795014824329e-06,
      "loss": 0.4137,
      "step": 1038
    },
    {
      "epoch": 2.107505070993915,
      "grad_norm": 0.23421604802814475,
      "learning_rate": 2.4626188997667224e-06,
      "loss": 0.4301,
      "step": 1039
    },
    {
      "epoch": 2.109533468559838,
      "grad_norm": 0.22056411137661872,
      "learning_rate": 2.452456920793645e-06,
      "loss": 0.4203,
      "step": 1040
    },
    {
      "epoch": 2.1115618661257605,
      "grad_norm": 0.2127408264949295,
      "learning_rate": 2.4423091345188244e-06,
      "loss": 0.413,
      "step": 1041
    },
    {
      "epoch": 2.1135902636916835,
      "grad_norm": 0.23231299470368866,
      "learning_rate": 2.432175597476912e-06,
      "loss": 0.4101,
      "step": 1042
    },
    {
      "epoch": 2.1156186612576064,
      "grad_norm": 0.22758053711971435,
      "learning_rate": 2.4220563661231793e-06,
      "loss": 0.4148,
      "step": 1043
    },
    {
      "epoch": 2.1176470588235294,
      "grad_norm": 0.22716515679858398,
      "learning_rate": 2.4119514968331994e-06,
      "loss": 0.4318,
      "step": 1044
    },
    {
      "epoch": 2.1196754563894524,
      "grad_norm": 0.25264405694357095,
      "learning_rate": 2.4018610459025317e-06,
      "loss": 0.4405,
      "step": 1045
    },
    {
      "epoch": 2.1217038539553754,
      "grad_norm": 0.22166021901109567,
      "learning_rate": 2.391785069546404e-06,
      "loss": 0.4326,
      "step": 1046
    },
    {
      "epoch": 2.123732251521298,
      "grad_norm": 0.20769275275955637,
      "learning_rate": 2.381723623899412e-06,
      "loss": 0.4103,
      "step": 1047
    },
    {
      "epoch": 2.125760649087221,
      "grad_norm": 0.2211888076844064,
      "learning_rate": 2.3716767650151958e-06,
      "loss": 0.4241,
      "step": 1048
    },
    {
      "epoch": 2.127789046653144,
      "grad_norm": 0.22618260235379734,
      "learning_rate": 2.361644548866127e-06,
      "loss": 0.4281,
      "step": 1049
    },
    {
      "epoch": 2.129817444219067,
      "grad_norm": 0.19681545765320668,
      "learning_rate": 2.3516270313430085e-06,
      "loss": 0.4096,
      "step": 1050
    },
    {
      "epoch": 2.13184584178499,
      "grad_norm": 0.22354550557480185,
      "learning_rate": 2.341624268254747e-06,
      "loss": 0.4191,
      "step": 1051
    },
    {
      "epoch": 2.133874239350913,
      "grad_norm": 0.24433598309707785,
      "learning_rate": 2.3316363153280545e-06,
      "loss": 0.4211,
      "step": 1052
    },
    {
      "epoch": 2.135902636916836,
      "grad_norm": 0.22223305356746414,
      "learning_rate": 2.3216632282071345e-06,
      "loss": 0.4168,
      "step": 1053
    },
    {
      "epoch": 2.1379310344827585,
      "grad_norm": 0.2463116720660682,
      "learning_rate": 2.3117050624533695e-06,
      "loss": 0.4216,
      "step": 1054
    },
    {
      "epoch": 2.1399594320486814,
      "grad_norm": 0.2292414141871627,
      "learning_rate": 2.3017618735450142e-06,
      "loss": 0.4189,
      "step": 1055
    },
    {
      "epoch": 2.1419878296146044,
      "grad_norm": 0.2293906261556795,
      "learning_rate": 2.291833716876882e-06,
      "loss": 0.4235,
      "step": 1056
    },
    {
      "epoch": 2.1440162271805274,
      "grad_norm": 0.24209271483842282,
      "learning_rate": 2.2819206477600462e-06,
      "loss": 0.4189,
      "step": 1057
    },
    {
      "epoch": 2.1460446247464504,
      "grad_norm": 0.22419065828817594,
      "learning_rate": 2.272022721421517e-06,
      "loss": 0.4135,
      "step": 1058
    },
    {
      "epoch": 2.1480730223123734,
      "grad_norm": 0.23145652984515344,
      "learning_rate": 2.2621399930039493e-06,
      "loss": 0.4341,
      "step": 1059
    },
    {
      "epoch": 2.150101419878296,
      "grad_norm": 0.21751191083300644,
      "learning_rate": 2.252272517565324e-06,
      "loss": 0.4181,
      "step": 1060
    },
    {
      "epoch": 2.152129817444219,
      "grad_norm": 0.22993435287332262,
      "learning_rate": 2.2424203500786473e-06,
      "loss": 0.4337,
      "step": 1061
    },
    {
      "epoch": 2.154158215010142,
      "grad_norm": 0.21899676864582868,
      "learning_rate": 2.232583545431646e-06,
      "loss": 0.4218,
      "step": 1062
    },
    {
      "epoch": 2.156186612576065,
      "grad_norm": 0.22392076910269204,
      "learning_rate": 2.2227621584264505e-06,
      "loss": 0.419,
      "step": 1063
    },
    {
      "epoch": 2.158215010141988,
      "grad_norm": 0.21348584393988604,
      "learning_rate": 2.2129562437793038e-06,
      "loss": 0.4247,
      "step": 1064
    },
    {
      "epoch": 2.160243407707911,
      "grad_norm": 0.23297988330029862,
      "learning_rate": 2.203165856120251e-06,
      "loss": 0.4224,
      "step": 1065
    },
    {
      "epoch": 2.162271805273834,
      "grad_norm": 0.23781493149231286,
      "learning_rate": 2.193391049992831e-06,
      "loss": 0.4341,
      "step": 1066
    },
    {
      "epoch": 2.1643002028397564,
      "grad_norm": 0.22275228638469205,
      "learning_rate": 2.183631879853776e-06,
      "loss": 0.4425,
      "step": 1067
    },
    {
      "epoch": 2.1663286004056794,
      "grad_norm": 0.22461646680037148,
      "learning_rate": 2.173888400072709e-06,
      "loss": 0.4147,
      "step": 1068
    },
    {
      "epoch": 2.1683569979716024,
      "grad_norm": 0.21049949309816482,
      "learning_rate": 2.164160664931843e-06,
      "loss": 0.4245,
      "step": 1069
    },
    {
      "epoch": 2.1703853955375254,
      "grad_norm": 0.21430039400242437,
      "learning_rate": 2.1544487286256683e-06,
      "loss": 0.4378,
      "step": 1070
    },
    {
      "epoch": 2.1724137931034484,
      "grad_norm": 0.23199598288899176,
      "learning_rate": 2.1447526452606658e-06,
      "loss": 0.4118,
      "step": 1071
    },
    {
      "epoch": 2.1744421906693714,
      "grad_norm": 0.21167889585988517,
      "learning_rate": 2.1350724688549906e-06,
      "loss": 0.4163,
      "step": 1072
    },
    {
      "epoch": 2.176470588235294,
      "grad_norm": 0.23017191200177672,
      "learning_rate": 2.125408253338183e-06,
      "loss": 0.4409,
      "step": 1073
    },
    {
      "epoch": 2.178498985801217,
      "grad_norm": 0.20799988508469375,
      "learning_rate": 2.1157600525508607e-06,
      "loss": 0.4189,
      "step": 1074
    },
    {
      "epoch": 2.18052738336714,
      "grad_norm": 0.23572807753235597,
      "learning_rate": 2.106127920244423e-06,
      "loss": 0.4164,
      "step": 1075
    },
    {
      "epoch": 2.182555780933063,
      "grad_norm": 0.22992700483669545,
      "learning_rate": 2.096511910080749e-06,
      "loss": 0.426,
      "step": 1076
    },
    {
      "epoch": 2.184584178498986,
      "grad_norm": 0.21588264876283472,
      "learning_rate": 2.086912075631896e-06,
      "loss": 0.4194,
      "step": 1077
    },
    {
      "epoch": 2.186612576064909,
      "grad_norm": 0.2214974761194497,
      "learning_rate": 2.0773284703798106e-06,
      "loss": 0.4146,
      "step": 1078
    },
    {
      "epoch": 2.1886409736308314,
      "grad_norm": 0.22844775756533736,
      "learning_rate": 2.067761147716017e-06,
      "loss": 0.426,
      "step": 1079
    },
    {
      "epoch": 2.1906693711967544,
      "grad_norm": 0.2177882735920948,
      "learning_rate": 2.0582101609413336e-06,
      "loss": 0.4253,
      "step": 1080
    },
    {
      "epoch": 2.1926977687626774,
      "grad_norm": 0.22816856383221945,
      "learning_rate": 2.0486755632655643e-06,
      "loss": 0.4181,
      "step": 1081
    },
    {
      "epoch": 2.1947261663286004,
      "grad_norm": 0.20572544759602782,
      "learning_rate": 2.0391574078072098e-06,
      "loss": 0.4338,
      "step": 1082
    },
    {
      "epoch": 2.1967545638945234,
      "grad_norm": 0.21715409172962596,
      "learning_rate": 2.029655747593169e-06,
      "loss": 0.4325,
      "step": 1083
    },
    {
      "epoch": 2.1987829614604464,
      "grad_norm": 0.22778904148629894,
      "learning_rate": 2.0201706355584438e-06,
      "loss": 0.4427,
      "step": 1084
    },
    {
      "epoch": 2.2008113590263694,
      "grad_norm": 0.21776424302179542,
      "learning_rate": 2.010702124545845e-06,
      "loss": 0.4379,
      "step": 1085
    },
    {
      "epoch": 2.202839756592292,
      "grad_norm": 0.22083100828192231,
      "learning_rate": 2.0012502673056916e-06,
      "loss": 0.4312,
      "step": 1086
    },
    {
      "epoch": 2.204868154158215,
      "grad_norm": 0.2181645756950587,
      "learning_rate": 1.9918151164955303e-06,
      "loss": 0.4054,
      "step": 1087
    },
    {
      "epoch": 2.206896551724138,
      "grad_norm": 0.21477177016520144,
      "learning_rate": 1.9823967246798247e-06,
      "loss": 0.4208,
      "step": 1088
    },
    {
      "epoch": 2.208924949290061,
      "grad_norm": 0.22089975720020907,
      "learning_rate": 1.9729951443296823e-06,
      "loss": 0.4193,
      "step": 1089
    },
    {
      "epoch": 2.210953346855984,
      "grad_norm": 0.2375180402570921,
      "learning_rate": 1.9636104278225414e-06,
      "loss": 0.4146,
      "step": 1090
    },
    {
      "epoch": 2.212981744421907,
      "grad_norm": 0.2280887107284783,
      "learning_rate": 1.9542426274418975e-06,
      "loss": 0.4472,
      "step": 1091
    },
    {
      "epoch": 2.2150101419878294,
      "grad_norm": 0.21004124329148144,
      "learning_rate": 1.9448917953769997e-06,
      "loss": 0.4297,
      "step": 1092
    },
    {
      "epoch": 2.2170385395537524,
      "grad_norm": 0.2439000591392261,
      "learning_rate": 1.9355579837225673e-06,
      "loss": 0.4216,
      "step": 1093
    },
