{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 12.987012987012987,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006493506493506494,
      "grad_norm": 33.13806915283203,
      "learning_rate": 9.999999013039593e-05,
      "loss": 5.645,
      "step": 1
    },
    {
      "epoch": 0.012987012987012988,
      "grad_norm": 23.576648712158203,
      "learning_rate": 9.99999605215876e-05,
      "loss": 3.9143,
      "step": 2
    },
    {
      "epoch": 0.01948051948051948,
      "grad_norm": 8.88882827758789,
      "learning_rate": 9.999991117358668e-05,
      "loss": 3.524,
      "step": 3
    },
    {
      "epoch": 0.025974025974025976,
      "grad_norm": 7.060589790344238,
      "learning_rate": 9.999984208641271e-05,
      "loss": 3.3346,
      "step": 4
    },
    {
      "epoch": 0.032467532467532464,
      "grad_norm": 6.797203063964844,
      "learning_rate": 9.999975326009292e-05,
      "loss": 3.097,
      "step": 5
    },
    {
      "epoch": 0.03896103896103896,
      "grad_norm": 5.928432464599609,
      "learning_rate": 9.999964469466236e-05,
      "loss": 3.0141,
      "step": 6
    },
    {
      "epoch": 0.045454545454545456,
      "grad_norm": 5.673449516296387,
      "learning_rate": 9.999951639016395e-05,
      "loss": 3.1291,
      "step": 7
    },
    {
      "epoch": 0.05194805194805195,
      "grad_norm": 5.7318596839904785,
      "learning_rate": 9.99993683466483e-05,
      "loss": 2.9019,
      "step": 8
    },
    {
      "epoch": 0.05844155844155844,
      "grad_norm": 6.2937493324279785,
      "learning_rate": 9.999920056417385e-05,
      "loss": 2.9567,
      "step": 9
    },
    {
      "epoch": 0.06493506493506493,
      "grad_norm": 5.941080570220947,
      "learning_rate": 9.999901304280685e-05,
      "loss": 3.2404,
      "step": 10
    },
    {
      "epoch": 0.07142857142857142,
      "grad_norm": 5.431388854980469,
      "learning_rate": 9.999880578262135e-05,
      "loss": 2.8637,
      "step": 11
    },
    {
      "epoch": 0.07792207792207792,
      "grad_norm": 5.87606143951416,
      "learning_rate": 9.999857878369916e-05,
      "loss": 2.9333,
      "step": 12
    },
    {
      "epoch": 0.08441558441558442,
      "grad_norm": 5.996065616607666,
      "learning_rate": 9.999833204612988e-05,
      "loss": 3.2531,
      "step": 13
    },
    {
      "epoch": 0.09090909090909091,
      "grad_norm": 5.6484832763671875,
      "learning_rate": 9.999806557001093e-05,
      "loss": 2.8224,
      "step": 14
    },
    {
      "epoch": 0.09740259740259741,
      "grad_norm": 5.376333713531494,
      "learning_rate": 9.99977793554475e-05,
      "loss": 2.7981,
      "step": 15
    },
    {
      "epoch": 0.1038961038961039,
      "grad_norm": 5.346993446350098,
      "learning_rate": 9.999747340255259e-05,
      "loss": 2.6952,
      "step": 16
    },
    {
      "epoch": 0.11038961038961038,
      "grad_norm": 5.37903356552124,
      "learning_rate": 9.999714771144701e-05,
      "loss": 2.7209,
      "step": 17
    },
    {
      "epoch": 0.11688311688311688,
      "grad_norm": 5.450172424316406,
      "learning_rate": 9.99968022822593e-05,
      "loss": 2.7421,
      "step": 18
    },
    {
      "epoch": 0.12337662337662338,
      "grad_norm": 5.602138042449951,
      "learning_rate": 9.999643711512586e-05,
      "loss": 2.4858,
      "step": 19
    },
    {
      "epoch": 0.12987012987012986,
      "grad_norm": 5.360761642456055,
      "learning_rate": 9.999605221019081e-05,
      "loss": 2.7158,
      "step": 20
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 6.177570343017578,
      "learning_rate": 9.999564756760615e-05,
      "loss": 2.6152,
      "step": 21
    },
    {
      "epoch": 0.14285714285714285,
      "grad_norm": 6.498959541320801,
      "learning_rate": 9.99952231875316e-05,
      "loss": 2.7611,
      "step": 22
    },
    {
      "epoch": 0.14935064935064934,
      "grad_norm": 5.865311622619629,
      "learning_rate": 9.999477907013473e-05,
      "loss": 2.6758,
      "step": 23
    },
    {
      "epoch": 0.15584415584415584,
      "grad_norm": 6.0329508781433105,
      "learning_rate": 9.999431521559082e-05,
      "loss": 2.6535,
      "step": 24
    },
    {
      "epoch": 0.16233766233766234,
      "grad_norm": 5.814820766448975,
      "learning_rate": 9.999383162408304e-05,
      "loss": 2.6313,
      "step": 25
    },
    {
      "epoch": 0.16883116883116883,
      "grad_norm": 6.224546432495117,
      "learning_rate": 9.999332829580226e-05,
      "loss": 2.627,
      "step": 26
    },
    {
      "epoch": 0.17532467532467533,
      "grad_norm": 5.967427730560303,
      "learning_rate": 9.999280523094724e-05,
      "loss": 2.5997,
      "step": 27
    },
    {
      "epoch": 0.18181818181818182,
      "grad_norm": 5.6386213302612305,
      "learning_rate": 9.999226242972444e-05,
      "loss": 2.4397,
      "step": 28
    },
    {
      "epoch": 0.18831168831168832,
      "grad_norm": 5.742101669311523,
      "learning_rate": 9.999169989234815e-05,
      "loss": 2.6343,
      "step": 29
    },
    {
      "epoch": 0.19480519480519481,
      "grad_norm": 5.685164928436279,
      "learning_rate": 9.999111761904046e-05,
      "loss": 2.3948,
      "step": 30
    },
    {
      "epoch": 0.2012987012987013,
      "grad_norm": 5.332027912139893,
      "learning_rate": 9.999051561003123e-05,
      "loss": 2.4506,
      "step": 31
    },
    {
      "epoch": 0.2077922077922078,
      "grad_norm": 5.785808086395264,
      "learning_rate": 9.998989386555814e-05,
      "loss": 2.7135,
      "step": 32
    },
    {
      "epoch": 0.21428571428571427,
      "grad_norm": 6.223892688751221,
      "learning_rate": 9.998925238586665e-05,
      "loss": 2.5245,
      "step": 33
    },
    {
      "epoch": 0.22077922077922077,
      "grad_norm": 5.84872579574585,
      "learning_rate": 9.998859117121e-05,
      "loss": 2.4981,
      "step": 34
    },
    {
      "epoch": 0.22727272727272727,
      "grad_norm": 5.572268486022949,
      "learning_rate": 9.998791022184922e-05,
      "loss": 2.608,
      "step": 35
    },
    {
      "epoch": 0.23376623376623376,
      "grad_norm": 5.090015411376953,
      "learning_rate": 9.998720953805312e-05,
      "loss": 2.4585,
      "step": 36
    },
    {
      "epoch": 0.24025974025974026,
      "grad_norm": 5.478267192840576,
      "learning_rate": 9.998648912009835e-05,
      "loss": 2.3653,
      "step": 37
    },
    {
      "epoch": 0.24675324675324675,
      "grad_norm": 5.772948741912842,
      "learning_rate": 9.998574896826931e-05,
      "loss": 2.438,
      "step": 38
    },
    {
      "epoch": 0.2532467532467532,
      "grad_norm": 4.907359600067139,
      "learning_rate": 9.998498908285819e-05,
      "loss": 2.2816,
      "step": 39
    },
    {
      "epoch": 0.2597402597402597,
      "grad_norm": 5.5392045974731445,
      "learning_rate": 9.9984209464165e-05,
      "loss": 2.3041,
      "step": 40
    },
    {
      "epoch": 0.2662337662337662,
      "grad_norm": 4.999871253967285,
      "learning_rate": 9.99834101124975e-05,
      "loss": 2.1686,
      "step": 41
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 6.051657676696777,
      "learning_rate": 9.998259102817129e-05,
      "loss": 2.3825,
      "step": 42
    },
    {
      "epoch": 0.2792207792207792,
      "grad_norm": 4.7146687507629395,
      "learning_rate": 9.99817522115097e-05,
      "loss": 2.2417,
      "step": 43
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 4.671162128448486,
      "learning_rate": 9.998089366284391e-05,
      "loss": 2.241,
      "step": 44
    },
    {
      "epoch": 0.2922077922077922,
      "grad_norm": 6.131312847137451,
      "learning_rate": 9.998001538251282e-05,
      "loss": 2.5224,
      "step": 45
    },
    {
      "epoch": 0.2987012987012987,
      "grad_norm": 5.212986469268799,
      "learning_rate": 9.997911737086322e-05,
      "loss": 2.452,
      "step": 46
    },
    {
      "epoch": 0.3051948051948052,
      "grad_norm": 4.900334358215332,
      "learning_rate": 9.997819962824957e-05,
      "loss": 2.4364,
      "step": 47
    },
    {
      "epoch": 0.3116883116883117,
      "grad_norm": 5.208124160766602,
      "learning_rate": 9.997726215503422e-05,
      "loss": 2.3829,
      "step": 48
    },
    {
      "epoch": 0.3181818181818182,
      "grad_norm": 4.627975940704346,
      "learning_rate": 9.997630495158728e-05,
      "loss": 2.0702,
      "step": 49
    },
    {
      "epoch": 0.3246753246753247,
      "grad_norm": 5.099819183349609,
      "learning_rate": 9.997532801828658e-05,
      "loss": 2.3067,
      "step": 50
    },
    {
      "epoch": 0.33116883116883117,
      "grad_norm": 4.694891929626465,
      "learning_rate": 9.997433135551786e-05,
      "loss": 2.3014,
      "step": 51
    },
    {
      "epoch": 0.33766233766233766,
      "grad_norm": 5.41646146774292,
      "learning_rate": 9.997331496367455e-05,
      "loss": 2.4805,
      "step": 52
    },
    {
      "epoch": 0.34415584415584416,
      "grad_norm": 5.233139514923096,
      "learning_rate": 9.997227884315791e-05,
      "loss": 2.2605,
      "step": 53
    },
    {
      "epoch": 0.35064935064935066,
      "grad_norm": 5.671755313873291,
      "learning_rate": 9.9971222994377e-05,
      "loss": 2.3303,
      "step": 54
    },
    {
      "epoch": 0.35714285714285715,
      "grad_norm": 5.070577144622803,
      "learning_rate": 9.997014741774866e-05,
      "loss": 2.3019,
      "step": 55
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 4.925657272338867,
      "learning_rate": 9.996905211369748e-05,
      "loss": 2.3937,
      "step": 56
    },
    {
      "epoch": 0.37012987012987014,
      "grad_norm": 5.051799774169922,
      "learning_rate": 9.996793708265586e-05,
      "loss": 2.358,
      "step": 57
    },
    {
      "epoch": 0.37662337662337664,
      "grad_norm": 4.3391828536987305,
      "learning_rate": 9.996680232506405e-05,
      "loss": 2.0576,
      "step": 58
    },
    {
      "epoch": 0.38311688311688313,
      "grad_norm": 4.852685451507568,
      "learning_rate": 9.996564784137e-05,
      "loss": 2.3678,
      "step": 59
    },
    {
      "epoch": 0.38961038961038963,
      "grad_norm": 4.842132091522217,
      "learning_rate": 9.996447363202946e-05,
      "loss": 2.3493,
      "step": 60
    },
    {
      "epoch": 0.3961038961038961,
      "grad_norm": 4.50392484664917,
      "learning_rate": 9.996327969750605e-05,
      "loss": 2.516,
      "step": 61
    },
    {
      "epoch": 0.4025974025974026,
      "grad_norm": 5.139745712280273,
      "learning_rate": 9.996206603827105e-05,
      "loss": 2.2679,
      "step": 62
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 4.663613319396973,
      "learning_rate": 9.996083265480365e-05,
      "loss": 2.2269,
      "step": 63
    },
    {
      "epoch": 0.4155844155844156,
      "grad_norm": 4.955394744873047,
      "learning_rate": 9.995957954759071e-05,
      "loss": 2.5408,
      "step": 64
    },
    {
      "epoch": 0.42207792207792205,
      "grad_norm": 4.935218334197998,
      "learning_rate": 9.9958306717127e-05,
      "loss": 2.4971,
      "step": 65
    },
    {
      "epoch": 0.42857142857142855,
      "grad_norm": 4.313268661499023,
      "learning_rate": 9.995701416391499e-05,
      "loss": 2.2366,
      "step": 66
    },
    {
      "epoch": 0.43506493506493504,
      "grad_norm": 4.81383752822876,
      "learning_rate": 9.995570188846495e-05,
      "loss": 2.7036,
      "step": 67
    },
    {
      "epoch": 0.44155844155844154,
      "grad_norm": 5.073368549346924,
      "learning_rate": 9.995436989129495e-05,
      "loss": 2.2867,
      "step": 68
    },
    {
      "epoch": 0.44805194805194803,
      "grad_norm": 4.764294147491455,
      "learning_rate": 9.995301817293084e-05,
      "loss": 2.5181,
      "step": 69
    },
    {
      "epoch": 0.45454545454545453,
      "grad_norm": 4.322338104248047,
      "learning_rate": 9.995164673390625e-05,
      "loss": 2.3062,
      "step": 70
    },
    {
      "epoch": 0.461038961038961,
      "grad_norm": 4.6427764892578125,
      "learning_rate": 9.995025557476261e-05,
      "loss": 2.2503,
      "step": 71
    },
    {
      "epoch": 0.4675324675324675,
      "grad_norm": 4.552596092224121,
      "learning_rate": 9.994884469604912e-05,
      "loss": 2.268,
      "step": 72
    },
    {
      "epoch": 0.474025974025974,
      "grad_norm": 3.8830368518829346,
      "learning_rate": 9.99474140983228e-05,
      "loss": 2.0846,
      "step": 73
    },
    {
      "epoch": 0.4805194805194805,
      "grad_norm": 4.275032043457031,
      "learning_rate": 9.994596378214841e-05,
      "loss": 2.2989,
      "step": 74
    },
    {
      "epoch": 0.487012987012987,
      "grad_norm": 4.56163215637207,
      "learning_rate": 9.994449374809851e-05,
      "loss": 2.0471,
      "step": 75
    },
    {
      "epoch": 0.4935064935064935,
      "grad_norm": 4.3414626121521,
      "learning_rate": 9.994300399675342e-05,
      "loss": 2.2404,
      "step": 76
    },
    {
      "epoch": 0.5,
      "grad_norm": 4.33914041519165,
      "learning_rate": 9.994149452870133e-05,
      "loss": 2.419,
      "step": 77
    },
    {
      "epoch": 0.5064935064935064,
      "grad_norm": 4.387986660003662,
      "learning_rate": 9.99399653445381e-05,
      "loss": 2.3415,
      "step": 78
    },
    {
      "epoch": 0.512987012987013,
      "grad_norm": 4.46196985244751,
      "learning_rate": 9.993841644486747e-05,
      "loss": 2.0966,
      "step": 79
    },
    {
      "epoch": 0.5194805194805194,
      "grad_norm": 4.377128601074219,
      "learning_rate": 9.993684783030088e-05,
      "loss": 2.1728,
      "step": 80
    },
    {
      "epoch": 0.525974025974026,
      "grad_norm": 4.3036789894104,
      "learning_rate": 9.99352595014576e-05,
      "loss": 2.3914,
      "step": 81
    },
    {
      "epoch": 0.5324675324675324,
      "grad_norm": 3.62605619430542,
      "learning_rate": 9.993365145896473e-05,
      "loss": 1.768,
      "step": 82
    },
    {
      "epoch": 0.538961038961039,
      "grad_norm": 4.524649143218994,
      "learning_rate": 9.993202370345705e-05,
      "loss": 2.3089,
      "step": 83
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 4.413171291351318,
      "learning_rate": 9.993037623557716e-05,
      "loss": 2.3156,
      "step": 84
    },
    {
      "epoch": 0.551948051948052,
      "grad_norm": 4.606533527374268,
      "learning_rate": 9.992870905597548e-05,
      "loss": 2.4141,
      "step": 85
    },
    {
      "epoch": 0.5584415584415584,
      "grad_norm": 4.309783935546875,
      "learning_rate": 9.99270221653102e-05,
      "loss": 2.1423,
      "step": 86
    },
    {
      "epoch": 0.564935064935065,
      "grad_norm": 4.384764671325684,
      "learning_rate": 9.992531556424726e-05,
      "loss": 2.5358,
      "step": 87
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 4.653176784515381,
      "learning_rate": 9.99235892534604e-05,
      "loss": 2.5152,
      "step": 88
    },
    {
      "epoch": 0.577922077922078,
      "grad_norm": 4.79496955871582,
      "learning_rate": 9.992184323363112e-05,
      "loss": 2.1936,
      "step": 89
    },
    {
      "epoch": 0.5844155844155844,
      "grad_norm": 3.893005847930908,
      "learning_rate": 9.992007750544876e-05,
      "loss": 1.916,
      "step": 90
    },
    {
      "epoch": 0.5909090909090909,
      "grad_norm": 4.456315040588379,
      "learning_rate": 9.991829206961037e-05,
      "loss": 2.6147,
      "step": 91
    },
    {
      "epoch": 0.5974025974025974,
      "grad_norm": 4.070108890533447,
      "learning_rate": 9.991648692682083e-05,
      "loss": 2.3899,
      "step": 92
    },
    {
      "epoch": 0.6038961038961039,
      "grad_norm": 4.310725212097168,
      "learning_rate": 9.991466207779278e-05,
      "loss": 1.9895,
      "step": 93
    },
    {
      "epoch": 0.6103896103896104,
      "grad_norm": 4.388233184814453,
      "learning_rate": 9.991281752324664e-05,
      "loss": 2.43,
      "step": 94
    },
    {
      "epoch": 0.6168831168831169,
      "grad_norm": 4.071033000946045,
      "learning_rate": 9.99109532639106e-05,
      "loss": 2.6234,
      "step": 95
    },
    {
      "epoch": 0.6233766233766234,
      "grad_norm": 4.230044841766357,
      "learning_rate": 9.990906930052064e-05,
      "loss": 2.2542,
      "step": 96
    },
    {
      "epoch": 0.6298701298701299,
      "grad_norm": 4.155112266540527,
      "learning_rate": 9.990716563382055e-05,
      "loss": 2.386,
      "step": 97
    },
    {
      "epoch": 0.6363636363636364,
      "grad_norm": 3.8967134952545166,
      "learning_rate": 9.990524226456182e-05,
      "loss": 2.1869,
      "step": 98
    },
    {
      "epoch": 0.6428571428571429,