    {
      "epoch": 2.2190669371196754,
      "grad_norm": 0.21245475884837753,
      "learning_rate": 1.926241244478496e-06,
      "loss": 0.4016,
      "step": 1094
    },
    {
      "epoch": 2.2210953346855984,
      "grad_norm": 0.20283048813735016,
      "learning_rate": 1.916941629549565e-06,
      "loss": 0.4339,
      "step": 1095
    },
    {
      "epoch": 2.2231237322515214,
      "grad_norm": 0.21444845419887068,
      "learning_rate": 1.9076591907451585e-06,
      "loss": 0.4187,
      "step": 1096
    },
    {
      "epoch": 2.2251521298174444,
      "grad_norm": 0.23397397011592466,
      "learning_rate": 1.8983939797789624e-06,
      "loss": 0.4272,
      "step": 1097
    },
    {
      "epoch": 2.227180527383367,
      "grad_norm": 0.2311083975180503,
      "learning_rate": 1.8891460482686903e-06,
      "loss": 0.4135,
      "step": 1098
    },
    {
      "epoch": 2.22920892494929,
      "grad_norm": 0.21985439610252197,
      "learning_rate": 1.8799154477357883e-06,
      "loss": 0.4317,
      "step": 1099
    },
    {
      "epoch": 2.231237322515213,
      "grad_norm": 0.21406075793840748,
      "learning_rate": 1.8707022296051463e-06,
      "loss": 0.4149,
      "step": 1100
    },
    {
      "epoch": 2.233265720081136,
      "grad_norm": 0.2121078579589814,
      "learning_rate": 1.8615064452048181e-06,
      "loss": 0.4148,
      "step": 1101
    },
    {
      "epoch": 2.235294117647059,
      "grad_norm": 0.2402438216812841,
      "learning_rate": 1.8523281457657304e-06,
      "loss": 0.4309,
      "step": 1102
    },
    {
      "epoch": 2.237322515212982,
      "grad_norm": 0.2035384251507312,
      "learning_rate": 1.8431673824214013e-06,
      "loss": 0.4286,
      "step": 1103
    },
    {
      "epoch": 2.239350912778905,
      "grad_norm": 0.2126170637708122,
      "learning_rate": 1.8340242062076464e-06,
      "loss": 0.4436,
      "step": 1104
    },
    {
      "epoch": 2.2413793103448274,
      "grad_norm": 0.22932310116502713,
      "learning_rate": 1.8248986680623077e-06,
      "loss": 0.4265,
      "step": 1105
    },
    {
      "epoch": 2.2434077079107504,
      "grad_norm": 0.24516642101197825,
      "learning_rate": 1.8157908188249629e-06,
      "loss": 0.4205,
      "step": 1106
    },
    {
      "epoch": 2.2454361054766734,
      "grad_norm": 0.21105143420310493,
      "learning_rate": 1.8067007092366368e-06,
      "loss": 0.4177,
      "step": 1107
    },
    {
      "epoch": 2.2474645030425964,
      "grad_norm": 0.2271843791938691,
      "learning_rate": 1.7976283899395314e-06,
      "loss": 0.4272,
      "step": 1108
    },
    {
      "epoch": 2.2494929006085194,
      "grad_norm": 0.2606829784750863,
      "learning_rate": 1.7885739114767292e-06,
      "loss": 0.4159,
      "step": 1109
    },
    {
      "epoch": 2.2515212981744424,
      "grad_norm": 0.22523984348112944,
      "learning_rate": 1.7795373242919262e-06,
      "loss": 0.4165,
      "step": 1110
    },
    {
      "epoch": 2.2535496957403653,
      "grad_norm": 0.21400931832177206,
      "learning_rate": 1.770518678729139e-06,
      "loss": 0.4105,
      "step": 1111
    },
    {
      "epoch": 2.255578093306288,
      "grad_norm": 0.2046129180145952,
      "learning_rate": 1.761518025032432e-06,
      "loss": 0.4032,
      "step": 1112
    },
    {
      "epoch": 2.257606490872211,
      "grad_norm": 0.2540386247962505,
      "learning_rate": 1.752535413345634e-06,
      "loss": 0.4296,
      "step": 1113
    },
    {
      "epoch": 2.259634888438134,
      "grad_norm": 0.25046920348554313,
      "learning_rate": 1.743570893712056e-06,
      "loss": 0.4357,
      "step": 1114
    },
    {
      "epoch": 2.261663286004057,
      "grad_norm": 0.23240471014026007,
      "learning_rate": 1.734624516074221e-06,
      "loss": 0.435,
      "step": 1115
    },
    {
      "epoch": 2.26369168356998,
      "grad_norm": 0.21284270868111912,
      "learning_rate": 1.7256963302735752e-06,
      "loss": 0.4317,
      "step": 1116
    },
    {
      "epoch": 2.2657200811359024,
      "grad_norm": 0.22550382028155588,
      "learning_rate": 1.716786386050221e-06,
      "loss": 0.424,
      "step": 1117
    },
    {
      "epoch": 2.2677484787018254,
      "grad_norm": 0.2546015476982716,
      "learning_rate": 1.7078947330426293e-06,
      "loss": 0.412,
      "step": 1118
    },
    {
      "epoch": 2.2697768762677484,
      "grad_norm": 0.2560688558925219,
      "learning_rate": 1.6990214207873723e-06,
      "loss": 0.4152,
      "step": 1119
    },
    {
      "epoch": 2.2718052738336714,
      "grad_norm": 0.21382840517176022,
      "learning_rate": 1.6901664987188427e-06,
      "loss": 0.4251,
      "step": 1120
    },
    {
      "epoch": 2.2738336713995944,
      "grad_norm": 0.21736624004739596,
      "learning_rate": 1.681330016168977e-06,
      "loss": 0.4217,
      "step": 1121
    },
    {
      "epoch": 2.2758620689655173,
      "grad_norm": 0.243880388171664,
      "learning_rate": 1.6725120223669873e-06,
      "loss": 0.4201,
      "step": 1122
    },
    {
      "epoch": 2.2778904665314403,
      "grad_norm": 0.22305657465002265,
      "learning_rate": 1.6637125664390747e-06,
      "loss": 0.4334,
      "step": 1123
    },
    {
      "epoch": 2.279918864097363,
      "grad_norm": 0.20585772712501824,
      "learning_rate": 1.654931697408172e-06,
      "loss": 0.4025,
      "step": 1124
    },
    {
      "epoch": 2.281947261663286,
      "grad_norm": 0.23498672742241497,
      "learning_rate": 1.6461694641936544e-06,
      "loss": 0.4247,
      "step": 1125
    },
    {
      "epoch": 2.283975659229209,
      "grad_norm": 0.2079416643428925,
      "learning_rate": 1.6374259156110812e-06,
      "loss": 0.4176,
      "step": 1126
    },
    {
      "epoch": 2.286004056795132,
      "grad_norm": 0.21014196973603883,
      "learning_rate": 1.6287011003719105e-06,
      "loss": 0.4218,
      "step": 1127
    },
    {
      "epoch": 2.288032454361055,
      "grad_norm": 0.2104285156431942,
      "learning_rate": 1.6199950670832404e-06,
      "loss": 0.433,
      "step": 1128
    },
    {
      "epoch": 2.290060851926978,
      "grad_norm": 0.21362029833390553,
      "learning_rate": 1.61130786424753e-06,
      "loss": 0.4141,
      "step": 1129
    },
    {
      "epoch": 2.292089249492901,
      "grad_norm": 0.20503106931484377,
      "learning_rate": 1.6026395402623274e-06,
      "loss": 0.4049,
      "step": 1130
    },
    {
      "epoch": 2.2941176470588234,
      "grad_norm": 0.22739701499174947,
      "learning_rate": 1.5939901434200145e-06,
      "loss": 0.4346,
      "step": 1131
    },
    {
      "epoch": 2.2961460446247464,
      "grad_norm": 0.22602108480176203,
      "learning_rate": 1.5853597219075155e-06,
      "loss": 0.4295,
      "step": 1132
    },
    {
      "epoch": 2.2981744421906694,
      "grad_norm": 0.21651509436587993,
      "learning_rate": 1.5767483238060498e-06,
      "loss": 0.4261,
      "step": 1133
    },
    {
      "epoch": 2.3002028397565923,
      "grad_norm": 0.21656747597299794,
      "learning_rate": 1.5681559970908483e-06,
      "loss": 0.4279,
      "step": 1134
    },
    {
      "epoch": 2.3022312373225153,
      "grad_norm": 0.20613067198582433,
      "learning_rate": 1.5595827896308968e-06,
      "loss": 0.4284,
      "step": 1135
    },
    {
      "epoch": 2.3042596348884383,
      "grad_norm": 0.23061541829921026,
      "learning_rate": 1.5510287491886656e-06,
      "loss": 0.4346,
      "step": 1136
    },
    {
      "epoch": 2.306288032454361,
      "grad_norm": 0.21118986472574205,
      "learning_rate": 1.5424939234198377e-06,
      "loss": 0.4138,
      "step": 1137
    },
    {
      "epoch": 2.308316430020284,
      "grad_norm": 0.21101560454971988,
      "learning_rate": 1.5339783598730568e-06,
      "loss": 0.4017,
      "step": 1138
    },
    {
      "epoch": 2.310344827586207,
      "grad_norm": 0.2366352721943451,
      "learning_rate": 1.5254821059896452e-06,
      "loss": 0.4303,
      "step": 1139
    },
    {
      "epoch": 2.31237322515213,
      "grad_norm": 0.2297908632847552,
      "learning_rate": 1.5170052091033554e-06,
      "loss": 0.4342,
      "step": 1140
    },
    {
      "epoch": 2.314401622718053,
      "grad_norm": 0.2159532223597863,
      "learning_rate": 1.5085477164400975e-06,
      "loss": 0.4289,
      "step": 1141
    },
    {
      "epoch": 2.316430020283976,
      "grad_norm": 0.20351434398936913,
      "learning_rate": 1.500109675117677e-06,
      "loss": 0.4298,
      "step": 1142
    },
    {
      "epoch": 2.3184584178498984,
      "grad_norm": 0.19734288132663438,
      "learning_rate": 1.4916911321455362e-06,
      "loss": 0.4194,
      "step": 1143
    },
    {
      "epoch": 2.3204868154158214,
      "grad_norm": 0.22819051733234882,
      "learning_rate": 1.4832921344244843e-06,
      "loss": 0.4272,
      "step": 1144
    },
    {
      "epoch": 2.3225152129817443,
      "grad_norm": 0.21986135245112998,
      "learning_rate": 1.4749127287464483e-06,
      "loss": 0.4235,
      "step": 1145
    },
    {
      "epoch": 2.3245436105476673,
      "grad_norm": 0.21115357827523498,
      "learning_rate": 1.4665529617941976e-06,
      "loss": 0.4298,
      "step": 1146
    },
    {
      "epoch": 2.3265720081135903,
      "grad_norm": 0.20737453947069742,
      "learning_rate": 1.458212880141099e-06,
      "loss": 0.411,
      "step": 1147
    },
    {
      "epoch": 2.3286004056795133,
      "grad_norm": 0.21843113365234493,
      "learning_rate": 1.4498925302508438e-06,
      "loss": 0.4424,
      "step": 1148
    },
    {
      "epoch": 2.3306288032454363,
      "grad_norm": 0.21555005074781752,
      "learning_rate": 1.4415919584771999e-06,
      "loss": 0.4248,
      "step": 1149
    },
    {
      "epoch": 2.332657200811359,
      "grad_norm": 0.20295914611794796,
      "learning_rate": 1.4333112110637454e-06,
      "loss": 0.4082,
      "step": 1150
    },
    {
      "epoch": 2.334685598377282,
      "grad_norm": 0.2212077477746175,
      "learning_rate": 1.425050334143616e-06,
      "loss": 0.459,
      "step": 1151
    },
    {
      "epoch": 2.336713995943205,
      "grad_norm": 0.1990636635495635,
      "learning_rate": 1.416809373739248e-06,
      "loss": 0.4367,
      "step": 1152
    },
    {
      "epoch": 2.338742393509128,
      "grad_norm": 0.20434262697032096,
      "learning_rate": 1.408588375762114e-06,
      "loss": 0.4347,
      "step": 1153
    },
    {
      "epoch": 2.340770791075051,
      "grad_norm": 0.2107095865680159,
      "learning_rate": 1.4003873860124806e-06,
      "loss": 0.4244,
      "step": 1154
    },
    {
      "epoch": 2.342799188640974,
      "grad_norm": 0.20562340407334298,
      "learning_rate": 1.39220645017914e-06,
      "loss": 0.4154,
      "step": 1155
    },
    {
      "epoch": 2.344827586206897,
      "grad_norm": 0.21546760065677545,
      "learning_rate": 1.3840456138391644e-06,
      "loss": 0.4124,
      "step": 1156
    },
    {
      "epoch": 2.3468559837728193,
      "grad_norm": 0.21415726313979708,
      "learning_rate": 1.3759049224576516e-06,
      "loss": 0.4166,
      "step": 1157
    },
    {
      "epoch": 2.3488843813387423,
      "grad_norm": 0.2081591059165186,
      "learning_rate": 1.3677844213874619e-06,
      "loss": 0.4345,
      "step": 1158
    },
    {
      "epoch": 2.3509127789046653,
      "grad_norm": 0.21507841843538988,
      "learning_rate": 1.3596841558689788e-06,
      "loss": 0.4385,
      "step": 1159
    },
    {
      "epoch": 2.3529411764705883,
      "grad_norm": 0.23309286567273924,
      "learning_rate": 1.35160417102985e-06,
      "loss": 0.4376,
      "step": 1160
    },
    {
      "epoch": 2.3549695740365113,
      "grad_norm": 0.21735749288258727,
      "learning_rate": 1.3435445118847362e-06,
      "loss": 0.4285,
      "step": 1161
    },
    {
      "epoch": 2.356997971602434,
      "grad_norm": 0.22032110844204283,
      "learning_rate": 1.3355052233350584e-06,
      "loss": 0.4329,
      "step": 1162
    },
    {
      "epoch": 2.359026369168357,
      "grad_norm": 0.21027720457562432,
      "learning_rate": 1.3274863501687546e-06,
      "loss": 0.424,
      "step": 1163
    },
    {
      "epoch": 2.36105476673428,
      "grad_norm": 0.20126240915091698,
      "learning_rate": 1.3194879370600244e-06,
      "loss": 0.4178,
      "step": 1164
    },
    {
      "epoch": 2.363083164300203,
      "grad_norm": 0.21934513306166634,
      "learning_rate": 1.3115100285690795e-06,
      "loss": 0.434,
      "step": 1165
    },
    {
      "epoch": 2.365111561866126,
      "grad_norm": 0.22784846954246407,
      "learning_rate": 1.3035526691419026e-06,
      "loss": 0.4261,
      "step": 1166
    },
    {
      "epoch": 2.367139959432049,
      "grad_norm": 0.21559757721652753,
      "learning_rate": 1.2956159031099874e-06,
      "loss": 0.4423,
      "step": 1167
    },
    {
      "epoch": 2.369168356997972,
      "grad_norm": 0.20375264741231824,
      "learning_rate": 1.2876997746901072e-06,
      "loss": 0.4231,
      "step": 1168
    },
    {
      "epoch": 2.3711967545638943,
      "grad_norm": 0.22582205345965,
      "learning_rate": 1.2798043279840544e-06,
      "loss": 0.4205,
      "step": 1169
    },
    {
      "epoch": 2.3732251521298173,
      "grad_norm": 0.20272260765019154,
      "learning_rate": 1.2719296069784066e-06,
      "loss": 0.4194,
      "step": 1170
    },
    {
      "epoch": 2.3752535496957403,
      "grad_norm": 0.21046929559447183,
      "learning_rate": 1.2640756555442684e-06,
      "loss": 0.4175,
      "step": 1171
    },
    {
      "epoch": 2.3772819472616633,
      "grad_norm": 0.1991009263265532,
      "learning_rate": 1.25624251743704e-06,
      "loss": 0.4075,
      "step": 1172
    },
    {
      "epoch": 2.3793103448275863,
      "grad_norm": 0.2014563409229188,
      "learning_rate": 1.248430236296168e-06,
      "loss": 0.4108,
      "step": 1173
    },
    {
      "epoch": 2.3813387423935093,
      "grad_norm": 0.19827163595478306,
      "learning_rate": 1.2406388556448973e-06,
      "loss": 0.4231,
      "step": 1174
    },
    {
      "epoch": 2.3833671399594323,
      "grad_norm": 0.19910455563869722,
      "learning_rate": 1.2328684188900392e-06,
      "loss": 0.4196,
      "step": 1175
    },
    {
      "epoch": 2.385395537525355,
      "grad_norm": 0.22764218872563768,
      "learning_rate": 1.2251189693217175e-06,
      "loss": 0.4279,
      "step": 1176
    },
    {
      "epoch": 2.387423935091278,
      "grad_norm": 0.20834845883699546,
      "learning_rate": 1.2173905501131395e-06,
      "loss": 0.4184,
      "step": 1177
    },
    {
      "epoch": 2.389452332657201,
      "grad_norm": 0.2076352957099338,
      "learning_rate": 1.2096832043203467e-06,
      "loss": 0.4208,
      "step": 1178
    },
    {
      "epoch": 2.391480730223124,
      "grad_norm": 0.2154005957901007,
      "learning_rate": 1.2019969748819783e-06,
      "loss": 0.4329,
      "step": 1179
    },
    {
      "epoch": 2.393509127789047,
      "grad_norm": 0.2127315882140304,
      "learning_rate": 1.1943319046190333e-06,
      "loss": 0.4094,
      "step": 1180
    },
    {
      "epoch": 2.3955375253549693,
      "grad_norm": 0.211980727846535,
      "learning_rate": 1.186688036234625e-06,
      "loss": 0.4286,
      "step": 1181
    },
    {
      "epoch": 2.3975659229208923,
      "grad_norm": 0.22193533826872705,
      "learning_rate": 1.1790654123137552e-06,
      "loss": 0.439,
      "step": 1182
    },
    {
      "epoch": 2.3995943204868153,
      "grad_norm": 0.2082270469299426,
      "learning_rate": 1.1714640753230628e-06,
      "loss": 0.4293,
      "step": 1183
    },
    {
      "epoch": 2.4016227180527383,
      "grad_norm": 0.21587359243665782,
      "learning_rate": 1.1638840676106012e-06,
      "loss": 0.4402,
      "step": 1184
    },
    {
      "epoch": 2.4036511156186613,
      "grad_norm": 0.21270639198488564,
      "learning_rate": 1.1563254314055893e-06,
      "loss": 0.4298,
      "step": 1185
    },
    {
      "epoch": 2.4056795131845843,
      "grad_norm": 0.2128115364202176,
      "learning_rate": 1.1487882088181878e-06,
      "loss": 0.4375,
      "step": 1186
    },
    {
      "epoch": 2.4077079107505073,
      "grad_norm": 0.22562701078001499,
      "learning_rate": 1.1412724418392562e-06,
      "loss": 0.4128,
      "step": 1187
    },
    {
      "epoch": 2.40973630831643,
      "grad_norm": 0.21306710666430376,
      "learning_rate": 1.1337781723401236e-06,
      "loss": 0.4259,
      "step": 1188
    },
    {
      "epoch": 2.411764705882353,
      "grad_norm": 0.21237019124958922,
      "learning_rate": 1.126305442072354e-06,
      "loss": 0.4291,
      "step": 1189
    },
    {
      "epoch": 2.413793103448276,
      "grad_norm": 0.2038177103425429,
      "learning_rate": 1.1188542926675105e-06,
      "loss": 0.4434,
      "step": 1190
    },
    {
      "epoch": 2.415821501014199,
      "grad_norm": 0.20235779272893278,
      "learning_rate": 1.1114247656369305e-06,
      "loss": 0.4196,
      "step": 1191
    },
    {
      "epoch": 2.417849898580122,
      "grad_norm": 0.207541771132307,
      "learning_rate": 1.1040169023714852e-06,
      "loss": 0.4233,
      "step": 1192
    },
    {
      "epoch": 2.4198782961460448,
      "grad_norm": 0.2069360210671329,
      "learning_rate": 1.0966307441413598e-06,
      "loss": 0.4175,
      "step": 1193
    },
    {
      "epoch": 2.4219066937119678,
      "grad_norm": 0.21660637980219738,
      "learning_rate": 1.0892663320958159e-06,
      "loss": 0.434,
      "step": 1194
    },
    {
      "epoch": 2.4239350912778903,
      "grad_norm": 0.2171494713897604,
      "learning_rate": 1.0819237072629606e-06,
      "loss": 0.4333,
      "step": 1195
    },
    {
      "epoch": 2.4259634888438133,
      "grad_norm": 0.19950581682621313,
      "learning_rate": 1.074602910549527e-06,
      "loss": 0.4293,
      "step": 1196
    },
    {
      "epoch": 2.4279918864097363,
      "grad_norm": 0.20241734562156613,
      "learning_rate": 1.0673039827406373e-06,
      "loss": 0.4196,
      "step": 1197
    },
    {
      "epoch": 2.4300202839756593,
      "grad_norm": 0.20765656671508048,
      "learning_rate": 1.0600269644995826e-06,
      "loss": 0.4257,
      "step": 1198
    },
    {
      "epoch": 2.4320486815415823,
      "grad_norm": 0.2170799504799897,
      "learning_rate": 1.0527718963675871e-06,
      "loss": 0.4191,
      "step": 1199
    },
    {
      "epoch": 2.4340770791075053,
      "grad_norm": 0.22418416746092423,
      "learning_rate": 1.0455388187635934e-06,
      "loss": 0.4343,
      "step": 1200
    },
    {
      "epoch": 2.436105476673428,
      "grad_norm": 0.19078362618577957,
      "learning_rate": 1.0383277719840318e-06,
      "loss": 0.4157,
      "step": 1201
    },
    {
      "epoch": 2.438133874239351,
      "grad_norm": 0.2003748966488225,
      "learning_rate": 1.0311387962025909e-06,
      "loss": 0.4215,
      "step": 1202
    },
    {
      "epoch": 2.440162271805274,
      "grad_norm": 0.19976310686453744,
      "learning_rate": 1.0239719314700052e-06,
      "loss": 0.4195,
      "step": 1203
    },
    {
      "epoch": 2.4421906693711968,
      "grad_norm": 0.22313888076921498,
      "learning_rate": 1.01682721771382e-06,