      "grad_norm": 3.701253890991211,
      "learning_rate": 9.99032991935038e-05,
      "loss": 2.0778,
      "step": 99
    },
    {
      "epoch": 0.6493506493506493,
      "grad_norm": 3.9027299880981445,
      "learning_rate": 9.990133642141359e-05,
      "loss": 2.3151,
      "step": 100
    },
    {
      "epoch": 0.6558441558441559,
      "grad_norm": 3.9109201431274414,
      "learning_rate": 9.989935394906602e-05,
      "loss": 2.2484,
      "step": 101
    },
    {
      "epoch": 0.6623376623376623,
      "grad_norm": 3.9390170574188232,
      "learning_rate": 9.989735177724378e-05,
      "loss": 2.1411,
      "step": 102
    },
    {
      "epoch": 0.6688311688311688,
      "grad_norm": 3.8148396015167236,
      "learning_rate": 9.989532990673728e-05,
      "loss": 2.238,
      "step": 103
    },
    {
      "epoch": 0.6753246753246753,
      "grad_norm": 3.8671321868896484,
      "learning_rate": 9.989328833834471e-05,
      "loss": 2.1264,
      "step": 104
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 4.0604448318481445,
      "learning_rate": 9.989122707287208e-05,
      "loss": 2.3146,
      "step": 105
    },
    {
      "epoch": 0.6883116883116883,
      "grad_norm": 4.460545539855957,
      "learning_rate": 9.988914611113311e-05,
      "loss": 2.2619,
      "step": 106
    },
    {
      "epoch": 0.6948051948051948,
      "grad_norm": 3.8163511753082275,
      "learning_rate": 9.988704545394936e-05,
      "loss": 2.3224,
      "step": 107
    },
    {
      "epoch": 0.7012987012987013,
      "grad_norm": 3.963921070098877,
      "learning_rate": 9.988492510215011e-05,
      "loss": 2.0558,
      "step": 108
    },
    {
      "epoch": 0.7077922077922078,
      "grad_norm": 3.638936996459961,
      "learning_rate": 9.988278505657247e-05,
      "loss": 2.2582,
      "step": 109
    },
    {
      "epoch": 0.7142857142857143,
      "grad_norm": 3.886962890625,
      "learning_rate": 9.988062531806126e-05,
      "loss": 2.369,
      "step": 110
    },
    {
      "epoch": 0.7207792207792207,
      "grad_norm": 3.7281506061553955,
      "learning_rate": 9.987844588746915e-05,
      "loss": 2.3923,
      "step": 111
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 4.045536041259766,
      "learning_rate": 9.987624676565652e-05,
      "loss": 2.2701,
      "step": 112
    },
    {
      "epoch": 0.7337662337662337,
      "grad_norm": 3.914747953414917,
      "learning_rate": 9.987402795349154e-05,
      "loss": 2.3457,
      "step": 113
    },
    {
      "epoch": 0.7402597402597403,
      "grad_norm": 3.742039203643799,
      "learning_rate": 9.98717894518502e-05,
      "loss": 2.1281,
      "step": 114
    },
    {
      "epoch": 0.7467532467532467,
      "grad_norm": 3.6615986824035645,
      "learning_rate": 9.986953126161619e-05,
      "loss": 2.2539,
      "step": 115
    },
    {
      "epoch": 0.7532467532467533,
      "grad_norm": 4.145374298095703,
      "learning_rate": 9.986725338368102e-05,
      "loss": 2.168,
      "step": 116
    },
    {
      "epoch": 0.7597402597402597,
      "grad_norm": 3.4575271606445312,
      "learning_rate": 9.986495581894395e-05,
      "loss": 2.2219,
      "step": 117
    },
    {
      "epoch": 0.7662337662337663,
      "grad_norm": 3.4362294673919678,
      "learning_rate": 9.986263856831204e-05,
      "loss": 2.1648,
      "step": 118
    },
    {
      "epoch": 0.7727272727272727,
      "grad_norm": 3.6311452388763428,
      "learning_rate": 9.986030163270011e-05,
      "loss": 2.2759,
      "step": 119
    },
    {
      "epoch": 0.7792207792207793,
      "grad_norm": 3.4129133224487305,
      "learning_rate": 9.98579450130307e-05,
      "loss": 1.9268,
      "step": 120
    },
    {
      "epoch": 0.7857142857142857,
      "grad_norm": 3.2100963592529297,
      "learning_rate": 9.98555687102342e-05,
      "loss": 2.0587,
      "step": 121
    },
    {
      "epoch": 0.7922077922077922,
      "grad_norm": 3.662796974182129,
      "learning_rate": 9.985317272524876e-05,
      "loss": 2.0628,
      "step": 122
    },
    {
      "epoch": 0.7987012987012987,
      "grad_norm": 3.4176554679870605,
      "learning_rate": 9.985075705902022e-05,
      "loss": 2.1833,
      "step": 123
    },
    {
      "epoch": 0.8051948051948052,
      "grad_norm": 3.2369673252105713,
      "learning_rate": 9.98483217125023e-05,
      "loss": 2.226,
      "step": 124
    },
    {
      "epoch": 0.8116883116883117,
      "grad_norm": 3.5990474224090576,
      "learning_rate": 9.98458666866564e-05,
      "loss": 2.2601,
      "step": 125
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 2.904496431350708,
      "learning_rate": 9.984339198245175e-05,
      "loss": 1.7978,
      "step": 126
    },
    {
      "epoch": 0.8246753246753247,
      "grad_norm": 3.4239206314086914,
      "learning_rate": 9.98408976008653e-05,
      "loss": 1.8985,
      "step": 127
    },
    {
      "epoch": 0.8311688311688312,
      "grad_norm": 3.4764034748077393,
      "learning_rate": 9.983838354288181e-05,
      "loss": 2.0324,
      "step": 128
    },
    {
      "epoch": 0.8376623376623377,
      "grad_norm": 3.777717351913452,
      "learning_rate": 9.98358498094938e-05,
      "loss": 2.35,
      "step": 129
    },
    {
      "epoch": 0.8441558441558441,
      "grad_norm": 3.3230550289154053,
      "learning_rate": 9.983329640170149e-05,
      "loss": 2.0381,
      "step": 130
    },
    {
      "epoch": 0.8506493506493507,
      "grad_norm": 3.5832202434539795,
      "learning_rate": 9.9830723320513e-05,
      "loss": 2.1329,
      "step": 131
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 3.628079414367676,
      "learning_rate": 9.982813056694412e-05,
      "loss": 2.1468,
      "step": 132
    },
    {
      "epoch": 0.8636363636363636,
      "grad_norm": 3.3164730072021484,
      "learning_rate": 9.982551814201839e-05,
      "loss": 2.1018,
      "step": 133
    },
    {
      "epoch": 0.8701298701298701,
      "grad_norm": 3.202061414718628,
      "learning_rate": 9.98228860467672e-05,
      "loss": 2.1209,
      "step": 134
    },
    {
      "epoch": 0.8766233766233766,
      "grad_norm": 3.5353541374206543,
      "learning_rate": 9.982023428222962e-05,
      "loss": 2.3157,
      "step": 135
    },
    {
      "epoch": 0.8831168831168831,
      "grad_norm": 3.284064292907715,
      "learning_rate": 9.981756284945256e-05,
      "loss": 2.1389,
      "step": 136
    },
    {
      "epoch": 0.8896103896103896,
      "grad_norm": 3.548656463623047,
      "learning_rate": 9.981487174949065e-05,
      "loss": 2.0996,
      "step": 137
    },
    {
      "epoch": 0.8961038961038961,
      "grad_norm": 3.6342179775238037,
      "learning_rate": 9.981216098340629e-05,
      "loss": 2.2534,
      "step": 138
    },
    {
      "epoch": 0.9025974025974026,
      "grad_norm": 3.3490617275238037,
      "learning_rate": 9.980943055226964e-05,
      "loss": 2.0916,
      "step": 139
    },
    {
      "epoch": 0.9090909090909091,
      "grad_norm": 3.469787120819092,
      "learning_rate": 9.980668045715864e-05,
      "loss": 2.0929,
      "step": 140
    },
    {
      "epoch": 0.9155844155844156,
      "grad_norm": 3.651165723800659,
      "learning_rate": 9.980391069915897e-05,
      "loss": 2.3875,
      "step": 141
    },
    {
      "epoch": 0.922077922077922,
      "grad_norm": 3.3916146755218506,
      "learning_rate": 9.980112127936409e-05,
      "loss": 2.2071,
      "step": 142
    },
    {
      "epoch": 0.9285714285714286,
      "grad_norm": 3.484081745147705,
      "learning_rate": 9.979831219887525e-05,
      "loss": 2.2033,
      "step": 143
    },
    {
      "epoch": 0.935064935064935,
      "grad_norm": 3.538928270339966,
      "learning_rate": 9.979548345880141e-05,
      "loss": 2.2221,
      "step": 144
    },
    {
      "epoch": 0.9415584415584416,
      "grad_norm": 3.1394541263580322,
      "learning_rate": 9.979263506025929e-05,
      "loss": 2.1657,
      "step": 145
    },
    {
      "epoch": 0.948051948051948,
      "grad_norm": 3.273376941680908,
      "learning_rate": 9.978976700437342e-05,
      "loss": 2.0307,
      "step": 146
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 3.474287748336792,
      "learning_rate": 9.978687929227606e-05,
      "loss": 2.2569,
      "step": 147
    },
    {
      "epoch": 0.961038961038961,
      "grad_norm": 3.40504789352417,
      "learning_rate": 9.978397192510721e-05,
      "loss": 2.2194,
      "step": 148
    },
    {
      "epoch": 0.9675324675324676,
      "grad_norm": 3.5432350635528564,
      "learning_rate": 9.978104490401467e-05,
      "loss": 2.208,
      "step": 149
    },
    {
      "epoch": 0.974025974025974,
      "grad_norm": 3.4987633228302,
      "learning_rate": 9.977809823015401e-05,
      "loss": 2.108,
      "step": 150
    },
    {
      "epoch": 0.9805194805194806,
      "grad_norm": 2.981435775756836,
      "learning_rate": 9.977513190468848e-05,
      "loss": 1.9615,
      "step": 151
    },
    {
      "epoch": 0.987012987012987,
      "grad_norm": 3.688192129135132,
      "learning_rate": 9.977214592878916e-05,
      "loss": 2.3847,
      "step": 152
    },
    {
      "epoch": 0.9935064935064936,
      "grad_norm": 3.7138729095458984,
      "learning_rate": 9.976914030363487e-05,
      "loss": 2.1349,
      "step": 153
    },
    {
      "epoch": 1.0,
      "grad_norm": 798.8204956054688,
      "learning_rate": 9.976611503041218e-05,
      "loss": 1.9848,
      "step": 154
    },
    {
      "epoch": 1.0064935064935066,
      "grad_norm": 3.2269575595855713,
      "learning_rate": 9.976307011031542e-05,
      "loss": 1.985,
      "step": 155
    },
    {
      "epoch": 1.0129870129870129,
      "grad_norm": 3.4152016639709473,
      "learning_rate": 9.976000554454668e-05,
      "loss": 1.8731,
      "step": 156
    },
    {
      "epoch": 1.0194805194805194,
      "grad_norm": 2.8445022106170654,
      "learning_rate": 9.975692133431579e-05,
      "loss": 1.6258,
      "step": 157
    },
    {
      "epoch": 1.025974025974026,
      "grad_norm": 3.289297580718994,
      "learning_rate": 9.975381748084035e-05,
      "loss": 1.9762,
      "step": 158
    },
    {
      "epoch": 1.0324675324675325,
      "grad_norm": 3.303457260131836,
      "learning_rate": 9.975069398534574e-05,
      "loss": 1.8754,
      "step": 159
    },
    {
      "epoch": 1.0389610389610389,
      "grad_norm": 3.0584588050842285,
      "learning_rate": 9.974755084906502e-05,
      "loss": 1.7928,
      "step": 160
    },
    {
      "epoch": 1.0454545454545454,
      "grad_norm": 3.4114456176757812,
      "learning_rate": 9.974438807323907e-05,
      "loss": 1.9171,
      "step": 161
    },
    {
      "epoch": 1.051948051948052,
      "grad_norm": 3.154326915740967,
      "learning_rate": 9.974120565911652e-05,
      "loss": 1.8924,
      "step": 162
    },
    {
      "epoch": 1.0584415584415585,
      "grad_norm": 4.066158771514893,
      "learning_rate": 9.973800360795372e-05,
      "loss": 2.1918,
      "step": 163
    },
    {
      "epoch": 1.0649350649350648,
      "grad_norm": 3.5308237075805664,
      "learning_rate": 9.97347819210148e-05,
      "loss": 2.0059,
      "step": 164
    },
    {
      "epoch": 1.0714285714285714,
      "grad_norm": 3.352773427963257,
      "learning_rate": 9.973154059957162e-05,
      "loss": 2.0407,
      "step": 165
    },
    {
      "epoch": 1.077922077922078,
      "grad_norm": 3.2745213508605957,
      "learning_rate": 9.972827964490381e-05,
      "loss": 1.9063,
      "step": 166
    },
    {
      "epoch": 1.0844155844155845,
      "grad_norm": 3.0867488384246826,
      "learning_rate": 9.972499905829875e-05,
      "loss": 1.7633,
      "step": 167
    },
    {
      "epoch": 1.0909090909090908,
      "grad_norm": 3.094118118286133,
      "learning_rate": 9.972169884105153e-05,
      "loss": 1.7058,
      "step": 168
    },
    {
      "epoch": 1.0974025974025974,
      "grad_norm": 3.409409284591675,
      "learning_rate": 9.971837899446505e-05,
      "loss": 1.9263,
      "step": 169
    },
    {
      "epoch": 1.103896103896104,
      "grad_norm": 3.7799603939056396,
      "learning_rate": 9.971503951984995e-05,
      "loss": 2.0513,
      "step": 170
    },
    {
      "epoch": 1.1103896103896105,
      "grad_norm": 3.531250238418579,
      "learning_rate": 9.971168041852456e-05,
      "loss": 1.7284,
      "step": 171
    },
    {
      "epoch": 1.1168831168831168,
      "grad_norm": 3.0355734825134277,
      "learning_rate": 9.970830169181505e-05,
      "loss": 1.7999,
      "step": 172
    },
    {
      "epoch": 1.1233766233766234,
      "grad_norm": 3.7033843994140625,
      "learning_rate": 9.970490334105524e-05,
      "loss": 2.1174,
      "step": 173
    },
    {
      "epoch": 1.12987012987013,
      "grad_norm": 3.9485671520233154,
      "learning_rate": 9.970148536758677e-05,
      "loss": 2.0429,
      "step": 174
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 3.2483878135681152,
      "learning_rate": 9.9698047772759e-05,
      "loss": 1.8513,
      "step": 175
    },
    {
      "epoch": 1.1428571428571428,
      "grad_norm": 3.4199907779693604,
      "learning_rate": 9.969459055792903e-05,
      "loss": 1.7395,
      "step": 176
    },
    {
      "epoch": 1.1493506493506493,
      "grad_norm": 3.7996020317077637,
      "learning_rate": 9.969111372446171e-05,
      "loss": 1.9037,
      "step": 177
    },
    {
      "epoch": 1.155844155844156,
      "grad_norm": 3.3956806659698486,
      "learning_rate": 9.968761727372964e-05,
      "loss": 1.6993,
      "step": 178
    },
    {
      "epoch": 1.1623376623376624,
      "grad_norm": 3.3464372158050537,
      "learning_rate": 9.96841012071132e-05,
      "loss": 1.7862,
      "step": 179
    },
    {
      "epoch": 1.1688311688311688,
      "grad_norm": 3.4040322303771973,
      "learning_rate": 9.968056552600043e-05,
      "loss": 2.0307,
      "step": 180
    },
    {
      "epoch": 1.1753246753246753,
      "grad_norm": 3.239704132080078,
      "learning_rate": 9.967701023178717e-05,
      "loss": 1.7453,
      "step": 181
    },
    {
      "epoch": 1.1818181818181819,
      "grad_norm": 3.682248115539551,
      "learning_rate": 9.967343532587702e-05,
      "loss": 1.8286,
      "step": 182
    },
    {
      "epoch": 1.1883116883116882,
      "grad_norm": 3.8498799800872803,
      "learning_rate": 9.966984080968128e-05,
      "loss": 2.072,
      "step": 183
    },
    {
      "epoch": 1.1948051948051948,
      "grad_norm": 3.3957226276397705,
      "learning_rate": 9.9666226684619e-05,
      "loss": 1.8372,
      "step": 184
    },
    {
      "epoch": 1.2012987012987013,
      "grad_norm": 3.5456008911132812,
      "learning_rate": 9.966259295211697e-05,
      "loss": 1.9703,
      "step": 185
    },
    {
      "epoch": 1.2077922077922079,
      "grad_norm": 3.291201591491699,
      "learning_rate": 9.965893961360976e-05,
      "loss": 1.9931,
      "step": 186
    },
    {
      "epoch": 1.2142857142857142,
      "grad_norm": 3.3770711421966553,
      "learning_rate": 9.965526667053963e-05,
      "loss": 1.9248,
      "step": 187
    },
    {
      "epoch": 1.2207792207792207,
      "grad_norm": 3.346139669418335,
      "learning_rate": 9.965157412435663e-05,
      "loss": 1.9951,
      "step": 188
    },
    {
      "epoch": 1.2272727272727273,
      "grad_norm": 2.8916828632354736,
      "learning_rate": 9.964786197651847e-05,
      "loss": 1.7675,
      "step": 189
    },
    {
      "epoch": 1.2337662337662338,
      "grad_norm": 3.3172147274017334,
      "learning_rate": 9.964413022849068e-05,
      "loss": 1.7783,
      "step": 190
    },
    {
      "epoch": 1.2402597402597402,
      "grad_norm": 3.2727859020233154,
      "learning_rate": 9.96403788817465e-05,
      "loss": 2.0687,
      "step": 191
    },
    {
      "epoch": 1.2467532467532467,
      "grad_norm": 2.8867673873901367,
      "learning_rate": 9.963660793776688e-05,
      "loss": 1.467,
      "step": 192
    },
    {
      "epoch": 1.2532467532467533,
      "grad_norm": 3.398193359375,
      "learning_rate": 9.963281739804054e-05,
      "loss": 2.084,
      "step": 193
    },
    {
      "epoch": 1.2597402597402598,
      "grad_norm": 3.4608664512634277,
      "learning_rate": 9.962900726406391e-05,
      "loss": 2.0284,
      "step": 194
    },
    {
      "epoch": 1.2662337662337662,
      "grad_norm": 2.9325497150421143,
      "learning_rate": 9.96251775373412e-05,
      "loss": 1.7167,
      "step": 195
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 3.263169765472412,
      "learning_rate": 9.96213282193843e-05,
      "loss": 1.9843,
      "step": 196
    },
    {
      "epoch": 1.2792207792207793,
      "grad_norm": 3.2453436851501465,