      "loss": 0.4515,
      "step": 1204
    },
    {
      "epoch": 2.4442190669371198,
      "grad_norm": 0.19563394709225057,
      "learning_rate": 1.0097046947381805e-06,
      "loss": 0.4251,
      "step": 1205
    },
    {
      "epoch": 2.4462474645030428,
      "grad_norm": 0.20296918027620575,
      "learning_rate": 1.002604402223597e-06,
      "loss": 0.4215,
      "step": 1206
    },
    {
      "epoch": 2.4482758620689653,
      "grad_norm": 0.22242437623452424,
      "learning_rate": 9.955263797267379e-07,
      "loss": 0.4265,
      "step": 1207
    },
    {
      "epoch": 2.4503042596348883,
      "grad_norm": 0.21163652577118733,
      "learning_rate": 9.88470666680198e-07,
      "loss": 0.4131,
      "step": 1208
    },
    {
      "epoch": 2.4523326572008113,
      "grad_norm": 0.20933229747285825,
      "learning_rate": 9.814373023922851e-07,
      "loss": 0.4236,
      "step": 1209
    },
    {
      "epoch": 2.4543610547667343,
      "grad_norm": 0.21315243831330988,
      "learning_rate": 9.744263260468006e-07,
      "loss": 0.435,
      "step": 1210
    },
    {
      "epoch": 2.4563894523326573,
      "grad_norm": 0.20562599681308738,
      "learning_rate": 9.674377767028142e-07,
      "loss": 0.4172,
      "step": 1211
    },
    {
      "epoch": 2.4584178498985803,
      "grad_norm": 0.20257895719514785,
      "learning_rate": 9.604716932944586e-07,
      "loss": 0.4134,
      "step": 1212
    },
    {
      "epoch": 2.4604462474645032,
      "grad_norm": 0.22547050777766386,
      "learning_rate": 9.53528114630699e-07,
      "loss": 0.4234,
      "step": 1213
    },
    {
      "epoch": 2.462474645030426,
      "grad_norm": 0.21068683683158246,
      "learning_rate": 9.466070793951309e-07,
      "loss": 0.412,
      "step": 1214
    },
    {
      "epoch": 2.464503042596349,
      "grad_norm": 0.19503947523999893,
      "learning_rate": 9.397086261457511e-07,
      "loss": 0.4068,
      "step": 1215
    },
    {
      "epoch": 2.4665314401622718,
      "grad_norm": 0.20131841157503924,
      "learning_rate": 9.328327933147529e-07,
      "loss": 0.4288,
      "step": 1216
    },
    {
      "epoch": 2.4685598377281948,
      "grad_norm": 0.20810392889112128,
      "learning_rate": 9.259796192083071e-07,
      "loss": 0.4365,
      "step": 1217
    },
    {
      "epoch": 2.4705882352941178,
      "grad_norm": 0.1981829866533378,
      "learning_rate": 9.191491420063498e-07,
      "loss": 0.4201,
      "step": 1218
    },
    {
      "epoch": 2.4726166328600407,
      "grad_norm": 0.20932763862229523,
      "learning_rate": 9.123413997623714e-07,
      "loss": 0.433,
      "step": 1219
    },
    {
      "epoch": 2.4746450304259637,
      "grad_norm": 0.20905639475173016,
      "learning_rate": 9.055564304031983e-07,
      "loss": 0.4177,
      "step": 1220
    },
    {
      "epoch": 2.4766734279918863,
      "grad_norm": 0.19957832379880908,
      "learning_rate": 8.987942717287923e-07,
      "loss": 0.4073,
      "step": 1221
    },
    {
      "epoch": 2.4787018255578093,
      "grad_norm": 0.20075531609285768,
      "learning_rate": 8.920549614120272e-07,
      "loss": 0.4113,
      "step": 1222
    },
    {
      "epoch": 2.4807302231237323,
      "grad_norm": 0.19723278260947325,
      "learning_rate": 8.853385369984901e-07,
      "loss": 0.4285,
      "step": 1223
    },
    {
      "epoch": 2.4827586206896552,
      "grad_norm": 0.20655268931559276,
      "learning_rate": 8.786450359062693e-07,
      "loss": 0.4098,
      "step": 1224
    },
    {
      "epoch": 2.4847870182555782,
      "grad_norm": 0.19325755932596467,
      "learning_rate": 8.719744954257375e-07,
      "loss": 0.4235,
      "step": 1225
    },
    {
      "epoch": 2.486815415821501,
      "grad_norm": 0.19576328976443377,
      "learning_rate": 8.65326952719357e-07,
      "loss": 0.4112,
      "step": 1226
    },
    {
      "epoch": 2.4888438133874238,
      "grad_norm": 0.20306071440695178,
      "learning_rate": 8.587024448214637e-07,
      "loss": 0.4235,
      "step": 1227
    },
    {
      "epoch": 2.4908722109533468,
      "grad_norm": 0.19904738781752856,
      "learning_rate": 8.521010086380665e-07,
      "loss": 0.4349,
      "step": 1228
    },
    {
      "epoch": 2.4929006085192698,
      "grad_norm": 0.20400745686884791,
      "learning_rate": 8.455226809466327e-07,
      "loss": 0.4397,
      "step": 1229
    },
    {
      "epoch": 2.4949290060851927,
      "grad_norm": 0.20381145721615376,
      "learning_rate": 8.389674983958951e-07,
      "loss": 0.4215,
      "step": 1230
    },
    {
      "epoch": 2.4969574036511157,
      "grad_norm": 0.20597031464686064,
      "learning_rate": 8.324354975056403e-07,
      "loss": 0.4432,
      "step": 1231
    },
    {
      "epoch": 2.4989858012170387,
      "grad_norm": 0.19503917415814825,
      "learning_rate": 8.25926714666504e-07,
      "loss": 0.4171,
      "step": 1232
    },
    {
      "epoch": 2.5010141987829613,
      "grad_norm": 0.20631145732787284,
      "learning_rate": 8.19441186139776e-07,
      "loss": 0.4151,
      "step": 1233
    },
    {
      "epoch": 2.5030425963488843,
      "grad_norm": 0.19804324671837636,
      "learning_rate": 8.129789480571871e-07,
      "loss": 0.4375,
      "step": 1234
    },
    {
      "epoch": 2.5050709939148073,
      "grad_norm": 0.2053538924682835,
      "learning_rate": 8.065400364207194e-07,
      "loss": 0.4204,
      "step": 1235
    },
    {
      "epoch": 2.5070993914807302,
      "grad_norm": 0.20507695328094316,
      "learning_rate": 8.001244871023978e-07,
      "loss": 0.4215,
      "step": 1236
    },
    {
      "epoch": 2.5091277890466532,
      "grad_norm": 0.2020778943161069,
      "learning_rate": 7.937323358440935e-07,
      "loss": 0.436,
      "step": 1237
    },
    {
      "epoch": 2.5111561866125762,
      "grad_norm": 0.19717927018492556,
      "learning_rate": 7.873636182573246e-07,
      "loss": 0.4269,
      "step": 1238
    },
    {
      "epoch": 2.513184584178499,
      "grad_norm": 0.20760771906484213,
      "learning_rate": 7.810183698230539e-07,
      "loss": 0.4162,
      "step": 1239
    },
    {
      "epoch": 2.5152129817444218,
      "grad_norm": 0.19554162970823757,
      "learning_rate": 7.74696625891499e-07,
      "loss": 0.4455,
      "step": 1240
    },
    {
      "epoch": 2.5172413793103448,
      "grad_norm": 0.19259653076526637,
      "learning_rate": 7.683984216819262e-07,
      "loss": 0.4199,
      "step": 1241
    },
    {
      "epoch": 2.5192697768762677,
      "grad_norm": 0.1903826526751136,
      "learning_rate": 7.621237922824637e-07,
      "loss": 0.4043,
      "step": 1242
    },
    {
      "epoch": 2.5212981744421907,
      "grad_norm": 0.19593576797951362,
      "learning_rate": 7.55872772649896e-07,
      "loss": 0.4218,
      "step": 1243
    },
    {
      "epoch": 2.5233265720081137,
      "grad_norm": 0.19309536052184137,
      "learning_rate": 7.496453976094792e-07,
      "loss": 0.4217,
      "step": 1244
    },
    {
      "epoch": 2.5253549695740363,
      "grad_norm": 0.18638659417397616,
      "learning_rate": 7.434417018547396e-07,
      "loss": 0.4104,
      "step": 1245
    },
    {
      "epoch": 2.5273833671399597,
      "grad_norm": 0.20032188128924508,
      "learning_rate": 7.372617199472854e-07,
      "loss": 0.4201,
      "step": 1246
    },
    {
      "epoch": 2.5294117647058822,
      "grad_norm": 0.20772086030950826,
      "learning_rate": 7.311054863166095e-07,
      "loss": 0.4343,
      "step": 1247
    },
    {
      "epoch": 2.5314401622718052,
      "grad_norm": 0.20485194909914736,
      "learning_rate": 7.249730352599e-07,
      "loss": 0.4455,
      "step": 1248
    },
    {
      "epoch": 2.5334685598377282,
      "grad_norm": 0.2092742376429762,
      "learning_rate": 7.188644009418517e-07,
      "loss": 0.4181,
      "step": 1249
    },
    {
      "epoch": 2.535496957403651,
      "grad_norm": 0.20106236828265725,
      "learning_rate": 7.127796173944695e-07,
      "loss": 0.415,
      "step": 1250
    },
    {
      "epoch": 2.537525354969574,
      "grad_norm": 0.19152259010024542,
      "learning_rate": 7.067187185168862e-07,
      "loss": 0.4112,
      "step": 1251
    },
    {
      "epoch": 2.5395537525354968,
      "grad_norm": 0.19986900172229802,
      "learning_rate": 7.006817380751663e-07,
      "loss": 0.415,
      "step": 1252
    },
    {
      "epoch": 2.5415821501014197,
      "grad_norm": 0.21022052574869007,
      "learning_rate": 6.946687097021249e-07,
      "loss": 0.4249,
      "step": 1253
    },
    {
      "epoch": 2.5436105476673427,
      "grad_norm": 0.19895766345507596,
      "learning_rate": 6.886796668971357e-07,
      "loss": 0.4141,
      "step": 1254
    },
    {
      "epoch": 2.5456389452332657,
      "grad_norm": 0.19335486789886364,
      "learning_rate": 6.827146430259446e-07,
      "loss": 0.4127,
      "step": 1255
    },
    {
      "epoch": 2.5476673427991887,
      "grad_norm": 0.208495695766129,
      "learning_rate": 6.767736713204875e-07,
      "loss": 0.4175,
      "step": 1256
    },
    {
      "epoch": 2.5496957403651117,
      "grad_norm": 0.20993490964717013,
      "learning_rate": 6.70856784878699e-07,
      "loss": 0.4128,
      "step": 1257
    },
    {
      "epoch": 2.5517241379310347,
      "grad_norm": 0.2075338414163081,
      "learning_rate": 6.649640166643345e-07,
      "loss": 0.4182,
      "step": 1258
    },
    {
      "epoch": 2.5537525354969572,
      "grad_norm": 0.2023688081634511,
      "learning_rate": 6.590953995067812e-07,
      "loss": 0.4204,
      "step": 1259
    },
    {
      "epoch": 2.5557809330628802,