      "learning_rate": 9.961745931171287e-05,
      "loss": 1.804,
      "step": 197
    },
    {
      "epoch": 1.2857142857142856,
      "grad_norm": 3.2929818630218506,
      "learning_rate": 9.96135708158543e-05,
      "loss": 2.0279,
      "step": 198
    },
    {
      "epoch": 1.2922077922077921,
      "grad_norm": 3.565657377243042,
      "learning_rate": 9.96096627333437e-05,
      "loss": 1.9242,
      "step": 199
    },
    {
      "epoch": 1.2987012987012987,
      "grad_norm": 3.3671059608459473,
      "learning_rate": 9.96057350657239e-05,
      "loss": 1.7902,
      "step": 200
    },
    {
      "epoch": 1.3051948051948052,
      "grad_norm": 3.2640137672424316,
      "learning_rate": 9.96017878145455e-05,
      "loss": 1.9485,
      "step": 201
    },
    {
      "epoch": 1.3116883116883118,
      "grad_norm": 3.6283884048461914,
      "learning_rate": 9.959782098136683e-05,
      "loss": 2.204,
      "step": 202
    },
    {
      "epoch": 1.3181818181818181,
      "grad_norm": 3.5066027641296387,
      "learning_rate": 9.959383456775391e-05,
      "loss": 2.0808,
      "step": 203
    },
    {
      "epoch": 1.3246753246753247,
      "grad_norm": 3.4553568363189697,
      "learning_rate": 9.958982857528052e-05,
      "loss": 1.8267,
      "step": 204
    },
    {
      "epoch": 1.3311688311688312,
      "grad_norm": 3.434098482131958,
      "learning_rate": 9.958580300552815e-05,
      "loss": 2.0149,
      "step": 205
    },
    {
      "epoch": 1.3376623376623376,
      "grad_norm": 3.090224504470825,
      "learning_rate": 9.958175786008604e-05,
      "loss": 1.6188,
      "step": 206
    },
    {
      "epoch": 1.344155844155844,
      "grad_norm": 3.104416847229004,
      "learning_rate": 9.957769314055117e-05,
      "loss": 1.8435,
      "step": 207
    },
    {
      "epoch": 1.3506493506493507,
      "grad_norm": 3.1267154216766357,
      "learning_rate": 9.957360884852817e-05,
      "loss": 1.839,
      "step": 208
    },
    {
      "epoch": 1.3571428571428572,
      "grad_norm": 3.384131908416748,
      "learning_rate": 9.956950498562953e-05,
      "loss": 2.0758,
      "step": 209
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 3.1655869483947754,
      "learning_rate": 9.956538155347534e-05,
      "loss": 1.6829,
      "step": 210
    },
    {
      "epoch": 1.37012987012987,
      "grad_norm": 3.373323917388916,
      "learning_rate": 9.956123855369346e-05,
      "loss": 1.8551,
      "step": 211
    },
    {
      "epoch": 1.3766233766233766,
      "grad_norm": 3.1319708824157715,
      "learning_rate": 9.955707598791952e-05,
      "loss": 1.7109,
      "step": 212
    },
    {
      "epoch": 1.3831168831168832,
      "grad_norm": 3.38543963432312,
      "learning_rate": 9.95528938577968e-05,
      "loss": 1.9705,
      "step": 213
    },
    {
      "epoch": 1.3896103896103895,
      "grad_norm": 3.442453145980835,
      "learning_rate": 9.954869216497635e-05,
      "loss": 1.7815,
      "step": 214
    },
    {
      "epoch": 1.396103896103896,
      "grad_norm": 3.13667368888855,
      "learning_rate": 9.954447091111694e-05,
      "loss": 1.9754,
      "step": 215
    },
    {
      "epoch": 1.4025974025974026,
      "grad_norm": 3.447659492492676,
      "learning_rate": 9.954023009788504e-05,
      "loss": 2.269,
      "step": 216
    },
    {
      "epoch": 1.4090909090909092,
      "grad_norm": 3.6094558238983154,
      "learning_rate": 9.953596972695487e-05,
      "loss": 2.0854,
      "step": 217
    },
    {
      "epoch": 1.4155844155844157,
      "grad_norm": 3.1890039443969727,
      "learning_rate": 9.953168980000835e-05,
      "loss": 1.993,
      "step": 218
    },
    {
      "epoch": 1.422077922077922,
      "grad_norm": 3.3757541179656982,
      "learning_rate": 9.952739031873512e-05,
      "loss": 2.0556,
      "step": 219
    },
    {
      "epoch": 1.4285714285714286,
      "grad_norm": 3.047400712966919,
      "learning_rate": 9.952307128483256e-05,
      "loss": 1.9276,
      "step": 220
    },
    {
      "epoch": 1.435064935064935,
      "grad_norm": 3.213284730911255,
      "learning_rate": 9.951873270000576e-05,
      "loss": 2.0443,
      "step": 221
    },
    {
      "epoch": 1.4415584415584415,
      "grad_norm": 3.250971794128418,
      "learning_rate": 9.95143745659675e-05,
      "loss": 1.9392,
      "step": 222
    },
    {
      "epoch": 1.448051948051948,
      "grad_norm": 3.3131914138793945,
      "learning_rate": 9.950999688443833e-05,
      "loss": 1.8581,
      "step": 223
    },
    {
      "epoch": 1.4545454545454546,
      "grad_norm": 2.970548629760742,
      "learning_rate": 9.950559965714648e-05,
      "loss": 1.7365,
      "step": 224
    },
    {
      "epoch": 1.4610389610389611,
      "grad_norm": 3.0069758892059326,
      "learning_rate": 9.950118288582788e-05,
      "loss": 1.4736,
      "step": 225
    },
    {
      "epoch": 1.4675324675324675,
      "grad_norm": 2.8806533813476562,
      "learning_rate": 9.949674657222624e-05,
      "loss": 1.6931,
      "step": 226
    },
    {
      "epoch": 1.474025974025974,
      "grad_norm": 2.8493435382843018,
      "learning_rate": 9.949229071809293e-05,
      "loss": 1.6771,
      "step": 227
    },
    {
      "epoch": 1.4805194805194806,
      "grad_norm": 3.4888033866882324,
      "learning_rate": 9.948781532518705e-05,
      "loss": 1.9593,
      "step": 228
    },
    {
      "epoch": 1.487012987012987,
      "grad_norm": 3.370680332183838,
      "learning_rate": 9.948332039527541e-05,
      "loss": 1.9331,
      "step": 229
    },
    {
      "epoch": 1.4935064935064934,
      "grad_norm": 3.5115890502929688,
      "learning_rate": 9.947880593013255e-05,
      "loss": 2.0011,
      "step": 230
    },
    {
      "epoch": 1.5,
      "grad_norm": 3.091661214828491,
      "learning_rate": 9.947427193154071e-05,
      "loss": 1.8513,
      "step": 231
    },
    {
      "epoch": 1.5064935064935066,
      "grad_norm": 3.062349796295166,
      "learning_rate": 9.946971840128981e-05,
      "loss": 1.7876,
      "step": 232
    },
    {
      "epoch": 1.512987012987013,
      "grad_norm": 3.1313397884368896,
      "learning_rate": 9.946514534117754e-05,
      "loss": 1.8766,
      "step": 233
    },
    {
      "epoch": 1.5194805194805194,
      "grad_norm": 3.2378571033477783,
      "learning_rate": 9.946055275300928e-05,
      "loss": 1.9461,
      "step": 234
    },
    {
      "epoch": 1.525974025974026,
      "grad_norm": 3.385910987854004,
      "learning_rate": 9.945594063859809e-05,
      "loss": 1.8297,
      "step": 235
    },
    {
      "epoch": 1.5324675324675323,
      "grad_norm": 3.524550676345825,
      "learning_rate": 9.945130899976477e-05,
      "loss": 1.9914,
      "step": 236
    },
    {
      "epoch": 1.5389610389610389,
      "grad_norm": 3.3748812675476074,
      "learning_rate": 9.944665783833782e-05,
      "loss": 1.8368,
      "step": 237
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 3.096031904220581,
      "learning_rate": 9.944198715615342e-05,
      "loss": 1.783,
      "step": 238
    },
    {
      "epoch": 1.551948051948052,
      "grad_norm": 3.375197410583496,
      "learning_rate": 9.943729695505552e-05,
      "loss": 2.073,
      "step": 239
    },
    {
      "epoch": 1.5584415584415585,
      "grad_norm": 2.887650966644287,
      "learning_rate": 9.94325872368957e-05,
      "loss": 1.8864,
      "step": 240
    },
    {
      "epoch": 1.564935064935065,
      "grad_norm": 3.060448169708252,
      "learning_rate": 9.942785800353332e-05,
      "loss": 1.7054,
      "step": 241
    },
    {
      "epoch": 1.5714285714285714,
      "grad_norm": 3.2923696041107178,
      "learning_rate": 9.942310925683538e-05,
      "loss": 1.999,
      "step": 242
    },
    {
      "epoch": 1.577922077922078,
      "grad_norm": 2.896110773086548,
      "learning_rate": 9.941834099867659e-05,
      "loss": 1.832,
      "step": 243
    },
    {
      "epoch": 1.5844155844155843,
      "grad_norm": 2.9442901611328125,
      "learning_rate": 9.941355323093943e-05,
      "loss": 2.1465,
      "step": 244
    },
    {
      "epoch": 1.5909090909090908,
      "grad_norm": 2.99206280708313,
      "learning_rate": 9.940874595551404e-05,
      "loss": 1.7449,
      "step": 245
    },
    {
      "epoch": 1.5974025974025974,
      "grad_norm": 3.049651622772217,
      "learning_rate": 9.940391917429818e-05,
      "loss": 1.5516,
      "step": 246
    },
    {
      "epoch": 1.603896103896104,
      "grad_norm": 3.3082385063171387,
      "learning_rate": 9.939907288919747e-05,
      "loss": 2.0167,
      "step": 247
    },
    {
      "epoch": 1.6103896103896105,
      "grad_norm": 3.1149022579193115,
      "learning_rate": 9.939420710212511e-05,
      "loss": 2.0022,
      "step": 248
    },
    {
      "epoch": 1.616883116883117,
      "grad_norm": 2.9124245643615723,
      "learning_rate": 9.938932181500205e-05,
      "loss": 1.9863,
      "step": 249
    },
    {
      "epoch": 1.6233766233766234,
      "grad_norm": 3.180870294570923,
      "learning_rate": 9.938441702975689e-05,
      "loss": 1.9872,
      "step": 250
    },
    {
      "epoch": 1.62987012987013,
      "grad_norm": 3.2378902435302734,
      "learning_rate": 9.9379492748326e-05,
      "loss": 1.9957,
      "step": 251
    },
    {
      "epoch": 1.6363636363636362,
      "grad_norm": 2.9901888370513916,
      "learning_rate": 9.937454897265337e-05,
      "loss": 1.7305,
      "step": 252
    },
    {
      "epoch": 1.6428571428571428,
      "grad_norm": 3.472109794616699,
      "learning_rate": 9.936958570469077e-05,
      "loss": 2.1072,
      "step": 253
    },
    {
      "epoch": 1.6493506493506493,
      "grad_norm": 3.1424944400787354,
      "learning_rate": 9.93646029463976e-05,
      "loss": 1.961,
      "step": 254
    },
    {
      "epoch": 1.655844155844156,
      "grad_norm": 3.177325487136841,
      "learning_rate": 9.935960069974096e-05,
      "loss": 1.7897,
      "step": 255
    },
    {
      "epoch": 1.6623376623376624,
      "grad_norm": 3.207707166671753,
      "learning_rate": 9.935457896669568e-05,
      "loss": 1.8971,
      "step": 256
    },
    {
      "epoch": 1.6688311688311688,
      "grad_norm": 3.142314910888672,
      "learning_rate": 9.934953774924424e-05,
      "loss": 1.8993,
      "step": 257
    },
    {
      "epoch": 1.6753246753246753,
      "grad_norm": 3.1800239086151123,
      "learning_rate": 9.934447704937684e-05,
      "loss": 1.9482,
      "step": 258
    },
    {
      "epoch": 1.6818181818181817,
      "grad_norm": 2.71478271484375,
      "learning_rate": 9.933939686909137e-05,
      "loss": 1.7173,
      "step": 259
    },
    {
      "epoch": 1.6883116883116882,
      "grad_norm": 3.2946617603302,
      "learning_rate": 9.93342972103934e-05,
      "loss": 2.0823,
      "step": 260
    },
    {
      "epoch": 1.6948051948051948,
      "grad_norm": 3.1132423877716064,
      "learning_rate": 9.93291780752962e-05,
      "loss": 1.9475,
      "step": 261
    },
    {
      "epoch": 1.7012987012987013,
      "grad_norm": 3.383772373199463,
      "learning_rate": 9.932403946582072e-05,
      "loss": 2.1187,
      "step": 262
    },
    {
      "epoch": 1.7077922077922079,
      "grad_norm": 3.0686593055725098,
      "learning_rate": 9.931888138399561e-05,
      "loss": 1.7843,
      "step": 263
    },
    {
      "epoch": 1.7142857142857144,
      "grad_norm": 3.1363070011138916,
      "learning_rate": 9.931370383185718e-05,
      "loss": 2.0091,
      "step": 264
    },
    {
      "epoch": 1.7207792207792207,
      "grad_norm": 3.5003199577331543,
      "learning_rate": 9.930850681144945e-05,
      "loss": 2.1529,
      "step": 265
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 3.456496000289917,
      "learning_rate": 9.930329032482413e-05,
      "loss": 2.0149,
      "step": 266
    },
    {
      "epoch": 1.7337662337662336,
      "grad_norm": 2.9732820987701416,
      "learning_rate": 9.92980543740406e-05,
      "loss": 1.7583,
      "step": 267
    },
    {
      "epoch": 1.7402597402597402,
      "grad_norm": 3.0580830574035645,
      "learning_rate": 9.929279896116594e-05,
      "loss": 1.9094,
      "step": 268
    },
    {
      "epoch": 1.7467532467532467,
      "grad_norm": 2.9448652267456055,
      "learning_rate": 9.92875240882749e-05,
      "loss": 1.8479,
      "step": 269
    },
    {
      "epoch": 1.7532467532467533,
      "grad_norm": 3.3782246112823486,
      "learning_rate": 9.928222975744991e-05,
      "loss": 1.7582,
      "step": 270
    },
    {
      "epoch": 1.7597402597402598,
      "grad_norm": 3.2021212577819824,
      "learning_rate": 9.927691597078108e-05,
      "loss": 2.0634,
      "step": 271
    },
    {
      "epoch": 1.7662337662337664,
      "grad_norm": 3.008000135421753,
      "learning_rate": 9.927158273036625e-05,
      "loss": 1.8569,
      "step": 272
    },
    {
      "epoch": 1.7727272727272727,
      "grad_norm": 3.255671501159668,
      "learning_rate": 9.926623003831084e-05,
      "loss": 1.8678,
      "step": 273
    },
    {
      "epoch": 1.7792207792207793,
      "grad_norm": 3.4665205478668213,
      "learning_rate": 9.926085789672806e-05,
      "loss": 2.1407,
      "step": 274
    },
    {
      "epoch": 1.7857142857142856,
      "grad_norm": 3.0679426193237305,
      "learning_rate": 9.92554663077387e-05,
      "loss": 2.0964,
      "step": 275
    },
    {
      "epoch": 1.7922077922077921,
      "grad_norm": 3.3258745670318604,
      "learning_rate": 9.92500552734713e-05,
      "loss": 2.0542,
      "step": 276
    },
    {
      "epoch": 1.7987012987012987,
      "grad_norm": 3.138080596923828,
      "learning_rate": 9.924462479606207e-05,
      "loss": 1.9672,
      "step": 277
    },
    {
      "epoch": 1.8051948051948052,
      "grad_norm": 3.225432872772217,
      "learning_rate": 9.923917487765484e-05,
      "loss": 1.9354,
      "step": 278
    },
    {
      "epoch": 1.8116883116883118,
      "grad_norm": 2.762915849685669,
      "learning_rate": 9.923370552040116e-05,
      "loss": 1.7821,
      "step": 279
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 3.107593059539795,
      "learning_rate": 9.922821672646027e-05,
      "loss": 1.8868,
      "step": 280
    },
    {
      "epoch": 1.8246753246753247,
      "grad_norm": 3.052213668823242,
      "learning_rate": 9.922270849799905e-05,
      "loss": 1.8148,
      "step": 281
    },
    {
      "epoch": 1.8311688311688312,
      "grad_norm": 3.04420804977417,
      "learning_rate": 9.921718083719203e-05,
      "loss": 2.0534,
      "step": 282
    },
    {
      "epoch": 1.8376623376623376,
      "grad_norm": 3.083042860031128,
      "learning_rate": 9.921163374622147e-05,
      "loss": 2.0427,
      "step": 283
    },
    {
      "epoch": 1.844155844155844,
      "grad_norm": 2.7837564945220947,
      "learning_rate": 9.920606722727725e-05,
      "loss": 1.5982,
      "step": 284
    },
    {
      "epoch": 1.8506493506493507,
      "grad_norm": 2.737576961517334,
      "learning_rate": 9.920048128255699e-05,
      "loss": 1.6942,
      "step": 285
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 2.9239234924316406,
      "learning_rate": 9.919487591426591e-05,
      "loss": 1.9588,
      "step": 286
    },
    {
      "epoch": 1.8636363636363638,
      "grad_norm": 2.5686419010162354,
      "learning_rate": 9.918925112461688e-05,
      "loss": 1.798,
      "step": 287
    },
    {
      "epoch": 1.87012987012987,
      "grad_norm": 3.106424570083618,
      "learning_rate": 9.918360691583056e-05,
      "loss": 1.9961,
      "step": 288
    },
    {
      "epoch": 1.8766233766233766,
      "grad_norm": 3.469996452331543,
      "learning_rate": 9.91779432901351e-05,
      "loss": 2.1417,
      "step": 289
    },
    {
      "epoch": 1.883116883116883,
      "grad_norm": 3.0456113815307617,
      "learning_rate": 9.917226024976649e-05,
      "loss": 1.922,
      "step": 290
    },
    {
      "epoch": 1.8896103896103895,
      "grad_norm": 3.158688545227051,
      "learning_rate": 9.916655779696826e-05,
      "loss": 1.8516,
      "step": 291
    },
    {
      "epoch": 1.896103896103896,
      "grad_norm": 2.9286715984344482,
      "learning_rate": 9.916083593399166e-05,
      "loss": 2.0368,
      "step": 292
    },
    {
      "epoch": 1.9025974025974026,
      "grad_norm": 2.7478342056274414,
      "learning_rate": 9.91550946630956e-05,
      "loss": 1.7571,
      "step": 293
    },
    {
      "epoch": 1.9090909090909092,
      "grad_norm": 2.5094292163848877,
      "learning_rate": 9.914933398654663e-05,
      "loss": 1.7007,
      "step": 294
    },
    {
      "epoch": 1.9155844155844157,
      "grad_norm": 2.9641265869140625,
      "learning_rate": 9.914355390661896e-05,
      "loss": 1.7764,
      "step": 295
    },
    {
      "epoch": 1.922077922077922,
      "grad_norm": 3.029686689376831,
      "learning_rate": 9.913775442559452e-05,
      "loss": 1.9938,
      "step": 296
    },
    {
      "epoch": 1.9285714285714286,
      "grad_norm": 3.169807195663452,
      "learning_rate": 9.91319355457628e-05,
      "loss": 1.9341,
      "step": 297
    },
    {
      "epoch": 1.935064935064935,
      "grad_norm": 2.8388686180114746,
      "learning_rate": 9.912609726942103e-05,
      "loss": 1.96,
      "step": 298
    },
    {
      "epoch": 1.9415584415584415,
      "grad_norm": 3.2075181007385254,
      "learning_rate": 9.912023959887408e-05,
      "loss": 2.0209,
      "step": 299
    },
    {
      "epoch": 1.948051948051948,
      "grad_norm": 2.870790481567383,
      "learning_rate": 9.911436253643445e-05,
      "loss": 1.7023,
      "step": 300
    },
    {
      "epoch": 1.9545454545454546,
      "grad_norm": 3.17559814453125,
      "learning_rate": 9.910846608442229e-05,
      "loss": 1.8861,
      "step": 301
    },
    {
      "epoch": 1.9610389610389611,
      "grad_norm": 3.2682406902313232,
      "learning_rate": 9.910255024516546e-05,
      "loss": 1.9807,
      "step": 302
    },
    {
      "epoch": 1.9675324675324677,
      "grad_norm": 3.0713870525360107,
      "learning_rate": 9.909661502099943e-05,
      "loss": 1.7816,
      "step": 303
    },
    {
      "epoch": 1.974025974025974,
      "grad_norm": 2.817713975906372,
      "learning_rate": 9.909066041426733e-05,
      "loss": 1.802,
      "step": 304
    },
    {
      "epoch": 1.9805194805194806,
      "grad_norm": 3.0280492305755615,
      "learning_rate": 9.908468642731995e-05,
      "loss": 2.1874,
      "step": 305
    },
    {
      "epoch": 1.987012987012987,
      "grad_norm": 2.799920082092285,
      "learning_rate": 9.907869306251572e-05,
      "loss": 1.6937,
      "step": 306
    },
    {
      "epoch": 1.9935064935064934,
      "grad_norm": 2.9983012676239014,
      "learning_rate": 9.907268032222071e-05,
      "loss": 1.9305,
      "step": 307
    },
    {
      "epoch": 2.0,
      "grad_norm": 4888.9140625,
      "learning_rate": 9.90666482088087e-05,
      "loss": 1.771,
      "step": 308
    },
    {
      "epoch": 2.0064935064935066,
      "grad_norm": 3.2918434143066406,
      "learning_rate": 9.906059672466101e-05,
      "loss": 1.651,
      "step": 309
    },
    {
      "epoch": 2.012987012987013,
      "grad_norm": 3.1957743167877197,
      "learning_rate": 9.90545258721667e-05,
      "loss": 1.7395,
      "step": 310
    },
    {
      "epoch": 2.0194805194805197,
      "grad_norm": 3.090162515640259,
      "learning_rate": 9.904843565372248e-05,
      "loss": 1.6515,
      "step": 311
    },
    {
      "epoch": 2.0259740259740258,
      "grad_norm": 3.333552598953247,
      "learning_rate": 9.904232607173262e-05,
      "loss": 1.6349,
      "step": 312
    },
    {
      "epoch": 2.0324675324675323,
      "grad_norm": 3.266960382461548,
      "learning_rate": 9.903619712860912e-05,
      "loss": 1.6787,
      "step": 313
    },
    {
      "epoch": 2.038961038961039,
      "grad_norm": 3.407783031463623,
      "learning_rate": 9.903004882677156e-05,
      "loss": 1.5224,
      "step": 314
    },
    {
      "epoch": 2.0454545454545454,
      "grad_norm": 3.5633609294891357,
      "learning_rate": 9.902388116864722e-05,
      "loss": 1.7562,
      "step": 315
    },
    {
      "epoch": 2.051948051948052,
      "grad_norm": 3.1980199813842773,
      "learning_rate": 9.901769415667099e-05,
      "loss": 1.6488,
      "step": 316
    },
    {
      "epoch": 2.0584415584415585,
      "grad_norm": 2.946018695831299,
      "learning_rate": 9.90114877932854e-05,
      "loss": 1.3956,
      "step": 317
    },
    {
      "epoch": 2.064935064935065,
      "grad_norm": 3.0959718227386475,
      "learning_rate": 9.900526208094061e-05,
      "loss": 1.6679,
      "step": 318
    },
    {
      "epoch": 2.0714285714285716,
      "grad_norm": 3.6357126235961914,
      "learning_rate": 9.899901702209445e-05,
      "loss": 1.6934,
      "step": 319
    },
    {
      "epoch": 2.0779220779220777,
      "grad_norm": 3.2969744205474854,
      "learning_rate": 9.899275261921234e-05,
      "loss": 1.7037,
      "step": 320
    },
    {
      "epoch": 2.0844155844155843,
      "grad_norm": 3.406505584716797,
      "learning_rate": 9.898646887476741e-05,
      "loss": 1.6802,
      "step": 321
    },
    {
      "epoch": 2.090909090909091,
      "grad_norm": 3.356058120727539,
      "learning_rate": 9.898016579124037e-05,
      "loss": 1.6779,
      "step": 322
    },
    {
      "epoch": 2.0974025974025974,
      "grad_norm": 3.2132608890533447,
      "learning_rate": 9.897384337111957e-05,
      "loss": 1.6753,
      "step": 323
    },
    {
      "epoch": 2.103896103896104,
      "grad_norm": 3.860316514968872,
      "learning_rate": 9.8967501616901e-05,
      "loss": 1.7702,
      "step": 324
    },
    {
      "epoch": 2.1103896103896105,
      "grad_norm": 3.283928871154785,
      "learning_rate": 9.896114053108829e-05,
      "loss": 1.874,
      "step": 325
    },
    {
      "epoch": 2.116883116883117,
      "grad_norm": 3.458469867706299,
      "learning_rate": 9.895476011619269e-05,
      "loss": 1.6123,
      "step": 326
    },
    {
      "epoch": 2.1233766233766236,
      "grad_norm": 3.1832275390625,
      "learning_rate": 9.89483603747331e-05,
      "loss": 1.5368,
      "step": 327
    },
    {
      "epoch": 2.1298701298701297,
      "grad_norm": 3.380354166030884,
      "learning_rate": 9.894194130923602e-05,
      "loss": 1.8633,
      "step": 328
    },
    {
      "epoch": 2.1363636363636362,
      "grad_norm": 3.0025081634521484,
      "learning_rate": 9.89355029222356e-05,
      "loss": 1.693,
      "step": 329
    },
    {
      "epoch": 2.142857142857143,
      "grad_norm": 3.223750591278076,
      "learning_rate": 9.892904521627361e-05,
      "loss": 1.6421,
      "step": 330
    },
    {
      "epoch": 2.1493506493506493,
      "grad_norm": 2.978912830352783,
      "learning_rate": 9.892256819389947e-05,
      "loss": 1.6442,
      "step": 331
    },
    {
      "epoch": 2.155844155844156,
      "grad_norm": 3.126190185546875,
      "learning_rate": 9.891607185767018e-05,
      "loss": 1.642,
      "step": 332
    },
    {
      "epoch": 2.1623376623376624,
      "grad_norm": 3.196380615234375,
      "learning_rate": 9.890955621015039e-05,
      "loss": 1.8094,
      "step": 333
    },
    {
      "epoch": 2.168831168831169,
      "grad_norm": 3.2463197708129883,
      "learning_rate": 9.890302125391239e-05,
      "loss": 1.7948,
      "step": 334
    },
    {
      "epoch": 2.175324675324675,
      "grad_norm": 3.05271053314209,
      "learning_rate": 9.88964669915361e-05,
      "loss": 1.7538,
      "step": 335
    },
    {
      "epoch": 2.1818181818181817,
      "grad_norm": 2.841326951980591,
      "learning_rate": 9.888989342560899e-05,
      "loss": 1.7039,
      "step": 336
    },
    {
      "epoch": 2.188311688311688,
      "grad_norm": 2.989499807357788,
      "learning_rate": 9.888330055872623e-05,
      "loss": 1.6692,
      "step": 337
    },
    {
      "epoch": 2.1948051948051948,
      "grad_norm": 3.3557872772216797,
      "learning_rate": 9.887668839349057e-05,
      "loss": 1.6721,
      "step": 338
    },
    {
      "epoch": 2.2012987012987013,
      "grad_norm": 3.342548370361328,
      "learning_rate": 9.88700569325124e-05,
      "loss": 1.83,
      "step": 339
    },
    {
      "epoch": 2.207792207792208,
      "grad_norm": 2.9942660331726074,
      "learning_rate": 9.886340617840968e-05,
      "loss": 1.8081,
      "step": 340
    },
    {
      "epoch": 2.2142857142857144,
      "grad_norm": 3.1562952995300293,
      "learning_rate": 9.885673613380806e-05,
      "loss": 1.7479,
      "step": 341
    },
    {
      "epoch": 2.220779220779221,
      "grad_norm": 2.7380170822143555,
      "learning_rate": 9.885004680134076e-05,
      "loss": 1.3577,
      "step": 342
    },
    {
      "epoch": 2.227272727272727,
      "grad_norm": 2.7151942253112793,
      "learning_rate": 9.884333818364861e-05,
      "loss": 1.3956,
      "step": 343
    },
    {
      "epoch": 2.2337662337662336,
      "grad_norm": 3.0294029712677,
      "learning_rate": 9.883661028338008e-05,
      "loss": 1.5567,
      "step": 344
    },
    {
      "epoch": 2.24025974025974,
      "grad_norm": 3.0623183250427246,
      "learning_rate": 9.882986310319124e-05,
      "loss": 1.6769,
      "step": 345
    },
    {
      "epoch": 2.2467532467532467,
      "grad_norm": 3.1786789894104004,
      "learning_rate": 9.882309664574575e-05,
      "loss": 1.595,
      "step": 346
    },
    {
      "epoch": 2.2532467532467533,
      "grad_norm": 3.2232789993286133,
      "learning_rate": 9.881631091371491e-05,
      "loss": 1.6652,
      "step": 347
    },
    {
      "epoch": 2.25974025974026,
      "grad_norm": 2.559070110321045,
      "learning_rate": 9.880950590977765e-05,
      "loss": 1.3795,
      "step": 348
    },
    {
      "epoch": 2.2662337662337664,
      "grad_norm": 3.3524162769317627,
      "learning_rate": 9.880268163662042e-05,
      "loss": 1.8738,
      "step": 349
    },
    {
      "epoch": 2.2727272727272725,
      "grad_norm": 3.0645384788513184,
      "learning_rate": 9.879583809693738e-05,
      "loss": 1.6108,
      "step": 350
    },
    {
      "epoch": 2.279220779220779,
      "grad_norm": 3.547124147415161,
      "learning_rate": 9.878897529343023e-05,
      "loss": 1.9958,
      "step": 351
    },
    {
      "epoch": 2.2857142857142856,
      "grad_norm": 2.875279188156128,
      "learning_rate": 9.87820932288083e-05,
      "loss": 1.5304,
      "step": 352
    },
    {
      "epoch": 2.292207792207792,
      "grad_norm": 3.039005756378174,
      "learning_rate": 9.877519190578852e-05,
      "loss": 1.6787,
      "step": 353
    },
    {
      "epoch": 2.2987012987012987,
      "grad_norm": 3.231783151626587,
      "learning_rate": 9.876827132709544e-05,
      "loss": 1.8304,
      "step": 354
    },
    {
      "epoch": 2.3051948051948052,
      "grad_norm": 3.3293938636779785,
      "learning_rate": 9.876133149546118e-05,
      "loss": 1.8101,
      "step": 355
    },
    {
      "epoch": 2.311688311688312,
      "grad_norm": 3.1368279457092285,
      "learning_rate": 9.875437241362546e-05,
      "loss": 1.7841,
      "step": 356
    },
    {
      "epoch": 2.3181818181818183,
      "grad_norm": 3.4561681747436523,
      "learning_rate": 9.874739408433565e-05,
      "loss": 1.8775,
      "step": 357
    },
    {
      "epoch": 2.324675324675325,
      "grad_norm": 2.352166175842285,
      "learning_rate": 9.874039651034666e-05,
      "loss": 1.4749,
      "step": 358
    },
    {
      "epoch": 2.331168831168831,
      "grad_norm": 2.7010676860809326,
      "learning_rate": 9.873337969442101e-05,
      "loss": 1.6247,
      "step": 359
    },
    {
      "epoch": 2.3376623376623376,
      "grad_norm": 2.420346260070801,
      "learning_rate": 9.872634363932887e-05,
      "loss": 1.2657,
      "step": 360
    },
    {
      "epoch": 2.344155844155844,
      "grad_norm": 3.2130322456359863,
      "learning_rate": 9.871928834784792e-05,
      "loss": 1.796,
      "step": 361
    },
    {
      "epoch": 2.3506493506493507,
      "grad_norm": 2.915154457092285,
      "learning_rate": 9.87122138227635e-05,
      "loss": 1.6665,
      "step": 362
    },
    {
      "epoch": 2.357142857142857,
      "grad_norm": 3.347503185272217,
      "learning_rate": 9.870512006686851e-05,
      "loss": 1.599,
      "step": 363
    },
    {
      "epoch": 2.3636363636363638,
      "grad_norm": 3.082561492919922,
      "learning_rate": 9.869800708296346e-05,
      "loss": 1.9103,
      "step": 364
    },
    {
      "epoch": 2.3701298701298703,
      "grad_norm": 2.9106686115264893,
      "learning_rate": 9.869087487385644e-05,
      "loss": 1.8083,
      "step": 365
    },
    {
      "epoch": 2.3766233766233764,
      "grad_norm": 3.256690740585327,
      "learning_rate": 9.868372344236313e-05,
      "loss": 1.6163,
      "step": 366
    },
    {
      "epoch": 2.383116883116883,
      "grad_norm": 3.0932395458221436,
      "learning_rate": 9.867655279130683e-05,
      "loss": 1.7318,
      "step": 367
    },
    {
      "epoch": 2.3896103896103895,
      "grad_norm": 2.975876569747925,
      "learning_rate": 9.866936292351836e-05,
      "loss": 1.8791,
      "step": 368
    },
    {
      "epoch": 2.396103896103896,
      "grad_norm": 3.221210241317749,
      "learning_rate": 9.866215384183619e-05,
      "loss": 1.8514,
      "step": 369
    },
    {
      "epoch": 2.4025974025974026,
      "grad_norm": 3.188079833984375,
      "learning_rate": 9.865492554910633e-05,
      "loss": 1.8689,
      "step": 370
    },
    {
      "epoch": 2.409090909090909,
      "grad_norm": 3.175229072570801,
      "learning_rate": 9.864767804818243e-05,
      "loss": 1.8805,
      "step": 371
    },
    {
      "epoch": 2.4155844155844157,
      "grad_norm": 3.077758312225342,
      "learning_rate": 9.864041134192563e-05,
      "loss": 1.9167,
      "step": 372
    },
    {
      "epoch": 2.4220779220779223,
      "grad_norm": 2.952587366104126,
      "learning_rate": 9.863312543320477e-05,
      "loss": 1.7636,
      "step": 373
    },
    {
      "epoch": 2.4285714285714284,
      "grad_norm": 2.8241958618164062,
      "learning_rate": 9.86258203248962e-05,
      "loss": 1.583,
      "step": 374
    },
    {
      "epoch": 2.435064935064935,
      "grad_norm": 3.0791385173797607,
      "learning_rate": 9.861849601988383e-05,
      "loss": 1.5679,
      "step": 375
    },
    {
      "epoch": 2.4415584415584415,
      "grad_norm": 3.0636885166168213,
      "learning_rate": 9.861115252105921e-05,
      "loss": 1.6682,
      "step": 376
    },
    {
      "epoch": 2.448051948051948,
      "grad_norm": 2.9893672466278076,
      "learning_rate": 9.860378983132143e-05,
      "loss": 1.8759,
      "step": 377
    },
    {
      "epoch": 2.4545454545454546,
      "grad_norm": 2.585669755935669,
      "learning_rate": 9.859640795357716e-05,
      "loss": 1.602,
      "step": 378
    },
    {
      "epoch": 2.461038961038961,
      "grad_norm": 2.8519511222839355,
      "learning_rate": 9.858900689074064e-05,
      "loss": 1.8531,
      "step": 379
    },
    {
      "epoch": 2.4675324675324677,
      "grad_norm": 2.6581485271453857,
      "learning_rate": 9.85815866457337e-05,
      "loss": 1.6298,
      "step": 380
    },
    {
      "epoch": 2.474025974025974,
      "grad_norm": 2.6713006496429443,
      "learning_rate": 9.857414722148574e-05,
      "loss": 1.7688,
      "step": 381
    },
    {
      "epoch": 2.4805194805194803,
      "grad_norm": 2.8108949661254883,
      "learning_rate": 9.856668862093372e-05,
      "loss": 1.7104,
      "step": 382
    },
    {
      "epoch": 2.487012987012987,
      "grad_norm": 2.4003746509552,
      "learning_rate": 9.855921084702219e-05,
      "loss": 1.4872,
      "step": 383
    },
    {
      "epoch": 2.4935064935064934,
      "grad_norm": 2.9604358673095703,
      "learning_rate": 9.855171390270324e-05,
      "loss": 1.6916,
      "step": 384
    },
    {
      "epoch": 2.5,
      "grad_norm": 2.9553635120391846,
      "learning_rate": 9.854419779093655e-05,
      "loss": 1.65,
      "step": 385
    },
    {
      "epoch": 2.5064935064935066,
      "grad_norm": 2.6719701290130615,
      "learning_rate": 9.853666251468937e-05,
      "loss": 1.5492,
      "step": 386
    },
    {
      "epoch": 2.512987012987013,
      "grad_norm": 2.918327808380127,
      "learning_rate": 9.85291080769365e-05,
      "loss": 1.8282,
      "step": 387
    },
    {
      "epoch": 2.5194805194805197,
      "grad_norm": 3.0886974334716797,
      "learning_rate": 9.852153448066032e-05,
      "loss": 1.7593,
      "step": 388
    },
    {
      "epoch": 2.525974025974026,
      "grad_norm": 2.843238592147827,
      "learning_rate": 9.851394172885074e-05,
      "loss": 1.7201,
      "step": 389
    },
    {
      "epoch": 2.5324675324675323,
      "grad_norm": 2.817976951599121,
      "learning_rate": 9.85063298245053e-05,
      "loss": 1.7933,
      "step": 390
    },
    {
      "epoch": 2.538961038961039,
      "grad_norm": 2.8269622325897217,
      "learning_rate": 9.849869877062902e-05,
      "loss": 1.7015,
      "step": 391
    },
    {
      "epoch": 2.5454545454545454,
      "grad_norm": 2.727435827255249,
      "learning_rate": 9.849104857023455e-05,
      "loss": 1.481,
      "step": 392
    },
    {
      "epoch": 2.551948051948052,
      "grad_norm": 2.7080495357513428,
      "learning_rate": 9.848337922634206e-05,