      "grad_norm": 0.19409850967219866,
      "learning_rate": 6.53250966100879e-07,
      "loss": 0.4322,
      "step": 1260
    },
    {
      "epoch": 2.5578093306288032,
      "grad_norm": 0.1913149226858275,
      "learning_rate": 6.474307490067383e-07,
      "loss": 0.4141,
      "step": 1261
    },
    {
      "epoch": 2.559837728194726,
      "grad_norm": 0.20165833585104256,
      "learning_rate": 6.416347806495532e-07,
      "loss": 0.4304,
      "step": 1262
    },
    {
      "epoch": 2.561866125760649,
      "grad_norm": 0.2133959810364879,
      "learning_rate": 6.358630933194282e-07,
      "loss": 0.4189,
      "step": 1263
    },
    {
      "epoch": 2.5638945233265718,
      "grad_norm": 0.19969874148953068,
      "learning_rate": 6.301157191711949e-07,
      "loss": 0.4174,
      "step": 1264
    },
    {
      "epoch": 2.565922920892495,
      "grad_norm": 0.20706252646253545,
      "learning_rate": 6.24392690224232e-07,
      "loss": 0.4393,
      "step": 1265
    },
    {
      "epoch": 2.5679513184584177,
      "grad_norm": 0.20300472037867065,
      "learning_rate": 6.186940383622875e-07,
      "loss": 0.4271,
      "step": 1266
    },
    {
      "epoch": 2.5699797160243407,
      "grad_norm": 0.19722975892183572,
      "learning_rate": 6.130197953333017e-07,
      "loss": 0.4199,
      "step": 1267
    },
    {
      "epoch": 2.5720081135902637,
      "grad_norm": 0.23271355241148972,
      "learning_rate": 6.073699927492322e-07,
      "loss": 0.4221,
      "step": 1268
    },
    {
      "epoch": 2.5740365111561867,
      "grad_norm": 0.20001430280233004,
      "learning_rate": 6.017446620858708e-07,
      "loss": 0.4302,
      "step": 1269
    },
    {
      "epoch": 2.5760649087221097,
      "grad_norm": 0.1940735172368702,
      "learning_rate": 5.961438346826792e-07,
      "loss": 0.4064,
      "step": 1270
    },
    {
      "epoch": 2.5780933062880322,
      "grad_norm": 0.200193611783582,
      "learning_rate": 5.905675417426027e-07,
      "loss": 0.4096,
      "step": 1271
    },
    {
      "epoch": 2.5801217038539552,
      "grad_norm": 0.1948400140677278,
      "learning_rate": 5.850158143319062e-07,
      "loss": 0.42,
      "step": 1272
    },
    {
      "epoch": 2.582150101419878,
      "grad_norm": 0.20095501303502,
      "learning_rate": 5.794886833799923e-07,
      "loss": 0.4165,
      "step": 1273
    },
    {
      "epoch": 2.584178498985801,
      "grad_norm": 0.19699496739617986,
      "learning_rate": 5.739861796792373e-07,
      "loss": 0.4251,
      "step": 1274
    },
    {
      "epoch": 2.586206896551724,
      "grad_norm": 0.18736498421780998,
      "learning_rate": 5.685083338848152e-07,
      "loss": 0.4219,
      "step": 1275
    },
    {
      "epoch": 2.588235294117647,
      "grad_norm": 0.19142161931318394,
      "learning_rate": 5.630551765145259e-07,
      "loss": 0.4226,
      "step": 1276
    },
    {
      "epoch": 2.59026369168357,
      "grad_norm": 0.21040783563799034,
      "learning_rate": 5.576267379486294e-07,
      "loss": 0.4204,
      "step": 1277
    },
    {
      "epoch": 2.5922920892494927,
      "grad_norm": 0.19734788641050938,
      "learning_rate": 5.522230484296698e-07,
      "loss": 0.4407,
      "step": 1278
    },
    {
      "epoch": 2.5943204868154157,
      "grad_norm": 0.20409932533075623,
      "learning_rate": 5.468441380623169e-07,
      "loss": 0.4238,
      "step": 1279
    },
    {
      "epoch": 2.5963488843813387,
      "grad_norm": 0.195188730928804,
      "learning_rate": 5.414900368131854e-07,
      "loss": 0.42,
      "step": 1280
    },
    {
      "epoch": 2.5983772819472617,
      "grad_norm": 0.20315590470116973,
      "learning_rate": 5.361607745106817e-07,
      "loss": 0.4488,
      "step": 1281
    },
    {
      "epoch": 2.6004056795131847,
      "grad_norm": 0.21427411849819394,
      "learning_rate": 5.308563808448286e-07,
      "loss": 0.4329,
      "step": 1282
    },
    {
      "epoch": 2.6024340770791072,
      "grad_norm": 0.19681631090283896,
      "learning_rate": 5.255768853671011e-07,
      "loss": 0.4166,
      "step": 1283
    },
    {
      "epoch": 2.6044624746450307,
      "grad_norm": 0.20164986783396652,
      "learning_rate": 5.203223174902644e-07,
      "loss": 0.4344,
      "step": 1284
    },
    {
      "epoch": 2.606490872210953,
      "grad_norm": 0.20118818795698007,
      "learning_rate": 5.150927064882089e-07,
      "loss": 0.4229,
      "step": 1285
    },
    {
      "epoch": 2.608519269776876,
      "grad_norm": 0.20349350089858653,
      "learning_rate": 5.098880814957879e-07,
      "loss": 0.439,
      "step": 1286
    },
    {
      "epoch": 2.610547667342799,
      "grad_norm": 0.1929837562210885,
      "learning_rate": 5.047084715086515e-07,
      "loss": 0.4245,
      "step": 1287
    },
    {
      "epoch": 2.612576064908722,
      "grad_norm": 0.19938952753702144,
      "learning_rate": 4.995539053830901e-07,
      "loss": 0.4376,
      "step": 1288
    },
    {
      "epoch": 2.614604462474645,
      "grad_norm": 0.25582670515994166,
      "learning_rate": 4.944244118358721e-07,
      "loss": 0.436,
      "step": 1289
    },
    {
      "epoch": 2.6166328600405677,
      "grad_norm": 0.18723962838536293,
      "learning_rate": 4.8932001944408e-07,
      "loss": 0.4097,
      "step": 1290
    },
    {
      "epoch": 2.6186612576064907,
      "grad_norm": 0.20201699238583065,
      "learning_rate": 4.842407566449591e-07,
      "loss": 0.4344,
      "step": 1291
    },
    {
      "epoch": 2.6206896551724137,
      "grad_norm": 0.2006121131507193,
      "learning_rate": 4.791866517357491e-07,
      "loss": 0.4311,
      "step": 1292
    },
    {
      "epoch": 2.6227180527383367,
      "grad_norm": 0.19268399289242696,
      "learning_rate": 4.741577328735364e-07,
      "loss": 0.4124,
      "step": 1293
    },
    {
      "epoch": 2.6247464503042597,
      "grad_norm": 0.19455837502266396,
      "learning_rate": 4.6915402807509004e-07,
      "loss": 0.4193,
      "step": 1294
    },
    {
      "epoch": 2.6267748478701827,
      "grad_norm": 0.1994485417420502,
      "learning_rate": 4.641755652167107e-07,
      "loss": 0.422,
      "step": 1295
    },
    {
      "epoch": 2.6288032454361057,
      "grad_norm": 0.2021925008643633,
      "learning_rate": 4.592223720340683e-07,
      "loss": 0.4204,
      "step": 1296
    },
    {
      "epoch": 2.630831643002028,
      "grad_norm": 0.2070469518526052,
      "learning_rate": 4.5429447612205635e-07,
      "loss": 0.4262,
      "step": 1297
    },
    {
      "epoch": 2.632860040567951,
      "grad_norm": 0.2050015178793271,
      "learning_rate": 4.4939190493463347e-07,
      "loss": 0.4297,
      "step": 1298
    },
    {
      "epoch": 2.634888438133874,
      "grad_norm": 0.19191998650799416,
      "learning_rate": 4.445146857846672e-07,
      "loss": 0.4123,
      "step": 1299
    },
    {
      "epoch": 2.636916835699797,
      "grad_norm": 0.20019677376698905,
      "learning_rate": 4.3966284584379125e-07,
      "loss": 0.4207,
      "step": 1300
    },
    {
      "epoch": 2.63894523326572,
      "grad_norm": 0.20154190760176235,
      "learning_rate": 4.3483641214224325e-07,
      "loss": 0.4214,
      "step": 1301
    },
    {
      "epoch": 2.640973630831643,
      "grad_norm": 0.19112163717005387,
      "learning_rate": 4.300354115687216e-07,
      "loss": 0.4066,
      "step": 1302
    },
    {
      "epoch": 2.643002028397566,
      "grad_norm": 0.2049065797345265,
      "learning_rate": 4.2525987087023433e-07,
      "loss": 0.4278,
      "step": 1303
    },
    {
      "epoch": 2.6450304259634887,
      "grad_norm": 0.20077792032685865,
      "learning_rate": 4.2050981665194725e-07,
      "loss": 0.4106,
      "step": 1304
    },
    {
      "epoch": 2.6470588235294117,
      "grad_norm": 0.19670886793857997,
      "learning_rate": 4.1578527537703973e-07,
      "loss": 0.4162,
      "step": 1305
    },
    {
      "epoch": 2.6490872210953347,
      "grad_norm": 0.20223517179246006,
      "learning_rate": 4.110862733665522e-07,
      "loss": 0.4274,
      "step": 1306
    },
    {
      "epoch": 2.6511156186612577,
      "grad_norm": 0.20246216729060276,
      "learning_rate": 4.064128367992459e-07,
      "loss": 0.401,
      "step": 1307
    },
    {
      "epoch": 2.6531440162271807,
      "grad_norm": 0.20479160804826305,
      "learning_rate": 4.0176499171144967e-07,
      "loss": 0.4322,
      "step": 1308
    },
    {
      "epoch": 2.655172413793103,
      "grad_norm": 0.19303447240987615,
      "learning_rate": 3.971427639969233e-07,
      "loss": 0.4116,
      "step": 1309
    },
    {
      "epoch": 2.6572008113590266,
      "grad_norm": 0.19532462070046253,
      "learning_rate": 3.925461794067048e-07,
      "loss": 0.4282,
      "step": 1310
    },
    {
      "epoch": 2.659229208924949,
      "grad_norm": 0.1932703039462651,
      "learning_rate": 3.879752635489736e-07,
      "loss": 0.4242,
      "step": 1311
    },
    {
      "epoch": 2.661257606490872,
      "grad_norm": 0.18995327484912145,
      "learning_rate": 3.834300418889042e-07,
      "loss": 0.4086,
      "step": 1312
    },
    {
      "epoch": 2.663286004056795,
      "grad_norm": 0.19712723620476189,
      "learning_rate": 3.7891053974852597e-07,
      "loss": 0.4355,
      "step": 1313
    },
    {
      "epoch": 2.665314401622718,
      "grad_norm": 0.18827899421413938,
      "learning_rate": 3.744167823065814e-07,
      "loss": 0.4132,
      "step": 1314
    },
    {
      "epoch": 2.667342799188641,
      "grad_norm": 0.1921283549700237,
      "learning_rate": 3.6994879459838375e-07,
      "loss": 0.4246,
      "step": 1315
    },
    {
      "epoch": 2.6693711967545637,
      "grad_norm": 0.19730595533946993,
      "learning_rate": 3.655066015156833e-07,
      "loss": 0.433,
      "step": 1316
    },
    {
      "epoch": 2.6713995943204867,
      "grad_norm": 0.18954426517934628,
      "learning_rate": 3.6109022780652147e-07,
      "loss": 0.4117,
      "step": 1317
    },
    {
      "epoch": 2.6734279918864097,
      "grad_norm": 0.21048616703476836,
      "learning_rate": 3.566996980750992e-07,
      "loss": 0.4377,
      "step": 1318
    },
    {
      "epoch": 2.6754563894523327,
      "grad_norm": 0.18474501088377832,
      "learning_rate": 3.5233503678163696e-07,
      "loss": 0.408,
      "step": 1319
    },
    {
      "epoch": 2.6774847870182557,
      "grad_norm": 0.19348118909977838,
      "learning_rate": 3.4799626824223666e-07,
      "loss": 0.4324,
      "step": 1320
    },
    {
      "epoch": 2.6795131845841786,
      "grad_norm": 0.19716223210839542,
      "learning_rate": 3.4368341662875004e-07,
      "loss": 0.4222,
      "step": 1321
    },
    {
      "epoch": 2.6815415821501016,
      "grad_norm": 0.19338655786479111,
      "learning_rate": 3.3939650596864216e-07,
      "loss": 0.4083,
      "step": 1322
    },
    {
      "epoch": 2.683569979716024,
      "grad_norm": 0.19911462949887845,
      "learning_rate": 3.3513556014485805e-07,
      "loss": 0.4316,
      "step": 1323
    },
    {
      "epoch": 2.685598377281947,
      "grad_norm": 0.18821017617931218,
      "learning_rate": 3.3090060289568683e-07,
      "loss": 0.4078,
      "step": 1324
    },
    {
      "epoch": 2.68762677484787,
      "grad_norm": 0.19504940656143685,
      "learning_rate": 3.26691657814634e-07,
      "loss": 0.4204,
      "step": 1325
    },
    {
      "epoch": 2.689655172413793,
      "grad_norm": 0.18928931518479788,
      "learning_rate": 3.225087483502892e-07,
      "loss": 0.4223,
      "step": 1326
    },
    {
      "epoch": 2.691683569979716,
      "grad_norm": 0.18860423501533546,
      "learning_rate": 3.183518978061895e-07,
      "loss": 0.399,
      "step": 1327
    },
    {
      "epoch": 2.6937119675456387,
      "grad_norm": 0.1941820210349148,
      "learning_rate": 3.1422112934069993e-07,
      "loss": 0.4154,
      "step": 1328
    },
    {
      "epoch": 2.695740365111562,
      "grad_norm": 0.21535273594293403,
      "learning_rate": 3.101164659668732e-07,
      "loss": 0.4494,
      "step": 1329
    },
    {
      "epoch": 2.6977687626774847,
      "grad_norm": 0.20103615770627253,
      "learning_rate": 3.06037930552332e-07,
      "loss": 0.4236,
      "step": 1330
    },
    {
      "epoch": 2.6997971602434077,
      "grad_norm": 0.1911279570846965,
      "learning_rate": 3.0198554581913343e-07,
      "loss": 0.4168,
      "step": 1331
    },
    {
      "epoch": 2.7018255578093306,
      "grad_norm": 0.20520611029748437,
      "learning_rate": 2.9795933434364764e-07,
      "loss": 0.4357,
      "step": 1332
    },
    {
      "epoch": 2.7038539553752536,
      "grad_norm": 0.1879292873114142,
      "learning_rate": 2.9395931855643043e-07,
      "loss": 0.4065,
      "step": 1333
    },
    {
      "epoch": 2.7058823529411766,
      "grad_norm": 0.1992887103544023,
      "learning_rate": 2.8998552074209485e-07,
      "loss": 0.4541,
      "step": 1334
    },
    {
      "epoch": 2.707910750507099,
      "grad_norm": 0.19828567877244627,
      "learning_rate": 2.860379630391935e-07,
      "loss": 0.4265,
      "step": 1335
    },
    {
      "epoch": 2.709939148073022,
      "grad_norm": 0.19656685567166624,
      "learning_rate": 2.821166674400905e-07,
      "loss": 0.4289,
      "step": 1336
    },
    {
      "epoch": 2.711967545638945,
      "grad_norm": 0.19974922625181465,
      "learning_rate": 2.7822165579084013e-07,
      "loss": 0.4289,
      "step": 1337
    },
    {
      "epoch": 2.713995943204868,
      "grad_norm": 0.18814907388199428,
      "learning_rate": 2.7435294979106574e-07,
      "loss": 0.4149,
      "step": 1338
    },
    {
      "epoch": 2.716024340770791,
      "grad_norm": 0.1922512093200878,
      "learning_rate": 2.705105709938388e-07,
      "loss": 0.4269,
      "step": 1339
    },
    {
      "epoch": 2.718052738336714,
      "grad_norm": 0.18947063467821607,
      "learning_rate": 2.666945408055571e-07,
      "loss": 0.4154,
      "step": 1340
    },
    {
      "epoch": 2.720081135902637,
      "grad_norm": 0.19466070100506686,
      "learning_rate": 2.629048804858275e-07,
      "loss": 0.4427,
      "step": 1341
    },
    {
      "epoch": 2.7221095334685597,
      "grad_norm": 0.1913208916466919,
      "learning_rate": 2.5914161114734937e-07,
      "loss": 0.4293,
      "step": 1342
    },
    {
      "epoch": 2.7241379310344827,
      "grad_norm": 0.20033829962288305,
      "learning_rate": 2.5540475375578967e-07,
      "loss": 0.4252,
      "step": 1343
    },
    {
      "epoch": 2.7261663286004056,
      "grad_norm": 0.21361197212257987,
      "learning_rate": 2.5169432912967527e-07,
      "loss": 0.428,
      "step": 1344
    },
    {
      "epoch": 2.7281947261663286,
      "grad_norm": 0.19583805550370662,
      "learning_rate": 2.4801035794026987e-07,
      "loss": 0.4218,
      "step": 1345
    },
    {
      "epoch": 2.7302231237322516,
      "grad_norm": 0.19591688102116775,
      "learning_rate": 2.4435286071146256e-07,
      "loss": 0.4449,
      "step": 1346
    },
    {
      "epoch": 2.732251521298174,
      "grad_norm": 0.202988338525561,
      "learning_rate": 2.407218578196524e-07,
      "loss": 0.4162,
      "step": 1347
    },
    {
      "epoch": 2.7342799188640976,
      "grad_norm": 0.20167583915118886,
      "learning_rate": 2.3711736949363516e-07,
      "loss": 0.4381,
      "step": 1348
    },
    {
      "epoch": 2.73630831643002,
      "grad_norm": 0.1950086217624508,
      "learning_rate": 2.3353941581449048e-07,
      "loss": 0.4057,
      "step": 1349
    },
    {
      "epoch": 2.738336713995943,
      "grad_norm": 0.18992097205004754,
      "learning_rate": 2.2998801671546945e-07,
      "loss": 0.4347,
      "step": 1350
    },
    {
      "epoch": 2.740365111561866,
      "grad_norm": 0.19172607731976585,
      "learning_rate": 2.2646319198188495e-07,
      "loss": 0.4141,
      "step": 1351
    },
    {
      "epoch": 2.742393509127789,
      "grad_norm": 0.18657418979649903,
      "learning_rate": 2.2296496125099987e-07,
      "loss": 0.4343,
      "step": 1352
    },
    {
      "epoch": 2.744421906693712,
      "grad_norm": 0.19706083656475834,
      "learning_rate": 2.1949334401192013e-07,
      "loss": 0.427,
      "step": 1353
    },
    {
      "epoch": 2.7464503042596347,
      "grad_norm": 0.1975365878719145,
      "learning_rate": 2.160483596054802e-07,
      "loss": 0.4103,
      "step": 1354
    },
    {
      "epoch": 2.7484787018255576,
      "grad_norm": 0.18540447331660262,
      "learning_rate": 2.1263002722414383e-07,
      "loss": 0.414,
      "step": 1355
    },
    {
      "epoch": 2.7505070993914806,
      "grad_norm": 0.1958009824810047,
      "learning_rate": 2.0923836591189183e-07,
      "loss": 0.4107,
      "step": 1356
    },
    {
      "epoch": 2.7525354969574036,
      "grad_norm": 0.18803433427235602,
      "learning_rate": 2.0587339456411503e-07,
      "loss": 0.4262,
      "step": 1357
    },
    {
      "epoch": 2.7545638945233266,
      "grad_norm": 0.18530679723442187,
      "learning_rate": 2.0253513192751374e-07,
      "loss": 0.4188,
      "step": 1358
    },
    {
      "epoch": 2.7565922920892496,
      "grad_norm": 0.1955785508469364,
      "learning_rate": 1.9922359659998724e-07,
      "loss": 0.4207,
      "step": 1359
    },
    {
      "epoch": 2.7586206896551726,
      "grad_norm": 0.1930595329665,
      "learning_rate": 1.9593880703053681e-07,
      "loss": 0.4101,
      "step": 1360
    },
    {
      "epoch": 2.760649087221095,
      "grad_norm": 0.19439057310650665,
      "learning_rate": 1.9268078151915724e-07,
      "loss": 0.4118,
      "step": 1361
    },
    {
      "epoch": 2.762677484787018,
      "grad_norm": 0.18831744906694275,
      "learning_rate": 1.8944953821673827e-07,
      "loss": 0.4209,
      "step": 1362
    },
    {
      "epoch": 2.764705882352941,
      "grad_norm": 0.18749485253668335,
      "learning_rate": 1.8624509512496336e-07,
      "loss": 0.4021,
      "step": 1363
    },
    {
      "epoch": 2.766734279918864,
      "grad_norm": 0.19663256425389558,
      "learning_rate": 1.830674700962054e-07,
      "loss": 0.4262,
      "step": 1364
    },
    {
      "epoch": 2.768762677484787,
      "grad_norm": 0.19138881364781526,
      "learning_rate": 1.799166808334335e-07,
      "loss": 0.417,
      "step": 1365
    },
    {
      "epoch": 2.77079107505071,
      "grad_norm": 0.19184748235308963,
      "learning_rate": 1.767927448901091e-07,
      "loss": 0.4175,
      "step": 1366
    },
    {
      "epoch": 2.772819472616633,
      "grad_norm": 0.18491343407274913,
      "learning_rate": 1.7369567967009226e-07,
      "loss": 0.4189,
      "step": 1367
    },
    {
      "epoch": 2.7748478701825556,
      "grad_norm": 0.20308571080295404,
      "learning_rate": 1.7062550242753994e-07,
      "loss": 0.446,
      "step": 1368
    },
    {
      "epoch": 2.7768762677484786,
      "grad_norm": 0.19715976534658522,
      "learning_rate": 1.6758223026681507e-07,
      "loss": 0.4288,
      "step": 1369
    },
    {
      "epoch": 2.7789046653144016,
      "grad_norm": 0.19132130518148485,
      "learning_rate": 1.6456588014238828e-07,
      "loss": 0.4386,
      "step": 1370
    },
    {
      "epoch": 2.7809330628803246,