      "loss": 1.4346,
      "step": 393
    },
    {
      "epoch": 2.5584415584415585,
      "grad_norm": 3.0474908351898193,
      "learning_rate": 9.847569074197926e-05,
      "loss": 1.6259,
      "step": 394
    },
    {
      "epoch": 2.564935064935065,
      "grad_norm": 2.7546396255493164,
      "learning_rate": 9.846798312018146e-05,
      "loss": 1.5626,
      "step": 395
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 2.9604411125183105,
      "learning_rate": 9.846025636399152e-05,
      "loss": 2.0086,
      "step": 396
    },
    {
      "epoch": 2.5779220779220777,
      "grad_norm": 3.0381553173065186,
      "learning_rate": 9.845251047645983e-05,
      "loss": 2.0266,
      "step": 397
    },
    {
      "epoch": 2.5844155844155843,
      "grad_norm": 2.540466070175171,
      "learning_rate": 9.844474546064435e-05,
      "loss": 1.4928,
      "step": 398
    },
    {
      "epoch": 2.590909090909091,
      "grad_norm": 2.5260584354400635,
      "learning_rate": 9.843696131961058e-05,
      "loss": 1.5486,
      "step": 399
    },
    {
      "epoch": 2.5974025974025974,
      "grad_norm": 2.7667641639709473,
      "learning_rate": 9.842915805643155e-05,
      "loss": 1.6139,
      "step": 400
    },
    {
      "epoch": 2.603896103896104,
      "grad_norm": 2.7333874702453613,
      "learning_rate": 9.842133567418792e-05,
      "loss": 1.747,
      "step": 401
    },
    {
      "epoch": 2.6103896103896105,
      "grad_norm": 2.620067596435547,
      "learning_rate": 9.841349417596779e-05,
      "loss": 1.6353,
      "step": 402
    },
    {
      "epoch": 2.616883116883117,
      "grad_norm": 2.723745584487915,
      "learning_rate": 9.84056335648669e-05,
      "loss": 1.5413,
      "step": 403
    },
    {
      "epoch": 2.6233766233766236,
      "grad_norm": 2.789571523666382,
      "learning_rate": 9.839775384398847e-05,
      "loss": 1.9104,
      "step": 404
    },
    {
      "epoch": 2.62987012987013,
      "grad_norm": 2.624312162399292,
      "learning_rate": 9.838985501644328e-05,
      "loss": 1.7595,
      "step": 405
    },
    {
      "epoch": 2.6363636363636362,
      "grad_norm": 2.454328775405884,
      "learning_rate": 9.838193708534968e-05,
      "loss": 1.5174,
      "step": 406
    },
    {
      "epoch": 2.642857142857143,
      "grad_norm": 2.664504051208496,
      "learning_rate": 9.837400005383354e-05,
      "loss": 1.5799,
      "step": 407
    },
    {
      "epoch": 2.6493506493506493,
      "grad_norm": 2.553387403488159,
      "learning_rate": 9.83660439250283e-05,
      "loss": 1.6922,
      "step": 408
    },
    {
      "epoch": 2.655844155844156,
      "grad_norm": 2.5766408443450928,
      "learning_rate": 9.835806870207487e-05,
      "loss": 1.8922,
      "step": 409
    },
    {
      "epoch": 2.6623376623376624,
      "grad_norm": 2.8266332149505615,
      "learning_rate": 9.835007438812177e-05,
      "loss": 1.6223,
      "step": 410
    },
    {
      "epoch": 2.6688311688311686,
      "grad_norm": 2.6766135692596436,
      "learning_rate": 9.834206098632499e-05,
      "loss": 1.6929,
      "step": 411
    },
    {
      "epoch": 2.675324675324675,
      "grad_norm": 2.5043816566467285,
      "learning_rate": 9.833402849984815e-05,
      "loss": 1.5207,
      "step": 412
    },
    {
      "epoch": 2.6818181818181817,
      "grad_norm": 2.6566975116729736,
      "learning_rate": 9.832597693186232e-05,
      "loss": 1.597,
      "step": 413
    },
    {
      "epoch": 2.688311688311688,
      "grad_norm": 2.4868650436401367,
      "learning_rate": 9.831790628554612e-05,
      "loss": 1.6636,
      "step": 414
    },
    {
      "epoch": 2.6948051948051948,
      "grad_norm": 2.5328221321105957,
      "learning_rate": 9.830981656408574e-05,
      "loss": 1.7525,
      "step": 415
    },
    {
      "epoch": 2.7012987012987013,
      "grad_norm": 2.5664961338043213,
      "learning_rate": 9.830170777067485e-05,
      "loss": 1.6539,
      "step": 416
    },
    {
      "epoch": 2.707792207792208,
      "grad_norm": 2.7295408248901367,
      "learning_rate": 9.829357990851468e-05,
      "loss": 1.8107,
      "step": 417
    },
    {
      "epoch": 2.7142857142857144,
      "grad_norm": 2.501190662384033,
      "learning_rate": 9.8285432980814e-05,
      "loss": 1.4882,
      "step": 418
    },
    {
      "epoch": 2.720779220779221,
      "grad_norm": 2.678788185119629,
      "learning_rate": 9.827726699078908e-05,
      "loss": 1.711,
      "step": 419
    },
    {
      "epoch": 2.7272727272727275,
      "grad_norm": 2.4673514366149902,
      "learning_rate": 9.82690819416637e-05,
      "loss": 1.6805,
      "step": 420
    },
    {
      "epoch": 2.7337662337662336,
      "grad_norm": 2.555209159851074,
      "learning_rate": 9.826087783666921e-05,
      "loss": 1.705,
      "step": 421
    },
    {
      "epoch": 2.74025974025974,
      "grad_norm": 2.5114858150482178,
      "learning_rate": 9.825265467904445e-05,
      "loss": 1.6545,
      "step": 422
    },
    {
      "epoch": 2.7467532467532467,
      "grad_norm": 2.7648122310638428,
      "learning_rate": 9.824441247203579e-05,
      "loss": 1.741,
      "step": 423
    },
    {
      "epoch": 2.7532467532467533,
      "grad_norm": 2.949734687805176,
      "learning_rate": 9.823615121889716e-05,
      "loss": 1.8766,
      "step": 424
    },
    {
      "epoch": 2.75974025974026,
      "grad_norm": 2.765125036239624,
      "learning_rate": 9.822787092288991e-05,
      "loss": 1.674,
      "step": 425
    },
    {
      "epoch": 2.7662337662337664,
      "grad_norm": 2.5355770587921143,
      "learning_rate": 9.821957158728301e-05,
      "loss": 1.7228,
      "step": 426
    },
    {
      "epoch": 2.7727272727272725,
      "grad_norm": 2.7192254066467285,
      "learning_rate": 9.82112532153529e-05,
      "loss": 1.8595,
      "step": 427
    },
    {
      "epoch": 2.779220779220779,
      "grad_norm": 2.6829566955566406,
      "learning_rate": 9.820291581038355e-05,
      "loss": 1.758,
      "step": 428
    },
    {
      "epoch": 2.7857142857142856,
      "grad_norm": 2.2499940395355225,
      "learning_rate": 9.819455937566642e-05,
      "loss": 1.427,
      "step": 429
    },
    {
      "epoch": 2.792207792207792,
      "grad_norm": 2.475795030593872,
      "learning_rate": 9.81861839145005e-05,
      "loss": 1.7117,
      "step": 430
    },
    {
      "epoch": 2.7987012987012987,
      "grad_norm": 2.8484835624694824,
      "learning_rate": 9.817778943019228e-05,
      "loss": 1.8138,
      "step": 431
    },
    {
      "epoch": 2.8051948051948052,
      "grad_norm": 2.821648359298706,
      "learning_rate": 9.816937592605579e-05,
      "loss": 1.6179,
      "step": 432
    },
    {
      "epoch": 2.811688311688312,
      "grad_norm": 2.4450900554656982,
      "learning_rate": 9.816094340541256e-05,
      "loss": 1.4323,
      "step": 433
    },
    {
      "epoch": 2.8181818181818183,
      "grad_norm": 2.6112494468688965,
      "learning_rate": 9.815249187159157e-05,
      "loss": 1.8037,
      "step": 434
    },
    {
      "epoch": 2.824675324675325,
      "grad_norm": 2.6128196716308594,
      "learning_rate": 9.814402132792939e-05,
      "loss": 1.6578,
      "step": 435
    },
    {
      "epoch": 2.8311688311688314,
      "grad_norm": 2.560913562774658,
      "learning_rate": 9.813553177777003e-05,
      "loss": 1.8626,
      "step": 436
    },
    {
      "epoch": 2.8376623376623376,
      "grad_norm": 2.2726757526397705,
      "learning_rate": 9.812702322446505e-05,
      "loss": 1.3101,
      "step": 437
    },
    {
      "epoch": 2.844155844155844,
      "grad_norm": 2.4498019218444824,
      "learning_rate": 9.81184956713735e-05,
      "loss": 1.6713,
      "step": 438
    },
    {
      "epoch": 2.8506493506493507,
      "grad_norm": 2.671379804611206,
      "learning_rate": 9.810994912186189e-05,
      "loss": 1.7618,
      "step": 439
    },
    {
      "epoch": 2.857142857142857,
      "grad_norm": 2.9830217361450195,
      "learning_rate": 9.81013835793043e-05,
      "loss": 1.8335,
      "step": 440
    },
    {
      "epoch": 2.8636363636363638,
      "grad_norm": 2.728466749191284,
      "learning_rate": 9.809279904708224e-05,
      "loss": 1.6927,
      "step": 441
    },
    {
      "epoch": 2.87012987012987,
      "grad_norm": 2.514505624771118,
      "learning_rate": 9.808419552858477e-05,
      "loss": 1.7542,
      "step": 442
    },
    {
      "epoch": 2.8766233766233764,
      "grad_norm": 2.831462860107422,
      "learning_rate": 9.80755730272084e-05,
      "loss": 1.6759,
      "step": 443
    },
    {
      "epoch": 2.883116883116883,
      "grad_norm": 2.8181638717651367,
      "learning_rate": 9.806693154635718e-05,
      "loss": 1.8944,
      "step": 444
    },
    {
      "epoch": 2.8896103896103895,
      "grad_norm": 2.7978835105895996,
      "learning_rate": 9.80582710894426e-05,
      "loss": 1.7783,
      "step": 445
    },
    {
      "epoch": 2.896103896103896,
      "grad_norm": 2.823338747024536,
      "learning_rate": 9.80495916598837e-05,
      "loss": 1.7405,
      "step": 446
    },
    {
      "epoch": 2.9025974025974026,
      "grad_norm": 2.665881395339966,
      "learning_rate": 9.804089326110697e-05,
      "loss": 1.6666,
      "step": 447
    },
    {
      "epoch": 2.909090909090909,
      "grad_norm": 2.69315767288208,
      "learning_rate": 9.80321758965464e-05,
      "loss": 1.759,
      "step": 448
    },
    {
      "epoch": 2.9155844155844157,
      "grad_norm": 2.6156058311462402,
      "learning_rate": 9.802343956964348e-05,
      "loss": 1.6456,
      "step": 449
    },
    {
      "epoch": 2.9220779220779223,
      "grad_norm": 2.9610443115234375,
      "learning_rate": 9.801468428384716e-05,
      "loss": 1.8471,
      "step": 450
    },
    {
      "epoch": 2.928571428571429,
      "grad_norm": 2.6673457622528076,
      "learning_rate": 9.800591004261388e-05,
      "loss": 1.6938,
      "step": 451
    },
    {
      "epoch": 2.935064935064935,
      "grad_norm": 2.542597532272339,
      "learning_rate": 9.79971168494076e-05,
      "loss": 1.5416,
      "step": 452
    },
    {
      "epoch": 2.9415584415584415,
      "grad_norm": 2.7253355979919434,
      "learning_rate": 9.79883047076997e-05,
      "loss": 1.7721,
      "step": 453
    },
    {
      "epoch": 2.948051948051948,
      "grad_norm": 2.6640748977661133,
      "learning_rate": 9.797947362096908e-05,
      "loss": 1.9404,
      "step": 454
    },
    {
      "epoch": 2.9545454545454546,
      "grad_norm": 2.5272445678710938,
      "learning_rate": 9.797062359270215e-05,
      "loss": 1.6494,
      "step": 455
    },
    {
      "epoch": 2.961038961038961,
      "grad_norm": 2.679551839828491,
      "learning_rate": 9.796175462639272e-05,
      "loss": 1.5135,
      "step": 456
    },
    {
      "epoch": 2.9675324675324677,
      "grad_norm": 2.832418918609619,
      "learning_rate": 9.795286672554213e-05,
      "loss": 1.7057,
      "step": 457
    },
    {
      "epoch": 2.974025974025974,
      "grad_norm": 2.6729843616485596,
      "learning_rate": 9.794395989365918e-05,
      "loss": 1.7592,
      "step": 458
    },
    {
      "epoch": 2.9805194805194803,
      "grad_norm": 2.605210065841675,
      "learning_rate": 9.793503413426015e-05,
      "loss": 1.7291,
      "step": 459
    },
    {
      "epoch": 2.987012987012987,
      "grad_norm": 2.906468391418457,
      "learning_rate": 9.79260894508688e-05,
      "loss": 1.6162,
      "step": 460
    },
    {
      "epoch": 2.9935064935064934,
      "grad_norm": 2.4108238220214844,
      "learning_rate": 9.791712584701634e-05,
      "loss": 1.5179,
      "step": 461
    },
    {
      "epoch": 3.0,
      "grad_norm": 3461.14306640625,
      "learning_rate": 9.790814332624143e-05,
      "loss": 1.6267,
      "step": 462
    },
    {
      "epoch": 3.0064935064935066,
      "grad_norm": 2.691493511199951,
      "learning_rate": 9.789914189209029e-05,
      "loss": 1.4611,
      "step": 463
    },
    {
      "epoch": 3.012987012987013,
      "grad_norm": 2.130276918411255,
      "learning_rate": 9.789012154811647e-05,
      "loss": 1.0821,
      "step": 464
    },
    {
      "epoch": 3.0194805194805197,
      "grad_norm": 2.652697801589966,
      "learning_rate": 9.788108229788111e-05,
      "loss": 1.4394,
      "step": 465
    },
    {
      "epoch": 3.0259740259740258,
      "grad_norm": 2.6161065101623535,
      "learning_rate": 9.787202414495276e-05,
      "loss": 1.5084,
      "step": 466
    },
    {
      "epoch": 3.0324675324675323,
      "grad_norm": 2.4646644592285156,
      "learning_rate": 9.786294709290741e-05,
      "loss": 1.375,
      "step": 467
    },
    {
      "epoch": 3.038961038961039,
      "grad_norm": 2.6410462856292725,
      "learning_rate": 9.785385114532857e-05,
      "loss": 1.4476,
      "step": 468
    },
    {
      "epoch": 3.0454545454545454,
      "grad_norm": 2.5742151737213135,
      "learning_rate": 9.784473630580713e-05,
      "loss": 1.4318,
      "step": 469
    },
    {
      "epoch": 3.051948051948052,
      "grad_norm": 2.7610526084899902,
      "learning_rate": 9.783560257794154e-05,
      "loss": 1.5746,
      "step": 470
    },
    {
      "epoch": 3.0584415584415585,
      "grad_norm": 2.6887526512145996,
      "learning_rate": 9.78264499653376e-05,
      "loss": 1.5727,
      "step": 471
    },
    {
      "epoch": 3.064935064935065,
      "grad_norm": 2.533906936645508,
      "learning_rate": 9.781727847160865e-05,
      "loss": 1.2746,
      "step": 472
    },
    {
      "epoch": 3.0714285714285716,
      "grad_norm": 2.960747241973877,
      "learning_rate": 9.780808810037543e-05,
      "loss": 1.5435,
      "step": 473
    },
    {
      "epoch": 3.0779220779220777,
      "grad_norm": 2.580984115600586,
      "learning_rate": 9.779887885526615e-05,
      "loss": 1.4734,
      "step": 474
    },
    {
      "epoch": 3.0844155844155843,
      "grad_norm": 2.5366058349609375,
      "learning_rate": 9.778965073991651e-05,
      "loss": 1.2849,
      "step": 475
    },
    {
      "epoch": 3.090909090909091,
      "grad_norm": 2.808206796646118,
      "learning_rate": 9.778040375796959e-05,
      "loss": 1.3771,
      "step": 476
    },
    {
      "epoch": 3.0974025974025974,
      "grad_norm": 3.073631763458252,
      "learning_rate": 9.777113791307598e-05,
      "loss": 1.3693,
      "step": 477
    },
    {
      "epoch": 3.103896103896104,
      "grad_norm": 2.810788869857788,
      "learning_rate": 9.776185320889363e-05,
      "loss": 1.6245,
      "step": 478
    },
    {
      "epoch": 3.1103896103896105,
      "grad_norm": 2.9978721141815186,
      "learning_rate": 9.775254964908807e-05,
      "loss": 1.5436,
      "step": 479
    },
    {
      "epoch": 3.116883116883117,
      "grad_norm": 2.744044065475464,
      "learning_rate": 9.774322723733216e-05,
      "loss": 1.6038,
      "step": 480
    },
    {
      "epoch": 3.1233766233766236,
      "grad_norm": 2.7648191452026367,
      "learning_rate": 9.773388597730623e-05,
      "loss": 1.3454,
      "step": 481
    },
    {
      "epoch": 3.1298701298701297,
      "grad_norm": 2.9355039596557617,
      "learning_rate": 9.772452587269808e-05,
      "loss": 1.5168,
      "step": 482
    },
    {
      "epoch": 3.1363636363636362,
      "grad_norm": 2.6012725830078125,
      "learning_rate": 9.771514692720293e-05,
      "loss": 1.4007,
      "step": 483
    },
    {
      "epoch": 3.142857142857143,
      "grad_norm": 2.895432949066162,
      "learning_rate": 9.770574914452343e-05,
      "loss": 1.5449,
      "step": 484
    },
    {
      "epoch": 3.1493506493506493,
      "grad_norm": 3.0441057682037354,
      "learning_rate": 9.769633252836969e-05,
      "loss": 1.7258,
      "step": 485
    },
    {
      "epoch": 3.155844155844156,
      "grad_norm": 2.6639978885650635,
      "learning_rate": 9.768689708245922e-05,
      "loss": 1.2904,
      "step": 486
    },
    {
      "epoch": 3.1623376623376624,
      "grad_norm": 2.972113847732544,
      "learning_rate": 9.767744281051701e-05,
      "loss": 1.5279,
      "step": 487
    },
    {
      "epoch": 3.168831168831169,
      "grad_norm": 2.602752685546875,
      "learning_rate": 9.766796971627543e-05,
      "loss": 1.4921,
      "step": 488
    },
    {
      "epoch": 3.175324675324675,
      "grad_norm": 2.472797393798828,
      "learning_rate": 9.765847780347432e-05,
      "loss": 1.4381,
      "step": 489
    },
    {
      "epoch": 3.1818181818181817,
      "grad_norm": 2.773179531097412,
      "learning_rate": 9.764896707586096e-05,
      "loss": 1.471,
      "step": 490
    },
    {
      "epoch": 3.188311688311688,
      "grad_norm": 2.8864376544952393,
      "learning_rate": 9.763943753718998e-05,