      "grad_norm": 0.20294465370974268,
      "learning_rate": 1.615764688587429e-07,
      "loss": 0.4215,
      "step": 1371
    },
    {
      "epoch": 2.7829614604462476,
      "grad_norm": 0.2003307363061287,
      "learning_rate": 1.58614013070284e-07,
      "loss": 0.425,
      "step": 1372
    },
    {
      "epoch": 2.78498985801217,
      "grad_norm": 0.18697137454765062,
      "learning_rate": 1.5567852928124237e-07,
      "loss": 0.4179,
      "step": 1373
    },
    {
      "epoch": 2.7870182555780936,
      "grad_norm": 0.19500366183324044,
      "learning_rate": 1.5277003384558619e-07,
      "loss": 0.4171,
      "step": 1374
    },
    {
      "epoch": 2.789046653144016,
      "grad_norm": 0.18607946403132988,
      "learning_rate": 1.4988854296692557e-07,
      "loss": 0.4294,
      "step": 1375
    },
    {
      "epoch": 2.791075050709939,
      "grad_norm": 0.19774431770318912,
      "learning_rate": 1.4703407269842707e-07,
      "loss": 0.4351,
      "step": 1376
    },
    {
      "epoch": 2.793103448275862,
      "grad_norm": 0.18792425115815417,
      "learning_rate": 1.442066389427199e-07,
      "loss": 0.41,
      "step": 1377
    },
    {
      "epoch": 2.795131845841785,
      "grad_norm": 0.18877120487538723,
      "learning_rate": 1.414062574518099e-07,
      "loss": 0.4091,
      "step": 1378
    },
    {
      "epoch": 2.797160243407708,
      "grad_norm": 0.18870202419904505,
      "learning_rate": 1.386329438269929e-07,
      "loss": 0.4213,
      "step": 1379
    },
    {
      "epoch": 2.7991886409736306,
      "grad_norm": 0.19106169493977537,
      "learning_rate": 1.358867135187636e-07,
      "loss": 0.4376,
      "step": 1380
    },
    {
      "epoch": 2.8012170385395536,
      "grad_norm": 0.1954784148429821,
      "learning_rate": 1.3316758182673307e-07,
      "loss": 0.4252,
      "step": 1381
    },
    {
      "epoch": 2.8032454361054766,
      "grad_norm": 0.18248394890680264,
      "learning_rate": 1.3047556389954252e-07,
      "loss": 0.412,
      "step": 1382
    },
    {
      "epoch": 2.8052738336713996,
      "grad_norm": 0.20191210767120943,
      "learning_rate": 1.2781067473477905e-07,
      "loss": 0.4287,
      "step": 1383
    },
    {
      "epoch": 2.8073022312373226,
      "grad_norm": 0.1955260915234217,
      "learning_rate": 1.2517292917889057e-07,
      "loss": 0.4321,
      "step": 1384
    },
    {
      "epoch": 2.8093306288032456,
      "grad_norm": 0.19452581677147418,
      "learning_rate": 1.225623419271055e-07,
      "loss": 0.4329,
      "step": 1385
    },
    {
      "epoch": 2.8113590263691686,
      "grad_norm": 0.19686261355743814,
      "learning_rate": 1.1997892752334983e-07,
      "loss": 0.4272,
      "step": 1386
    },
    {
      "epoch": 2.813387423935091,
      "grad_norm": 0.1913708123294258,
      "learning_rate": 1.1742270036016523e-07,
      "loss": 0.4253,
      "step": 1387
    },
    {
      "epoch": 2.815415821501014,
      "grad_norm": 0.18919649952758794,
      "learning_rate": 1.1489367467862999e-07,
      "loss": 0.4155,
      "step": 1388
    },
    {
      "epoch": 2.817444219066937,
      "grad_norm": 0.2000542267041068,
      "learning_rate": 1.1239186456828033e-07,
      "loss": 0.4127,
      "step": 1389
    },
    {
      "epoch": 2.81947261663286,
      "grad_norm": 0.20262754302798033,
      "learning_rate": 1.0991728396702983e-07,
      "loss": 0.4161,
      "step": 1390
    },
    {
      "epoch": 2.821501014198783,
      "grad_norm": 0.18943590183283665,
      "learning_rate": 1.0746994666109234e-07,
      "loss": 0.4036,
      "step": 1391
    },
    {
      "epoch": 2.8235294117647056,
      "grad_norm": 0.1877473430647031,
      "learning_rate": 1.0504986628490755e-07,
      "loss": 0.4094,
      "step": 1392
    },
    {
      "epoch": 2.825557809330629,
      "grad_norm": 0.18516136881694376,
      "learning_rate": 1.0265705632106216e-07,
      "loss": 0.4266,
      "step": 1393
    },
    {
      "epoch": 2.8275862068965516,
      "grad_norm": 0.19279809778715779,
      "learning_rate": 1.0029153010021497e-07,
      "loss": 0.4327,
      "step": 1394
    },
    {
      "epoch": 2.8296146044624746,
      "grad_norm": 0.1957175279600048,
      "learning_rate": 9.795330080102527e-08,
      "loss": 0.4235,
      "step": 1395
    },
    {
      "epoch": 2.8316430020283976,
      "grad_norm": 0.19794394883880334,
      "learning_rate": 9.564238145007621e-08,
      "loss": 0.4076,
      "step": 1396
    },
    {
      "epoch": 2.8336713995943206,
      "grad_norm": 0.18947336533787218,
      "learning_rate": 9.335878492180373e-08,
      "loss": 0.4197,
      "step": 1397
    },
    {
      "epoch": 2.8356997971602436,
      "grad_norm": 0.19575017702319386,
      "learning_rate": 9.110252393842556e-08,
      "loss": 0.4209,
      "step": 1398
    },
    {
      "epoch": 2.837728194726166,
      "grad_norm": 0.19280384306766873,
      "learning_rate": 8.887361106986848e-08,
      "loss": 0.4159,
      "step": 1399
    },
    {
      "epoch": 2.839756592292089,
      "grad_norm": 0.20013633912077497,
      "learning_rate": 8.66720587337011e-08,
      "loss": 0.4276,
      "step": 1400
    },
    {
      "epoch": 2.841784989858012,
      "grad_norm": 0.18831900548624658,
      "learning_rate": 8.44978791950607e-08,
      "loss": 0.4296,
      "step": 1401
    },
    {
      "epoch": 2.843813387423935,
      "grad_norm": 0.1952338613740282,
      "learning_rate": 8.235108456658814e-08,
      "loss": 0.4428,
      "step": 1402
    },
    {
      "epoch": 2.845841784989858,
      "grad_norm": 0.19450781423272948,
      "learning_rate": 8.023168680835913e-08,
      "loss": 0.423,
      "step": 1403
    },
    {
      "epoch": 2.847870182555781,
      "grad_norm": 0.20034839079037273,
      "learning_rate": 7.813969772781815e-08,
      "loss": 0.4337,
      "step": 1404
    },
    {
      "epoch": 2.849898580121704,
      "grad_norm": 0.19001314626781304,
      "learning_rate": 7.60751289797118e-08,
      "loss": 0.4188,
      "step": 1405
    },
    {
      "epoch": 2.8519269776876266,
      "grad_norm": 0.19422310577822363,
      "learning_rate": 7.403799206602447e-08,
      "loss": 0.4268,
      "step": 1406
    },
    {
      "epoch": 2.8539553752535496,
      "grad_norm": 0.19508194111876698,
      "learning_rate": 7.202829833591496e-08,
      "loss": 0.412,
      "step": 1407
    },
    {
      "epoch": 2.8559837728194726,
      "grad_norm": 0.18644532964953248,
      "learning_rate": 7.004605898565164e-08,
      "loss": 0.4119,
      "step": 1408
    },
    {
      "epoch": 2.8580121703853956,
      "grad_norm": 0.19225715159601525,
      "learning_rate": 6.809128505855189e-08,
      "loss": 0.4186,
      "step": 1409
    },
    {
      "epoch": 2.8600405679513186,
      "grad_norm": 0.19321951190955997,
      "learning_rate": 6.616398744491826e-08,
      "loss": 0.4301,
      "step": 1410
    },
    {
      "epoch": 2.862068965517241,
      "grad_norm": 0.19887311549978218,
      "learning_rate": 6.426417688197961e-08,
      "loss": 0.4306,
      "step": 1411
    },
    {
      "epoch": 2.8640973630831645,
      "grad_norm": 0.19653596741882617,
      "learning_rate": 6.239186395383067e-08,
      "loss": 0.4129,
      "step": 1412
    },
    {
      "epoch": 2.866125760649087,
      "grad_norm": 0.2001723150087194,
      "learning_rate": 6.054705909137426e-08,
      "loss": 0.4126,
      "step": 1413
    },
    {
      "epoch": 2.86815415821501,
      "grad_norm": 0.20424192805001368,
      "learning_rate": 5.872977257226187e-08,
      "loss": 0.4377,
      "step": 1414
    },
    {
      "epoch": 2.870182555780933,
      "grad_norm": 0.19465843477776992,
      "learning_rate": 5.6940014520834865e-08,
      "loss": 0.4256,
      "step": 1415
    },
    {
      "epoch": 2.872210953346856,
      "grad_norm": 0.1912485134946282,
      "learning_rate": 5.5177794908071736e-08,
      "loss": 0.4165,
      "step": 1416
    },
    {
      "epoch": 2.874239350912779,
      "grad_norm": 0.19097040357923942,
      "learning_rate": 5.344312355153036e-08,
      "loss": 0.417,
      "step": 1417
    },
    {
      "epoch": 2.8762677484787016,
      "grad_norm": 0.19258558333948456,
      "learning_rate": 5.173601011529361e-08,
      "loss": 0.4265,
      "step": 1418
    },
    {
      "epoch": 2.8782961460446246,
      "grad_norm": 0.1947158303138187,
      "learning_rate": 5.005646410991549e-08,
      "loss": 0.4266,
      "step": 1419
    },
    {
      "epoch": 2.8803245436105476,
      "grad_norm": 0.19130522750587425,
      "learning_rate": 4.840449489236787e-08,
      "loss": 0.4217,
      "step": 1420
    },
    {
      "epoch": 2.8823529411764706,
      "grad_norm": 0.20204978542617627,
      "learning_rate": 4.678011166598884e-08,
      "loss": 0.426,
      "step": 1421
    },
    {
      "epoch": 2.8843813387423936,
      "grad_norm": 0.19057314933757366,
      "learning_rate": 4.518332348043219e-08,
      "loss": 0.4245,
      "step": 1422
    },
    {
      "epoch": 2.8864097363083165,
      "grad_norm": 0.18989032994024574,
      "learning_rate": 4.3614139231614725e-08,
      "loss": 0.4026,
      "step": 1423
    },
    {
      "epoch": 2.8884381338742395,
      "grad_norm": 0.19631186230064648,
      "learning_rate": 4.207256766166845e-08,
      "loss": 0.4195,
      "step": 1424
    },
    {
      "epoch": 2.890466531440162,
      "grad_norm": 0.19014835986471243,
      "learning_rate": 4.0558617358892326e-08,
      "loss": 0.4182,
      "step": 1425
    },
    {
      "epoch": 2.892494929006085,
      "grad_norm": 0.19581876372885243,
      "learning_rate": 3.9072296757702857e-08,