      "loss": 1.4751,
      "step": 491
    },
    {
      "epoch": 3.1948051948051948,
      "grad_norm": 2.8007543087005615,
      "learning_rate": 9.762988919122355e-05,
      "loss": 1.3703,
      "step": 492
    },
    {
      "epoch": 3.2012987012987013,
      "grad_norm": 2.5158145427703857,
      "learning_rate": 9.762032204173116e-05,
      "loss": 1.2792,
      "step": 493
    },
    {
      "epoch": 3.207792207792208,
      "grad_norm": 2.662209987640381,
      "learning_rate": 9.761073609248981e-05,
      "loss": 1.5026,
      "step": 494
    },
    {
      "epoch": 3.2142857142857144,
      "grad_norm": 2.8098185062408447,
      "learning_rate": 9.760113134728384e-05,
      "loss": 1.5946,
      "step": 495
    },
    {
      "epoch": 3.220779220779221,
      "grad_norm": 2.5345005989074707,
      "learning_rate": 9.759150780990507e-05,
      "loss": 1.4414,
      "step": 496
    },
    {
      "epoch": 3.227272727272727,
      "grad_norm": 2.4922263622283936,
      "learning_rate": 9.758186548415273e-05,
      "loss": 1.4034,
      "step": 497
    },
    {
      "epoch": 3.2337662337662336,
      "grad_norm": 2.613332509994507,
      "learning_rate": 9.757220437383346e-05,
      "loss": 1.5602,
      "step": 498
    },
    {
      "epoch": 3.24025974025974,
      "grad_norm": 2.9960076808929443,
      "learning_rate": 9.756252448276127e-05,
      "loss": 1.4437,
      "step": 499
    },
    {
      "epoch": 3.2467532467532467,
      "grad_norm": 2.567506790161133,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.5557,
      "step": 500
    },
    {
      "epoch": 3.2532467532467533,
      "grad_norm": 2.7986555099487305,
      "learning_rate": 9.754310837365155e-05,
      "loss": 1.458,
      "step": 501
    },
    {
      "epoch": 3.25974025974026,
      "grad_norm": 2.6835241317749023,
      "learning_rate": 9.753337216327917e-05,
      "loss": 1.4782,
      "step": 502
    },
    {
      "epoch": 3.2662337662337664,
      "grad_norm": 2.7536139488220215,
      "learning_rate": 9.752361718748423e-05,
      "loss": 1.5209,
      "step": 503
    },
    {
      "epoch": 3.2727272727272725,
      "grad_norm": 2.784055709838867,
      "learning_rate": 9.751384345011787e-05,
      "loss": 1.5919,
      "step": 504
    },
    {
      "epoch": 3.279220779220779,
      "grad_norm": 2.5098867416381836,
      "learning_rate": 9.750405095503859e-05,
      "loss": 1.5246,
      "step": 505
    },
    {
      "epoch": 3.2857142857142856,
      "grad_norm": 2.397177219390869,
      "learning_rate": 9.749423970611231e-05,
      "loss": 1.3737,
      "step": 506
    },
    {
      "epoch": 3.292207792207792,
      "grad_norm": 2.790895938873291,
      "learning_rate": 9.748440970721236e-05,
      "loss": 1.5794,
      "step": 507
    },
    {
      "epoch": 3.2987012987012987,
      "grad_norm": 2.7680368423461914,
      "learning_rate": 9.747456096221945e-05,
      "loss": 1.5599,
      "step": 508
    },
    {
      "epoch": 3.3051948051948052,
      "grad_norm": 2.880476951599121,
      "learning_rate": 9.746469347502174e-05,
      "loss": 1.6697,
      "step": 509
    },
    {
      "epoch": 3.311688311688312,
      "grad_norm": 2.7532639503479004,
      "learning_rate": 9.745480724951473e-05,
      "loss": 1.5952,
      "step": 510
    },
    {
      "epoch": 3.3181818181818183,
      "grad_norm": 2.685209035873413,
      "learning_rate": 9.744490228960138e-05,
      "loss": 1.5274,
      "step": 511
    },
    {
      "epoch": 3.324675324675325,
      "grad_norm": 2.940244674682617,
      "learning_rate": 9.743497859919196e-05,
      "loss": 1.5009,
      "step": 512
    },
    {
      "epoch": 3.331168831168831,
      "grad_norm": 2.9467644691467285,
      "learning_rate": 9.742503618220422e-05,
      "loss": 1.6353,
      "step": 513
    },
    {
      "epoch": 3.3376623376623376,
      "grad_norm": 2.8322834968566895,
      "learning_rate": 9.741507504256327e-05,
      "loss": 1.653,
      "step": 514
    },
    {
      "epoch": 3.344155844155844,
      "grad_norm": 3.078629493713379,
      "learning_rate": 9.74050951842016e-05,
      "loss": 1.6408,
      "step": 515
    },
    {
      "epoch": 3.3506493506493507,
      "grad_norm": 2.8227908611297607,
      "learning_rate": 9.739509661105912e-05,
      "loss": 1.6134,
      "step": 516
    },
    {
      "epoch": 3.357142857142857,
      "grad_norm": 2.8084356784820557,
      "learning_rate": 9.738507932708307e-05,
      "loss": 1.6953,
      "step": 517
    },
    {
      "epoch": 3.3636363636363638,
      "grad_norm": 2.492335557937622,
      "learning_rate": 9.737504333622813e-05,
      "loss": 1.534,
      "step": 518
    },
    {
      "epoch": 3.3701298701298703,
      "grad_norm": 2.6720969676971436,
      "learning_rate": 9.736498864245638e-05,
      "loss": 1.4763,
      "step": 519
    },
    {
      "epoch": 3.3766233766233764,
      "grad_norm": 2.631711006164551,
      "learning_rate": 9.735491524973722e-05,
      "loss": 1.5452,
      "step": 520
    },
    {
      "epoch": 3.383116883116883,
      "grad_norm": 3.2638559341430664,
      "learning_rate": 9.734482316204747e-05,
      "loss": 1.5939,
      "step": 521
    },
    {
      "epoch": 3.3896103896103895,
      "grad_norm": 2.783334255218506,
      "learning_rate": 9.733471238337136e-05,
      "loss": 1.448,
      "step": 522
    },
    {
      "epoch": 3.396103896103896,
      "grad_norm": 2.8106093406677246,
      "learning_rate": 9.73245829177004e-05,
      "loss": 1.378,
      "step": 523
    },
    {
      "epoch": 3.4025974025974026,
      "grad_norm": 2.796281099319458,
      "learning_rate": 9.73144347690336e-05,
      "loss": 1.5344,
      "step": 524
    },
    {
      "epoch": 3.409090909090909,
      "grad_norm": 2.7715847492218018,
      "learning_rate": 9.730426794137727e-05,
      "loss": 1.3997,
      "step": 525
    },
    {
      "epoch": 3.4155844155844157,
      "grad_norm": 2.7550740242004395,
      "learning_rate": 9.729408243874511e-05,
      "loss": 1.6974,
      "step": 526
    },
    {
      "epoch": 3.4220779220779223,
      "grad_norm": 2.8042678833007812,
      "learning_rate": 9.728387826515819e-05,
      "loss": 1.432,
      "step": 527
    },
    {
      "epoch": 3.4285714285714284,
      "grad_norm": 2.7147512435913086,
      "learning_rate": 9.727365542464497e-05,
      "loss": 1.6309,
      "step": 528
    },
    {
      "epoch": 3.435064935064935,
      "grad_norm": 2.6216771602630615,
      "learning_rate": 9.726341392124127e-05,
      "loss": 1.4175,
      "step": 529
    },
    {
      "epoch": 3.4415584415584415,
      "grad_norm": 2.668849468231201,
      "learning_rate": 9.725315375899024e-05,
      "loss": 1.6321,
      "step": 530
    },
    {
      "epoch": 3.448051948051948,
      "grad_norm": 2.8898913860321045,
      "learning_rate": 9.724287494194247e-05,
      "loss": 1.5724,
      "step": 531
    },
    {
      "epoch": 3.4545454545454546,
      "grad_norm": 2.313223361968994,
      "learning_rate": 9.723257747415584e-05,
      "loss": 1.357,
      "step": 532
    },
    {
      "epoch": 3.461038961038961,
      "grad_norm": 2.627986431121826,
      "learning_rate": 9.722226135969566e-05,
      "loss": 1.5693,
      "step": 533
    },
    {
      "epoch": 3.4675324675324677,
      "grad_norm": 2.4512863159179688,
      "learning_rate": 9.721192660263453e-05,
      "loss": 1.6062,
      "step": 534
    },
    {
      "epoch": 3.474025974025974,
      "grad_norm": 2.7816243171691895,
      "learning_rate": 9.72015732070525e-05,
      "loss": 1.6978,
      "step": 535
    },
    {
      "epoch": 3.4805194805194803,
      "grad_norm": 2.9086880683898926,
      "learning_rate": 9.719120117703687e-05,
      "loss": 1.5865,
      "step": 536
    },
    {
      "epoch": 3.487012987012987,
      "grad_norm": 2.2172327041625977,
      "learning_rate": 9.718081051668239e-05,
      "loss": 1.3989,
      "step": 537
    },
    {
      "epoch": 3.4935064935064934,
      "grad_norm": 2.8032913208007812,
      "learning_rate": 9.717040123009111e-05,
      "loss": 1.6331,
      "step": 538
    },
    {
      "epoch": 3.5,
      "grad_norm": 2.4231390953063965,
      "learning_rate": 9.715997332137248e-05,
      "loss": 1.5196,
      "step": 539
    },
    {
      "epoch": 3.5064935064935066,
      "grad_norm": 2.8790225982666016,
      "learning_rate": 9.714952679464323e-05,
      "loss": 1.5077,
      "step": 540
    },
    {
      "epoch": 3.512987012987013,
      "grad_norm": 2.6722092628479004,
      "learning_rate": 9.713906165402751e-05,
      "loss": 1.6147,
      "step": 541
    },
    {
      "epoch": 3.5194805194805197,
      "grad_norm": 2.7589612007141113,
      "learning_rate": 9.71285779036568e-05,
      "loss": 1.428,
      "step": 542
    },
    {
      "epoch": 3.525974025974026,
      "grad_norm": 2.4026832580566406,
      "learning_rate": 9.71180755476699e-05,
      "loss": 1.2818,
      "step": 543
    },
    {
      "epoch": 3.5324675324675323,
      "grad_norm": 2.9387869834899902,
      "learning_rate": 9.710755459021296e-05,
      "loss": 1.5423,
      "step": 544
    },
    {
      "epoch": 3.538961038961039,
      "grad_norm": 2.7179949283599854,
      "learning_rate": 9.709701503543954e-05,
      "loss": 1.5384,
      "step": 545
    },
    {
      "epoch": 3.5454545454545454,
      "grad_norm": 2.889209508895874,
      "learning_rate": 9.708645688751044e-05,
      "loss": 1.6434,
      "step": 546
    },
    {
      "epoch": 3.551948051948052,
      "grad_norm": 2.6073267459869385,
      "learning_rate": 9.707588015059386e-05,
      "loss": 1.729,
      "step": 547
    },
    {
      "epoch": 3.5584415584415585,
      "grad_norm": 2.7963855266571045,
      "learning_rate": 9.706528482886535e-05,
      "loss": 1.7076,
      "step": 548
    },
    {
      "epoch": 3.564935064935065,
      "grad_norm": 2.9273459911346436,
      "learning_rate": 9.705467092650775e-05,
      "loss": 1.5023,
      "step": 549
    },
    {
      "epoch": 3.571428571428571,
      "grad_norm": 2.3651621341705322,
      "learning_rate": 9.704403844771128e-05,
      "loss": 1.3576,
      "step": 550
    },
    {
      "epoch": 3.5779220779220777,
      "grad_norm": 2.7819669246673584,
      "learning_rate": 9.703338739667346e-05,
      "loss": 1.7064,
      "step": 551
    },
    {
      "epoch": 3.5844155844155843,
      "grad_norm": 2.7776331901550293,
      "learning_rate": 9.702271777759916e-05,
      "loss": 1.6858,
      "step": 552
    },
    {
      "epoch": 3.590909090909091,
      "grad_norm": 2.456737995147705,
      "learning_rate": 9.701202959470058e-05,
      "loss": 1.5001,
      "step": 553
    },
    {
      "epoch": 3.5974025974025974,
      "grad_norm": 2.6922426223754883,
      "learning_rate": 9.700132285219724e-05,
      "loss": 1.6717,
      "step": 554
    },
    {
      "epoch": 3.603896103896104,
      "grad_norm": 2.968545913696289,
      "learning_rate": 9.699059755431598e-05,
      "loss": 1.6364,
      "step": 555
    },
    {
      "epoch": 3.6103896103896105,
      "grad_norm": 2.709141731262207,
      "learning_rate": 9.697985370529101e-05,
      "loss": 1.7164,
      "step": 556
    },
    {
      "epoch": 3.616883116883117,
      "grad_norm": 2.5271835327148438,
      "learning_rate": 9.696909130936382e-05,
      "loss": 1.3943,
      "step": 557
    },
    {
      "epoch": 3.6233766233766236,
      "grad_norm": 2.51953125,
      "learning_rate": 9.695831037078322e-05,
      "loss": 1.5367,
      "step": 558
    },
    {
      "epoch": 3.62987012987013,
      "grad_norm": 2.145310401916504,
      "learning_rate": 9.694751089380536e-05,
      "loss": 1.2712,
      "step": 559
    },
    {
      "epoch": 3.6363636363636362,
      "grad_norm": 2.6842525005340576,
      "learning_rate": 9.693669288269372e-05,
      "loss": 1.5694,
      "step": 560
    },
    {
      "epoch": 3.642857142857143,
      "grad_norm": 2.610161781311035,
      "learning_rate": 9.692585634171905e-05,
      "loss": 1.6228,
      "step": 561
    },
    {
      "epoch": 3.6493506493506493,
      "grad_norm": 2.376155376434326,
      "learning_rate": 9.691500127515945e-05,
      "loss": 1.438,
      "step": 562
    },
    {
      "epoch": 3.655844155844156,
      "grad_norm": 3.225393533706665,
      "learning_rate": 9.690412768730035e-05,
      "loss": 1.7239,
      "step": 563
    },
    {
      "epoch": 3.6623376623376624,
      "grad_norm": 2.3960697650909424,
      "learning_rate": 9.689323558243446e-05,
      "loss": 1.4696,
      "step": 564
    },
    {
      "epoch": 3.6688311688311686,
      "grad_norm": 2.8606388568878174,
      "learning_rate": 9.688232496486178e-05,
      "loss": 1.5147,
      "step": 565
    },
    {
      "epoch": 3.675324675324675,
      "grad_norm": 2.504814863204956,
      "learning_rate": 9.687139583888972e-05,
      "loss": 1.5208,
      "step": 566
    },
    {
      "epoch": 3.6818181818181817,
      "grad_norm": 2.563558340072632,
      "learning_rate": 9.686044820883285e-05,
      "loss": 1.3693,
      "step": 567
    },
    {
      "epoch": 3.688311688311688,
      "grad_norm": 2.8867974281311035,
      "learning_rate": 9.684948207901315e-05,
      "loss": 1.6681,
      "step": 568
    },
    {
      "epoch": 3.6948051948051948,
      "grad_norm": 2.5309319496154785,
      "learning_rate": 9.68384974537599e-05,
      "loss": 1.5182,
      "step": 569
    },
    {
      "epoch": 3.7012987012987013,
      "grad_norm": 2.7548654079437256,
      "learning_rate": 9.682749433740962e-05,
      "loss": 1.7318,
      "step": 570
    },
    {
      "epoch": 3.707792207792208,
      "grad_norm": 2.2360949516296387,
      "learning_rate": 9.681647273430618e-05,
      "loss": 1.223,
      "step": 571
    },
    {
      "epoch": 3.7142857142857144,
      "grad_norm": 2.6854865550994873,
      "learning_rate": 9.680543264880076e-05,
      "loss": 1.6428,
      "step": 572
    },
    {
      "epoch": 3.720779220779221,
      "grad_norm": 2.4563329219818115,
      "learning_rate": 9.679437408525174e-05,
      "loss": 1.5267,
      "step": 573
    },
    {
      "epoch": 3.7272727272727275,
      "grad_norm": 2.505563735961914,
      "learning_rate": 9.678329704802494e-05,
      "loss": 1.4726,
      "step": 574
    },
    {
      "epoch": 3.7337662337662336,
      "grad_norm": 2.7705132961273193,
      "learning_rate": 9.677220154149336e-05,
      "loss": 1.6072,
      "step": 575
    },
    {
      "epoch": 3.74025974025974,
      "grad_norm": 2.2069296836853027,
      "learning_rate": 9.676108757003735e-05,
      "loss": 1.1699,
      "step": 576
    },
    {
      "epoch": 3.7467532467532467,
      "grad_norm": 2.6705710887908936,
      "learning_rate": 9.674995513804452e-05,
      "loss": 1.5892,
      "step": 577
    },
    {
      "epoch": 3.7532467532467533,
      "grad_norm": 2.477724313735962,
      "learning_rate": 9.673880424990977e-05,
      "loss": 1.4687,
      "step": 578
    },
    {
      "epoch": 3.75974025974026,
      "grad_norm": 2.465447425842285,
      "learning_rate": 9.672763491003531e-05,
      "loss": 1.5392,
      "step": 579
    },
    {
      "epoch": 3.7662337662337664,
      "grad_norm": 2.462146282196045,
      "learning_rate": 9.671644712283061e-05,
      "loss": 1.5552,
      "step": 580
    },
    {
      "epoch": 3.7727272727272725,
      "grad_norm": 2.4764628410339355,
      "learning_rate": 9.670524089271242e-05,
      "loss": 1.6651,
      "step": 581
    },
    {
      "epoch": 3.779220779220779,
      "grad_norm": 2.6047585010528564,
      "learning_rate": 9.669401622410482e-05,
      "loss": 1.4174,
      "step": 582
    },
    {
      "epoch": 3.7857142857142856,
      "grad_norm": 2.6200366020202637,
      "learning_rate": 9.668277312143907e-05,
      "loss": 1.5273,
      "step": 583
    },
    {
      "epoch": 3.792207792207792,
      "grad_norm": 3.028610944747925,
      "learning_rate": 9.667151158915382e-05,
      "loss": 1.698,
      "step": 584
    },
    {
      "epoch": 3.7987012987012987,
      "grad_norm": 2.632977247238159,
      "learning_rate": 9.666023163169493e-05,
      "loss": 1.6539,
      "step": 585
    },
    {
      "epoch": 3.8051948051948052,
      "grad_norm": 2.5790421962738037,
      "learning_rate": 9.664893325351555e-05,
      "loss": 1.5818,
      "step": 586
    },
    {
      "epoch": 3.811688311688312,
      "grad_norm": 2.546786069869995,
      "learning_rate": 9.663761645907609e-05,
      "loss": 1.6406,
      "step": 587
    },
    {
      "epoch": 3.8181818181818183,
      "grad_norm": 2.581007957458496,
      "learning_rate": 9.662628125284425e-05,
      "loss": 1.4747,
      "step": 588
    },
    {
      "epoch": 3.824675324675325,
      "grad_norm": 2.5376641750335693,