      "loss": 0.4327,
      "step": 1426
    },
    {
      "epoch": 2.894523326572008,
      "grad_norm": 0.19177574162346972,
      "learning_rate": 3.7613614138587995e-08,
      "loss": 0.4171,
      "step": 1427
    },
    {
      "epoch": 2.896551724137931,
      "grad_norm": 0.18722304142442478,
      "learning_rate": 3.61825776280611e-08,
      "loss": 0.4238,
      "step": 1428
    },
    {
      "epoch": 2.898580121703854,
      "grad_norm": 0.186595812334048,
      "learning_rate": 3.477919519861428e-08,
      "loss": 0.4217,
      "step": 1429
    },
    {
      "epoch": 2.900608519269777,
      "grad_norm": 0.18702345267169543,
      "learning_rate": 3.3403474668677326e-08,
      "loss": 0.4157,
      "step": 1430
    },
    {
      "epoch": 2.9026369168357,
      "grad_norm": 0.19294603945623523,
      "learning_rate": 3.205542370256997e-08,
      "loss": 0.4104,
      "step": 1431
    },
    {
      "epoch": 2.9046653144016226,
      "grad_norm": 0.19182700099643712,
      "learning_rate": 3.073504981046138e-08,
      "loss": 0.428,
      "step": 1432
    },
    {
      "epoch": 2.9066937119675456,
      "grad_norm": 0.1890339457428133,
      "learning_rate": 2.944236034832959e-08,
      "loss": 0.4163,
      "step": 1433
    },
    {
      "epoch": 2.9087221095334685,
      "grad_norm": 0.1778377014337382,
      "learning_rate": 2.817736251791714e-08,
      "loss": 0.4266,
      "step": 1434
    },
    {
      "epoch": 2.9107505070993915,
      "grad_norm": 0.18778548693164254,
      "learning_rate": 2.6940063366693303e-08,
      "loss": 0.428,
      "step": 1435
    },
    {
      "epoch": 2.9127789046653145,
      "grad_norm": 0.19207608083645855,
      "learning_rate": 2.5730469787815792e-08,
      "loss": 0.4334,
      "step": 1436
    },
    {
      "epoch": 2.914807302231237,
      "grad_norm": 0.19428391472359505,
      "learning_rate": 2.4548588520089123e-08,
      "loss": 0.4259,
      "step": 1437
    },
    {
      "epoch": 2.9168356997971605,
      "grad_norm": 0.19286802128939184,
      "learning_rate": 2.339442614792964e-08,
      "loss": 0.4196,
      "step": 1438
    },
    {
      "epoch": 2.918864097363083,
      "grad_norm": 0.1852280670971229,
      "learning_rate": 2.2267989101328878e-08,
      "loss": 0.4268,
      "step": 1439
    },
    {
      "epoch": 2.920892494929006,
      "grad_norm": 0.19282442187634316,
      "learning_rate": 2.1169283655815276e-08,
      "loss": 0.4173,
      "step": 1440
    },
    {
      "epoch": 2.922920892494929,
      "grad_norm": 0.1875434390917842,
      "learning_rate": 2.0098315932421952e-08,
      "loss": 0.4306,
      "step": 1441
    },
    {
      "epoch": 2.924949290060852,
      "grad_norm": 0.19204308241376192,
      "learning_rate": 1.9055091897650647e-08,
      "loss": 0.4109,
      "step": 1442
    },
    {
      "epoch": 2.926977687626775,
      "grad_norm": 0.1868956535018828,
      "learning_rate": 1.803961736344062e-08,
      "loss": 0.4313,
      "step": 1443
    },
    {
      "epoch": 2.9290060851926976,
      "grad_norm": 0.1912973244220651,
      "learning_rate": 1.7051897987134248e-08,
      "loss": 0.4219,
      "step": 1444
    },
    {
      "epoch": 2.9310344827586206,
      "grad_norm": 0.19299185687248652,
      "learning_rate": 1.6091939271446478e-08,
      "loss": 0.4216,
      "step": 1445
    },
    {
      "epoch": 2.9330628803245435,
      "grad_norm": 0.18388558554699813,
      "learning_rate": 1.51597465644332e-08,
      "loss": 0.418,
      "step": 1446
    },
    {
      "epoch": 2.9350912778904665,
      "grad_norm": 0.18652715721456958,
      "learning_rate": 1.4255325059463477e-08,
      "loss": 0.4159,
      "step": 1447
    },
    {
      "epoch": 2.9371196754563895,
      "grad_norm": 0.18859199804756302,
      "learning_rate": 1.337867979518903e-08,
      "loss": 0.4313,
      "step": 1448
    },
    {
      "epoch": 2.9391480730223125,
      "grad_norm": 0.1928256719637985,
      "learning_rate": 1.252981565551481e-08,
      "loss": 0.4374,
      "step": 1449
    },
    {
      "epoch": 2.9411764705882355,
      "grad_norm": 0.19632173282729626,
      "learning_rate": 1.1708737369576229e-08,
      "loss": 0.4075,
      "step": 1450
    },
    {
      "epoch": 2.943204868154158,
      "grad_norm": 0.19395475353789127,
      "learning_rate": 1.0915449511708088e-08,
      "loss": 0.4343,
      "step": 1451
    },
    {
      "epoch": 2.945233265720081,
      "grad_norm": 0.20354008234643953,
      "learning_rate": 1.0149956501421254e-08,
      "loss": 0.4508,
      "step": 1452
    },
    {
      "epoch": 2.947261663286004,
      "grad_norm": 0.20075751130921743,
      "learning_rate": 9.412262603378797e-09,
      "loss": 0.4328,
      "step": 1453
    },
    {
      "epoch": 2.949290060851927,
      "grad_norm": 0.1930917253981488,
      "learning_rate": 8.70237192736989e-09,
      "loss": 0.3979,
      "step": 1454
    },
    {
      "epoch": 2.95131845841785,
      "grad_norm": 0.18998008575223552,
      "learning_rate": 8.020288428289836e-09,
      "loss": 0.4287,
      "step": 1455
    },
    {
      "epoch": 2.9533468559837726,
      "grad_norm": 0.18953130616620864,
      "learning_rate": 7.366015906115076e-09,
      "loss": 0.4133,
      "step": 1456
    },
    {
      "epoch": 2.955375253549696,
      "grad_norm": 0.19550511112622204,
      "learning_rate": 6.739558005884883e-09,
      "loss": 0.4431,
      "step": 1457
    },
    {
      "epoch": 2.9574036511156185,
      "grad_norm": 0.19025862274734395,
      "learning_rate": 6.140918217678038e-09,
      "loss": 0.4309,
      "step": 1458
    },
    {
      "epoch": 2.9594320486815415,
      "grad_norm": 0.19095759274861568,
      "learning_rate": 5.570099876595625e-09,
      "loss": 0.4192,
      "step": 1459
    },
    {
      "epoch": 2.9614604462474645,
      "grad_norm": 0.18250896451135992,
      "learning_rate": 5.027106162742712e-09,
      "loss": 0.4182,
      "step": 1460
    },
    {
      "epoch": 2.9634888438133875,
      "grad_norm": 0.18865329308168316,
      "learning_rate": 4.511940101207812e-09,
      "loss": 0.4309,
      "step": 1461
    },
    {
      "epoch": 2.9655172413793105,
      "grad_norm": 0.19895645690126995,
      "learning_rate": 4.024604562049561e-09,
      "loss": 0.4134,
      "step": 1462
    },
    {
      "epoch": 2.967545638945233,
      "grad_norm": 0.1902048155441232,
      "learning_rate": 3.565102260278397e-09,
      "loss": 0.4383,
      "step": 1463
    },
    {
      "epoch": 2.969574036511156,
      "grad_norm": 0.1950856225445358,
      "learning_rate": 3.133435755842129e-09,
      "loss": 0.4357,
      "step": 1464
    },
    {
      "epoch": 2.971602434077079,
      "grad_norm": 0.1831335830060314,
      "learning_rate": 2.72960745361206e-09,
      "loss": 0.4222,
      "step": 1465
    },
    {
      "epoch": 2.973630831643002,
      "grad_norm": 0.197573053955611,
      "learning_rate": 2.353619603368551e-09,
      "loss": 0.4395,
      "step": 1466
    },
    {
      "epoch": 2.975659229208925,
      "grad_norm": 0.18623911863884565,
      "learning_rate": 2.0054742997893674e-09,
      "loss": 0.4198,
      "step": 1467
    },
    {
      "epoch": 2.977687626774848,
      "grad_norm": 0.1912957354502942,
      "learning_rate": 1.6851734824380184e-09,
      "loss": 0.4238,
      "step": 1468
    },
    {
      "epoch": 2.979716024340771,
      "grad_norm": 0.19044609934187226,
      "learning_rate": 1.392718935752102e-09,
      "loss": 0.4338,
      "step": 1469
    },
    {
      "epoch": 2.9817444219066935,
      "grad_norm": 0.20165285531563396,
      "learning_rate": 1.1281122890355322e-09,
      "loss": 0.4292,
      "step": 1470
    },
    {
      "epoch": 2.9837728194726165,
      "grad_norm": 0.19638825924128533,
      "learning_rate": 8.913550164463269e-10,
      "loss": 0.413,
      "step": 1471
    },
    {
      "epoch": 2.9858012170385395,
      "grad_norm": 0.19168021075678254,
      "learning_rate": 6.824484369899464e-10,
      "loss": 0.4042,
      "step": 1472
    },
    {
      "epoch": 2.9878296146044625,
      "grad_norm": 0.18931224521963913,
      "learning_rate": 5.013937145131875e-10,
      "loss": 0.4073,
      "step": 1473
    },
    {
      "epoch": 2.9898580121703855,
      "grad_norm": 0.1806184973102561,
      "learning_rate": 3.481918576953014e-10,
      "loss": 0.4005,
      "step": 1474
    },
    {
      "epoch": 2.991886409736308,
      "grad_norm": 0.1896458718843294,
      "learning_rate": 2.2284372004410804e-10,
      "loss": 0.4232,
      "step": 1475
    },
    {
      "epoch": 2.9939148073022315,
      "grad_norm": 0.19004539989164482,
      "learning_rate": 1.2534999989044506e-10,
      "loss": 0.4214,
      "step": 1476
    },
    {
      "epoch": 2.995943204868154,
      "grad_norm": 0.19969067065942334,
      "learning_rate": 5.5711240385392106e-11,
      "loss": 0.4295,
      "step": 1477
    },
    {
      "epoch": 2.997971602434077,
      "grad_norm": 0.19274656926441538,
      "learning_rate": 1.3927829494719825e-11,
      "loss": 0.4385,
      "step": 1478
    },
    {
      "epoch": 3.0,
      "grad_norm": 0.20434491810091698,
      "learning_rate": 0.0,
      "loss": 0.4267,
      "step": 1479
    },
    {
      "epoch": 3.0,
      "step": 1479,
      "total_flos": 4652233949970432.0,
      "train_loss": 0.4637200323772882,
      "train_runtime": 19251.6289,
      "train_samples_per_second": 7.369,
      "train_steps_per_second": 0.077
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 1479,
  "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": 4652233949970432.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}