      "learning_rate": 9.6614927639295e-05,
      "loss": 1.5814,
      "step": 589
    },
    {
      "epoch": 3.8311688311688314,
      "grad_norm": 2.6359288692474365,
      "learning_rate": 9.660355562291055e-05,
      "loss": 1.5488,
      "step": 590
    },
    {
      "epoch": 3.8376623376623376,
      "grad_norm": 2.6092121601104736,
      "learning_rate": 9.65921652081804e-05,
      "loss": 1.4414,
      "step": 591
    },
    {
      "epoch": 3.844155844155844,
      "grad_norm": 2.4724388122558594,
      "learning_rate": 9.65807563996013e-05,
      "loss": 1.4849,
      "step": 592
    },
    {
      "epoch": 3.8506493506493507,
      "grad_norm": 2.303741693496704,
      "learning_rate": 9.656932920167727e-05,
      "loss": 1.6192,
      "step": 593
    },
    {
      "epoch": 3.857142857142857,
      "grad_norm": 2.2833895683288574,
      "learning_rate": 9.65578836189196e-05,
      "loss": 1.389,
      "step": 594
    },
    {
      "epoch": 3.8636363636363638,
      "grad_norm": 2.262861728668213,
      "learning_rate": 9.654641965584678e-05,
      "loss": 1.5055,
      "step": 595
    },
    {
      "epoch": 3.87012987012987,
      "grad_norm": 2.427997350692749,
      "learning_rate": 9.653493731698467e-05,
      "loss": 1.5428,
      "step": 596
    },
    {
      "epoch": 3.8766233766233764,
      "grad_norm": 2.410557270050049,
      "learning_rate": 9.652343660686626e-05,
      "loss": 1.5879,
      "step": 597
    },
    {
      "epoch": 3.883116883116883,
      "grad_norm": 2.3627400398254395,
      "learning_rate": 9.651191753003186e-05,
      "loss": 1.4858,
      "step": 598
    },
    {
      "epoch": 3.8896103896103895,
      "grad_norm": 2.57161283493042,
      "learning_rate": 9.650038009102905e-05,
      "loss": 1.6244,
      "step": 599
    },
    {
      "epoch": 3.896103896103896,
      "grad_norm": 2.7876200675964355,
      "learning_rate": 9.648882429441257e-05,
      "loss": 1.6798,
      "step": 600
    },
    {
      "epoch": 3.9025974025974026,
      "grad_norm": 2.5130650997161865,
      "learning_rate": 9.647725014474452e-05,
      "loss": 1.4941,
      "step": 601
    },
    {
      "epoch": 3.909090909090909,
      "grad_norm": 2.831350326538086,
      "learning_rate": 9.646565764659417e-05,
      "loss": 1.6509,
      "step": 602
    },
    {
      "epoch": 3.9155844155844157,
      "grad_norm": 2.643336534500122,
      "learning_rate": 9.645404680453805e-05,
      "loss": 1.5174,
      "step": 603
    },
    {
      "epoch": 3.9220779220779223,
      "grad_norm": 2.471973180770874,
      "learning_rate": 9.644241762315995e-05,
      "loss": 1.6618,
      "step": 604
    },
    {
      "epoch": 3.928571428571429,
      "grad_norm": 2.491856575012207,
      "learning_rate": 9.643077010705087e-05,
      "loss": 1.4538,
      "step": 605
    },
    {
      "epoch": 3.935064935064935,
      "grad_norm": 2.444056749343872,
      "learning_rate": 9.641910426080908e-05,
      "loss": 1.6486,
      "step": 606
    },
    {
      "epoch": 3.9415584415584415,
      "grad_norm": 2.5531413555145264,
      "learning_rate": 9.640742008904005e-05,
      "loss": 1.5494,
      "step": 607
    },
    {
      "epoch": 3.948051948051948,
      "grad_norm": 2.229311466217041,
      "learning_rate": 9.639571759635654e-05,
      "loss": 1.3976,
      "step": 608
    },
    {
      "epoch": 3.9545454545454546,
      "grad_norm": 2.342977285385132,
      "learning_rate": 9.638399678737848e-05,
      "loss": 1.6625,
      "step": 609
    },
    {
      "epoch": 3.961038961038961,
      "grad_norm": 2.172034978866577,
      "learning_rate": 9.637225766673307e-05,
      "loss": 1.3824,
      "step": 610
    },
    {
      "epoch": 3.9675324675324677,
      "grad_norm": 2.5296199321746826,
      "learning_rate": 9.636050023905473e-05,
      "loss": 1.6165,
      "step": 611
    },
    {
      "epoch": 3.974025974025974,
      "grad_norm": 2.536747455596924,
      "learning_rate": 9.63487245089851e-05,
      "loss": 1.7385,
      "step": 612
    },
    {
      "epoch": 3.9805194805194803,
      "grad_norm": 2.177907705307007,
      "learning_rate": 9.633693048117306e-05,
      "loss": 1.4146,
      "step": 613
    },
    {
      "epoch": 3.987012987012987,
      "grad_norm": 2.305320978164673,
      "learning_rate": 9.632511816027469e-05,
      "loss": 1.4506,
      "step": 614
    },
    {
      "epoch": 3.9935064935064934,
      "grad_norm": 2.482697010040283,
      "learning_rate": 9.631328755095333e-05,
      "loss": 1.5763,
      "step": 615
    },
    {
      "epoch": 4.0,
      "grad_norm": 5.43980073928833,
      "learning_rate": 9.630143865787951e-05,
      "loss": 1.6993,
      "step": 616
    },
    {
      "epoch": 4.0064935064935066,
      "grad_norm": 2.2191660404205322,
      "learning_rate": 9.628957148573098e-05,
      "loss": 1.2847,
      "step": 617
    },
    {
      "epoch": 4.012987012987013,
      "grad_norm": 2.345017671585083,
      "learning_rate": 9.62776860391927e-05,
      "loss": 1.2782,
      "step": 618
    },
    {
      "epoch": 4.01948051948052,
      "grad_norm": 2.5423858165740967,
      "learning_rate": 9.626578232295689e-05,
      "loss": 1.6394,
      "step": 619
    },
    {
      "epoch": 4.025974025974026,
      "grad_norm": 2.2237322330474854,
      "learning_rate": 9.62538603417229e-05,
      "loss": 1.2239,
      "step": 620
    },
    {
      "epoch": 4.032467532467533,
      "grad_norm": 2.228304147720337,
      "learning_rate": 9.62419201001974e-05,
      "loss": 1.3274,
      "step": 621
    },
    {
      "epoch": 4.038961038961039,
      "grad_norm": 2.5428571701049805,
      "learning_rate": 9.622996160309414e-05,
      "loss": 1.2906,
      "step": 622
    },
    {
      "epoch": 4.045454545454546,
      "grad_norm": 2.316067695617676,
      "learning_rate": 9.62179848551342e-05,
      "loss": 1.2922,
      "step": 623
    },
    {
      "epoch": 4.0519480519480515,
      "grad_norm": 2.2502663135528564,
      "learning_rate": 9.620598986104578e-05,
      "loss": 1.2758,
      "step": 624
    },
    {
      "epoch": 4.058441558441558,
      "grad_norm": 2.4511966705322266,
      "learning_rate": 9.619397662556435e-05,
      "loss": 1.3631,
      "step": 625
    },
    {
      "epoch": 4.064935064935065,
      "grad_norm": 2.5539820194244385,
      "learning_rate": 9.61819451534325e-05,
      "loss": 1.3669,
      "step": 626
    },
    {
      "epoch": 4.071428571428571,
      "grad_norm": 2.432616949081421,
      "learning_rate": 9.616989544940009e-05,
      "loss": 1.4103,
      "step": 627
    },
    {
      "epoch": 4.077922077922078,
      "grad_norm": 2.256044387817383,
      "learning_rate": 9.615782751822413e-05,
      "loss": 1.2974,
      "step": 628
    },
    {
      "epoch": 4.084415584415584,
      "grad_norm": 2.4633290767669678,
      "learning_rate": 9.614574136466888e-05,
      "loss": 1.3829,
      "step": 629
    },
    {
      "epoch": 4.090909090909091,
      "grad_norm": 2.8581559658050537,
      "learning_rate": 9.613363699350575e-05,
      "loss": 1.4883,
      "step": 630
    },
    {
      "epoch": 4.097402597402597,
      "grad_norm": 2.2781195640563965,
      "learning_rate": 9.612151440951334e-05,
      "loss": 1.362,
      "step": 631
    },
    {
      "epoch": 4.103896103896104,
      "grad_norm": 2.3701205253601074,
      "learning_rate": 9.610937361747748e-05,
      "loss": 1.2678,
      "step": 632
    },
    {
      "epoch": 4.1103896103896105,
      "grad_norm": 2.225470542907715,
      "learning_rate": 9.609721462219114e-05,
      "loss": 1.2274,
      "step": 633
    },
    {
      "epoch": 4.116883116883117,
      "grad_norm": 2.603336811065674,
      "learning_rate": 9.60850374284545e-05,
      "loss": 1.4969,
      "step": 634
    },
    {
      "epoch": 4.123376623376624,
      "grad_norm": 2.4754090309143066,
      "learning_rate": 9.607284204107493e-05,
      "loss": 1.2693,
      "step": 635
    },
    {
      "epoch": 4.12987012987013,
      "grad_norm": 2.260408639907837,
      "learning_rate": 9.606062846486698e-05,
      "loss": 1.2367,
      "step": 636
    },
    {
      "epoch": 4.136363636363637,
      "grad_norm": 2.4500088691711426,
      "learning_rate": 9.604839670465236e-05,
      "loss": 1.3531,
      "step": 637
    },
    {
      "epoch": 4.142857142857143,
      "grad_norm": 2.718536853790283,
      "learning_rate": 9.603614676526e-05,
      "loss": 1.4347,
      "step": 638
    },
    {
      "epoch": 4.14935064935065,
      "grad_norm": 2.782520294189453,
      "learning_rate": 9.602387865152597e-05,
      "loss": 1.3553,
      "step": 639
    },
    {
      "epoch": 4.1558441558441555,
      "grad_norm": 2.552777051925659,
      "learning_rate": 9.601159236829352e-05,
      "loss": 1.4462,
      "step": 640
    },
    {
      "epoch": 4.162337662337662,
      "grad_norm": 2.2942702770233154,
      "learning_rate": 9.599928792041308e-05,
      "loss": 1.2738,
      "step": 641
    },
    {
      "epoch": 4.1688311688311686,
      "grad_norm": 2.392411231994629,
      "learning_rate": 9.598696531274227e-05,
      "loss": 1.3295,
      "step": 642
    },
    {
      "epoch": 4.175324675324675,
      "grad_norm": 2.619590997695923,
      "learning_rate": 9.597462455014585e-05,
      "loss": 1.3489,
      "step": 643
    },
    {
      "epoch": 4.181818181818182,
      "grad_norm": 2.656822443008423,
      "learning_rate": 9.596226563749575e-05,
      "loss": 1.511,
      "step": 644
    },
    {
      "epoch": 4.188311688311688,
      "grad_norm": 2.4362051486968994,
      "learning_rate": 9.594988857967106e-05,
      "loss": 1.2914,
      "step": 645
    },
    {
      "epoch": 4.194805194805195,
      "grad_norm": 2.1643314361572266,
      "learning_rate": 9.593749338155809e-05,
      "loss": 1.1751,
      "step": 646
    },
    {
      "epoch": 4.201298701298701,
      "grad_norm": 2.725790023803711,
      "learning_rate": 9.592508004805023e-05,
      "loss": 1.5275,
      "step": 647
    },
    {
      "epoch": 4.207792207792208,
      "grad_norm": 2.422140598297119,
      "learning_rate": 9.59126485840481e-05,
      "loss": 1.3667,
      "step": 648
    },
    {
      "epoch": 4.214285714285714,
      "grad_norm": 2.537302017211914,
      "learning_rate": 9.59001989944594e-05,
      "loss": 1.4091,
      "step": 649
    },
    {
      "epoch": 4.220779220779221,
      "grad_norm": 2.6183526515960693,
      "learning_rate": 9.588773128419906e-05,
      "loss": 1.4391,
      "step": 650
    },
    {
      "epoch": 4.2272727272727275,
      "grad_norm": 2.544766426086426,
      "learning_rate": 9.587524545818913e-05,
      "loss": 1.4811,
      "step": 651
    },
    {
      "epoch": 4.233766233766234,
      "grad_norm": 2.4699995517730713,
      "learning_rate": 9.586274152135884e-05,
      "loss": 1.3246,
      "step": 652
    },
    {
      "epoch": 4.240259740259741,
      "grad_norm": 2.2006990909576416,
      "learning_rate": 9.58502194786445e-05,
      "loss": 1.2254,
      "step": 653
    },
    {
      "epoch": 4.246753246753247,
      "grad_norm": 2.6973538398742676,
      "learning_rate": 9.583767933498964e-05,
      "loss": 1.4668,
      "step": 654
    },
    {
      "epoch": 4.253246753246753,
      "grad_norm": 2.2840769290924072,
      "learning_rate": 9.58251210953449e-05,
      "loss": 1.2807,
      "step": 655
    },
    {
      "epoch": 4.259740259740259,
      "grad_norm": 2.4915285110473633,
      "learning_rate": 9.58125447646681e-05,
      "loss": 1.4697,
      "step": 656
    },
    {
      "epoch": 4.266233766233766,
      "grad_norm": 2.6363418102264404,
      "learning_rate": 9.579995034792414e-05,
      "loss": 1.3831,
      "step": 657
    },
    {
      "epoch": 4.2727272727272725,
      "grad_norm": 2.5724446773529053,
      "learning_rate": 9.578733785008513e-05,
      "loss": 1.4606,
      "step": 658
    },
    {
      "epoch": 4.279220779220779,
      "grad_norm": 2.6677348613739014,
      "learning_rate": 9.577470727613025e-05,
      "loss": 1.488,
      "step": 659
    },
    {
      "epoch": 4.285714285714286,
      "grad_norm": 2.5872716903686523,
      "learning_rate": 9.576205863104588e-05,
      "loss": 1.4098,
      "step": 660
    },
    {
      "epoch": 4.292207792207792,
      "grad_norm": 2.636482000350952,
      "learning_rate": 9.57493919198255e-05,
      "loss": 1.4056,
      "step": 661
    },
    {
      "epoch": 4.298701298701299,
      "grad_norm": 2.4787089824676514,
      "learning_rate": 9.573670714746972e-05,
      "loss": 1.3252,
      "step": 662
    },
    {
      "epoch": 4.305194805194805,
      "grad_norm": 2.544703245162964,
      "learning_rate": 9.572400431898627e-05,
      "loss": 1.213,
      "step": 663
    },
    {
      "epoch": 4.311688311688312,
      "grad_norm": 2.3949763774871826,
      "learning_rate": 9.571128343939005e-05,
      "loss": 1.4002,
      "step": 664
    },
    {
      "epoch": 4.318181818181818,
      "grad_norm": 2.5433011054992676,
      "learning_rate": 9.569854451370307e-05,
      "loss": 1.4224,
      "step": 665
    },
    {
      "epoch": 4.324675324675325,
      "grad_norm": 2.485722780227661,
      "learning_rate": 9.568578754695442e-05,
      "loss": 1.3944,
      "step": 666
    },
    {
      "epoch": 4.3311688311688314,
      "grad_norm": 2.7001330852508545,
      "learning_rate": 9.567301254418038e-05,
      "loss": 1.3847,
      "step": 667
    },
    {
      "epoch": 4.337662337662338,
      "grad_norm": 2.6091723442077637,
      "learning_rate": 9.566021951042433e-05,
      "loss": 1.48,
      "step": 668
    },
    {
      "epoch": 4.3441558441558445,
      "grad_norm": 2.37465238571167,
      "learning_rate": 9.56474084507367e-05,
      "loss": 1.2859,
      "step": 669
    },
    {
      "epoch": 4.35064935064935,
      "grad_norm": 2.298042058944702,
      "learning_rate": 9.563457937017515e-05,
      "loss": 1.2833,
      "step": 670
    },
    {
      "epoch": 4.357142857142857,
      "grad_norm": 2.418095827102661,
      "learning_rate": 9.562173227380436e-05,
      "loss": 1.2609,
      "step": 671
    },
    {
      "epoch": 4.363636363636363,
      "grad_norm": 2.5434913635253906,
      "learning_rate": 9.56088671666962e-05,
      "loss": 1.2695,
      "step": 672
    },
    {
      "epoch": 4.37012987012987,
      "grad_norm": 2.607151508331299,
      "learning_rate": 9.559598405392958e-05,
      "loss": 1.4324,
      "step": 673
    },
    {
      "epoch": 4.376623376623376,
      "grad_norm": 2.42596173286438,
      "learning_rate": 9.558308294059054e-05,
      "loss": 1.4373,
      "step": 674
    },
    {
      "epoch": 4.383116883116883,
      "grad_norm": 2.508871078491211,
      "learning_rate": 9.557016383177227e-05,
      "loss": 1.4397,
      "step": 675
    },
    {
      "epoch": 4.3896103896103895,
      "grad_norm": 2.110358715057373,
      "learning_rate": 9.555722673257501e-05,
      "loss": 1.139,
      "step": 676
    },
    {
      "epoch": 4.396103896103896,
      "grad_norm": 2.3169806003570557,
      "learning_rate": 9.554427164810611e-05,
      "loss": 1.231,
      "step": 677
    },
    {
      "epoch": 4.402597402597403,
      "grad_norm": 2.418757200241089,
      "learning_rate": 9.553129858348006e-05,
      "loss": 1.4744,
      "step": 678
    },
    {
      "epoch": 4.409090909090909,
      "grad_norm": 2.496831178665161,
      "learning_rate": 9.55183075438184e-05,
      "loss": 1.4325,
      "step": 679
    },
    {
      "epoch": 4.415584415584416,
      "grad_norm": 2.282007932662964,
      "learning_rate": 9.550529853424979e-05,
      "loss": 1.2713,
      "step": 680
    },
    {
      "epoch": 4.422077922077922,
      "grad_norm": 2.46980619430542,
      "learning_rate": 9.549227155990999e-05,
      "loss": 1.426,
      "step": 681
    },
    {
      "epoch": 4.428571428571429,
      "grad_norm": 2.3510262966156006,
      "learning_rate": 9.547922662594183e-05,
      "loss": 1.2731,
      "step": 682
    },
    {
      "epoch": 4.435064935064935,
      "grad_norm": 2.6384055614471436,
      "learning_rate": 9.546616373749525e-05,
      "loss": 1.4716,
      "step": 683
    },
    {
      "epoch": 4.441558441558442,
      "grad_norm": 2.60732102394104,
      "learning_rate": 9.545308289972728e-05,
      "loss": 1.4314,
      "step": 684
    },
    {
      "epoch": 4.448051948051948,
      "grad_norm": 2.352541923522949,
      "learning_rate": 9.543998411780201e-05,
      "loss": 1.3909,
      "step": 685
    },
    {
      "epoch": 4.454545454545454,
      "grad_norm": 2.4203426837921143,
      "learning_rate": 9.542686739689065e-05,
      "loss": 1.3476,
      "step": 686
    },
    {
      "epoch": 4.461038961038961,
      "grad_norm": 2.63857364654541,
      "learning_rate": 9.541373274217145e-05,
      "loss": 1.4261,
      "step": 687
    },
    {
      "epoch": 4.467532467532467,
      "grad_norm": 2.3278229236602783,
      "learning_rate": 9.540058015882979e-05,
      "loss": 1.2848,
      "step": 688
    },
    {
      "epoch": 4.474025974025974,
      "grad_norm": 2.334977388381958,
      "learning_rate": 9.538740965205808e-05,
      "loss": 1.3625,
      "step": 689
    },
    {
      "epoch": 4.48051948051948,
      "grad_norm": 2.4834561347961426,
      "learning_rate": 9.537422122705585e-05,
      "loss": 1.302,
      "step": 690
    },
    {
      "epoch": 4.487012987012987,
      "grad_norm": 2.0580313205718994,
      "learning_rate": 9.536101488902966e-05,
      "loss": 1.1823,
      "step": 691
    },
    {
      "epoch": 4.4935064935064934,
      "grad_norm": 2.5352213382720947,
      "learning_rate": 9.534779064319318e-05,
      "loss": 1.3543,
      "step": 692
    },
    {
      "epoch": 4.5,
      "grad_norm": 2.424487829208374,
      "learning_rate": 9.533454849476712e-05,
      "loss": 1.4382,
      "step": 693
    },
    {
      "epoch": 4.5064935064935066,
      "grad_norm": 2.486064910888672,
      "learning_rate": 9.532128844897928e-05,
      "loss": 1.3491,
      "step": 694
    },
    {
      "epoch": 4.512987012987013,
      "grad_norm": 2.4160115718841553,
      "learning_rate": 9.530801051106449e-05,
      "loss": 1.401,
      "step": 695
    },
    {
      "epoch": 4.51948051948052,
      "grad_norm": 2.027177095413208,
      "learning_rate": 9.529471468626472e-05,
      "loss": 1.1387,
      "step": 696
    },
    {
      "epoch": 4.525974025974026,
      "grad_norm": 2.3802859783172607,
      "learning_rate": 9.528140097982889e-05,
      "loss": 1.504,
      "step": 697
    },
    {
      "epoch": 4.532467532467533,
      "grad_norm": 2.4218432903289795,
      "learning_rate": 9.526806939701309e-05,
      "loss": 1.3862,
      "step": 698
    },
    {
      "epoch": 4.538961038961039,
      "grad_norm": 2.2717883586883545,
      "learning_rate": 9.52547199430804e-05,
      "loss": 1.1086,
      "step": 699
    },
    {
      "epoch": 4.545454545454545,
      "grad_norm": 2.2099993228912354,
      "learning_rate": 9.524135262330098e-05,
      "loss": 1.3446,
      "step": 700
    },
    {
      "epoch": 4.551948051948052,
      "grad_norm": 2.0593318939208984,
      "learning_rate": 9.522796744295202e-05,
      "loss": 1.0246,
      "step": 701
    },
    {
      "epoch": 4.558441558441558,
      "grad_norm": 2.5161662101745605,
      "learning_rate": 9.52145644073178e-05,
      "loss": 1.51,
      "step": 702
    },
    {
      "epoch": 4.564935064935065,
      "grad_norm": 2.523601531982422,
      "learning_rate": 9.520114352168958e-05,
      "loss": 1.4339,
      "step": 703
    },
    {
      "epoch": 4.571428571428571,
      "grad_norm": 2.5355663299560547,
      "learning_rate": 9.518770479136578e-05,
      "loss": 1.3231,
      "step": 704
    },
    {
      "epoch": 4.577922077922078,
      "grad_norm": 2.8767991065979004,
      "learning_rate": 9.517424822165175e-05,
      "loss": 1.5899,
      "step": 705
    },
    {
      "epoch": 4.584415584415584,
      "grad_norm": 2.6482222080230713,
      "learning_rate": 9.516077381785994e-05,
      "loss": 1.337,
      "step": 706
    },
    {
      "epoch": 4.590909090909091,
      "grad_norm": 2.7159810066223145,
      "learning_rate": 9.514728158530983e-05,
      "loss": 1.4175,
      "step": 707
    },
    {
      "epoch": 4.597402597402597,
      "grad_norm": 2.59430193901062,
      "learning_rate": 9.513377152932796e-05,
      "loss": 1.4217,
      "step": 708
    },
    {
      "epoch": 4.603896103896104,
      "grad_norm": 2.4445676803588867,
      "learning_rate": 9.512024365524787e-05,
      "loss": 1.4127,
      "step": 709
    },
    {
      "epoch": 4.6103896103896105,
      "grad_norm": 2.483778953552246,
      "learning_rate": 9.510669796841014e-05,
      "loss": 1.4122,
      "step": 710
    },
    {
      "epoch": 4.616883116883117,
      "grad_norm": 2.4489808082580566,
      "learning_rate": 9.509313447416242e-05,
      "loss": 1.5403,
      "step": 711
    },
    {
      "epoch": 4.623376623376624,
      "grad_norm": 2.4049267768859863,
      "learning_rate": 9.507955317785934e-05,
      "loss": 1.4104,
      "step": 712
    },
    {
      "epoch": 4.62987012987013,
      "grad_norm": 2.906292200088501,
      "learning_rate": 9.506595408486259e-05,
      "loss": 1.7337,
      "step": 713
    },
    {
      "epoch": 4.636363636363637,
      "grad_norm": 2.2860770225524902,
      "learning_rate": 9.505233720054087e-05,
      "loss": 1.3166,
      "step": 714
    },
    {
      "epoch": 4.642857142857143,
      "grad_norm": 2.660557270050049,
      "learning_rate": 9.503870253026991e-05,
      "loss": 1.4689,
      "step": 715
    },
    {
      "epoch": 4.64935064935065,
      "grad_norm": 2.4646763801574707,
      "learning_rate": 9.502505007943248e-05,
      "loss": 1.4584,
      "step": 716
    },
    {
      "epoch": 4.6558441558441555,
      "grad_norm": 2.287764072418213,
      "learning_rate": 9.501137985341832e-05,
      "loss": 1.4074,
      "step": 717
    },
    {
      "epoch": 4.662337662337662,
      "grad_norm": 3.6270928382873535,
      "learning_rate": 9.499769185762425e-05,
      "loss": 1.5872,
      "step": 718
    },
    {
      "epoch": 4.6688311688311686,
      "grad_norm": 2.5962812900543213,
      "learning_rate": 9.498398609745405e-05,
      "loss": 1.5825,
      "step": 719
    },
    {
      "epoch": 4.675324675324675,
      "grad_norm": 2.382645845413208,
      "learning_rate": 9.497026257831855e-05,
      "loss": 1.3595,
      "step": 720
    },
    {
      "epoch": 4.681818181818182,
      "grad_norm": 2.0103063583374023,
      "learning_rate": 9.49565213056356e-05,
      "loss": 1.1057,
      "step": 721
    },
    {
      "epoch": 4.688311688311688,
      "grad_norm": 2.2708091735839844,
      "learning_rate": 9.494276228482998e-05,
      "loss": 1.3579,
      "step": 722
    },
    {
      "epoch": 4.694805194805195,
      "grad_norm": 2.530606508255005,
      "learning_rate": 9.492898552133358e-05,
      "loss": 1.4606,
      "step": 723
    },
    {
      "epoch": 4.701298701298701,
      "grad_norm": 2.033069372177124,
      "learning_rate": 9.491519102058522e-05,
      "loss": 1.1266,
      "step": 724
    },
    {
      "epoch": 4.707792207792208,
      "grad_norm": 2.533560037612915,
      "learning_rate": 9.490137878803079e-05,
      "loss": 1.5121,
      "step": 725
    },
    {
      "epoch": 4.714285714285714,
      "grad_norm": 2.3064355850219727,
      "learning_rate": 9.48875488291231e-05,
      "loss": 1.3564,
      "step": 726
    },
    {
      "epoch": 4.720779220779221,
      "grad_norm": 2.3468894958496094,
      "learning_rate": 9.487370114932202e-05,
      "loss": 1.4063,
      "step": 727
    },
    {
      "epoch": 4.7272727272727275,
      "grad_norm": 2.2847111225128174,
      "learning_rate": 9.485983575409438e-05,
      "loss": 1.3642,
      "step": 728
    },
    {
      "epoch": 4.733766233766234,
      "grad_norm": 2.188500165939331,
      "learning_rate": 9.484595264891402e-05,
      "loss": 1.2824,
      "step": 729
    },
    {
      "epoch": 4.740259740259741,
      "grad_norm": 2.442934036254883,
      "learning_rate": 9.483205183926181e-05,
      "loss": 1.4042,
      "step": 730
    },
    {
      "epoch": 4.746753246753247,
      "grad_norm": 2.4196629524230957,
      "learning_rate": 9.48181333306255e-05,
      "loss": 1.4626,
      "step": 731
    },
    {
      "epoch": 4.753246753246753,
      "grad_norm": 2.290989637374878,
      "learning_rate": 9.480419712849995e-05,
      "loss": 1.2912,
      "step": 732
    },
    {
      "epoch": 4.759740259740259,
      "grad_norm": 2.430495262145996,
      "learning_rate": 9.479024323838693e-05,
      "loss": 1.4117,
      "step": 733
    },
    {
      "epoch": 4.766233766233766,
      "grad_norm": 2.0131633281707764,
      "learning_rate": 9.477627166579522e-05,
      "loss": 1.1689,
      "step": 734
    },
    {
      "epoch": 4.7727272727272725,
      "grad_norm": 2.582155466079712,
      "learning_rate": 9.476228241624059e-05,
      "loss": 1.6276,
      "step": 735
    },
    {
      "epoch": 4.779220779220779,
      "grad_norm": 2.3215701580047607,
      "learning_rate": 9.474827549524574e-05,
      "loss": 1.3326,
      "step": 736
    },
    {
      "epoch": 4.785714285714286,
      "grad_norm": 2.2553138732910156,
      "learning_rate": 9.473425090834041e-05,
      "loss": 1.3568,
      "step": 737
    },
    {
      "epoch": 4.792207792207792,
      "grad_norm": 2.5642342567443848,
      "learning_rate": 9.472020866106128e-05,
      "loss": 1.4779,
      "step": 738
    },
    {
      "epoch": 4.798701298701299,
      "grad_norm": 2.465707778930664,
      "learning_rate": 9.470614875895201e-05,
      "loss": 1.4478,
      "step": 739
    },
    {
      "epoch": 4.805194805194805,
      "grad_norm": 2.395329236984253,
      "learning_rate": 9.46920712075632e-05,
      "loss": 1.3339,
      "step": 740
    },
    {
      "epoch": 4.811688311688312,
      "grad_norm": 2.5474603176116943,
      "learning_rate": 9.467797601245246e-05,
      "loss": 1.4824,
      "step": 741
    },
    {
      "epoch": 4.818181818181818,
      "grad_norm": 2.498077154159546,
      "learning_rate": 9.466386317918436e-05,
      "loss": 1.4558,
      "step": 742
    },
    {
      "epoch": 4.824675324675325,
      "grad_norm": 2.258089065551758,
      "learning_rate": 9.464973271333042e-05,
      "loss": 1.3571,
      "step": 743
    },
    {
      "epoch": 4.8311688311688314,
      "grad_norm": 2.5701329708099365,
      "learning_rate": 9.463558462046912e-05,
      "loss": 1.5302,
      "step": 744
    },
    {
      "epoch": 4.837662337662338,
      "grad_norm": 2.347843885421753,
      "learning_rate": 9.46214189061859e-05,
      "loss": 1.3368,
      "step": 745
    },
    {
      "epoch": 4.8441558441558445,
      "grad_norm": 2.3211004734039307,
      "learning_rate": 9.460723557607316e-05,
      "loss": 1.3966,
      "step": 746
    },
    {
      "epoch": 4.85064935064935,
      "grad_norm": 2.381417989730835,
      "learning_rate": 9.459303463573026e-05,
      "loss": 1.267,
      "step": 747
    },
    {
      "epoch": 4.857142857142857,
      "grad_norm": 2.3821098804473877,
      "learning_rate": 9.457881609076352e-05,
      "loss": 1.4147,
      "step": 748
    },
    {
      "epoch": 4.863636363636363,
      "grad_norm": 2.3875110149383545,
      "learning_rate": 9.456457994678616e-05,
      "loss": 1.3116,
      "step": 749
    },
    {
      "epoch": 4.87012987012987,
      "grad_norm": 2.3099076747894287,
      "learning_rate": 9.45503262094184e-05,
      "loss": 1.2902,
      "step": 750
    },
    {
      "epoch": 4.876623376623376,
      "grad_norm": 2.0546610355377197,
      "learning_rate": 9.45360548842874e-05,
      "loss": 1.1813,
      "step": 751
    },
    {
      "epoch": 4.883116883116883,
      "grad_norm": 2.4475390911102295,
      "learning_rate": 9.452176597702725e-05,
      "loss": 1.5366,
      "step": 752
    },
    {
      "epoch": 4.8896103896103895,
      "grad_norm": 2.145540237426758,
      "learning_rate": 9.450745949327896e-05,
      "loss": 1.1983,
      "step": 753
    },
    {
      "epoch": 4.896103896103896,
      "grad_norm": 2.694192886352539,
      "learning_rate": 9.449313543869055e-05,
      "loss": 1.5095,
      "step": 754
    },
    {
      "epoch": 4.902597402597403,
      "grad_norm": 2.1562998294830322,
      "learning_rate": 9.447879381891692e-05,
      "loss": 1.1382,
      "step": 755
    },
    {
      "epoch": 4.909090909090909,
      "grad_norm": 2.4446427822113037,
      "learning_rate": 9.446443463961986e-05,
      "loss": 1.4053,
      "step": 756
    },
    {
      "epoch": 4.915584415584416,
      "grad_norm": 2.055360794067383,
      "learning_rate": 9.445005790646819e-05,
      "loss": 1.3194,
      "step": 757
    },
    {
      "epoch": 4.922077922077922,
      "grad_norm": 2.397583246231079,
      "learning_rate": 9.443566362513763e-05,
      "loss": 1.5682,
      "step": 758
    },
    {
      "epoch": 4.928571428571429,
      "grad_norm": 2.185490846633911,
      "learning_rate": 9.442125180131078e-05,
      "loss": 1.4359,
      "step": 759
    },
    {
      "epoch": 4.935064935064935,
      "grad_norm": 2.365079402923584,
      "learning_rate": 9.440682244067724e-05,
      "loss": 1.4141,
      "step": 760
    },
    {
      "epoch": 4.941558441558442,
      "grad_norm": 2.510624408721924,
      "learning_rate": 9.439237554893344e-05,
      "loss": 1.5375,
      "step": 761
    },
    {
      "epoch": 4.948051948051948,
      "grad_norm": 2.369779109954834,
      "learning_rate": 9.437791113178282e-05,
      "loss": 1.4662,
      "step": 762
    },
    {
      "epoch": 4.954545454545455,
      "grad_norm": 2.063880443572998,
      "learning_rate": 9.43634291949357e-05,
      "loss": 1.3133,
      "step": 763
    },
    {
      "epoch": 4.961038961038961,
      "grad_norm": 2.3315935134887695,
      "learning_rate": 9.434892974410932e-05,
      "loss": 1.4277,
      "step": 764
    },
    {
      "epoch": 4.967532467532467,
      "grad_norm": 2.5497238636016846,
      "learning_rate": 9.433441278502783e-05,
      "loss": 1.4894,
      "step": 765
    },
    {
      "epoch": 4.974025974025974,
      "grad_norm": 2.4642364978790283,
      "learning_rate": 9.431987832342228e-05,
      "loss": 1.4023,
      "step": 766
    },
    {
      "epoch": 4.98051948051948,
      "grad_norm": 2.380721092224121,
      "learning_rate": 9.430532636503068e-05,
      "loss": 1.3227,
      "step": 767
    },
    {
      "epoch": 4.987012987012987,
      "grad_norm": 2.2699685096740723,
      "learning_rate": 9.429075691559787e-05,
      "loss": 1.352,
      "step": 768
    },
    {
      "epoch": 4.9935064935064934,
      "grad_norm": 2.156005382537842,
      "learning_rate": 9.427616998087568e-05,
      "loss": 1.3047,
      "step": 769
    },
    {
      "epoch": 5.0,
      "grad_norm": 1483.0968017578125,
      "learning_rate": 9.426156556662276e-05,
      "loss": 1.3802,
      "step": 770
    },
    {
      "epoch": 5.0064935064935066,
      "grad_norm": 2.4799044132232666,
      "learning_rate": 9.424694367860473e-05,
      "loss": 1.214,
      "step": 771
    },
    {
      "epoch": 5.012987012987013,
      "grad_norm": 3.611436128616333,
      "learning_rate": 9.423230432259409e-05,
      "loss": 1.1361,
      "step": 772
    },
    {
      "epoch": 5.01948051948052,
      "grad_norm": 2.477595567703247,
      "learning_rate": 9.421764750437019e-05,
      "loss": 1.322,
      "step": 773
    },
    {
      "epoch": 5.025974025974026,
      "grad_norm": 2.0933218002319336,
      "learning_rate": 9.420297322971933e-05,
      "loss": 1.1123,
      "step": 774
    },
    {
      "epoch": 5.032467532467533,
      "grad_norm": 2.1223325729370117,
      "learning_rate": 9.418828150443467e-05,
      "loss": 1.2504,
      "step": 775
    },
    {
      "epoch": 5.038961038961039,
      "grad_norm": 2.640428304672241,
      "learning_rate": 9.41735723343163e-05,
      "loss": 1.3819,
      "step": 776
    },
    {
      "epoch": 5.045454545454546,
      "grad_norm": 2.309741258621216,
      "learning_rate": 9.415884572517113e-05,
      "loss": 1.2354,
      "step": 777
    },
    {
      "epoch": 5.0519480519480515,
      "grad_norm": 2.4606881141662598,
      "learning_rate": 9.414410168281302e-05,
      "loss": 1.4091,
      "step": 778
    },
    {
      "epoch": 5.058441558441558,
      "grad_norm": 2.1588926315307617,
      "learning_rate": 9.412934021306267e-05,
      "loss": 1.1284,
      "step": 779
    },
    {
      "epoch": 5.064935064935065,
      "grad_norm": 2.3062281608581543,
      "learning_rate": 9.411456132174767e-05,
      "loss": 1.2039,
      "step": 780
    },
    {
      "epoch": 5.071428571428571,
      "grad_norm": 2.493637800216675,
      "learning_rate": 9.40997650147025e-05,
      "loss": 1.2556,
      "step": 781
    },
    {
      "epoch": 5.077922077922078,
      "grad_norm": 2.254223346710205,
      "learning_rate": 9.408495129776852e-05,
      "loss": 1.14,
      "step": 782
    },
    {
      "epoch": 5.084415584415584,
      "grad_norm": 2.187127113342285,
      "learning_rate": 9.407012017679393e-05,
      "loss": 1.1906,
      "step": 783
    },
    {
      "epoch": 5.090909090909091,
      "grad_norm": 2.017028570175171,
      "learning_rate": 9.405527165763384e-05,
      "loss": 1.0582,
      "step": 784
    },
    {
      "epoch": 5.097402597402597,
      "grad_norm": 2.3100154399871826,
      "learning_rate": 9.404040574615018e-05,
      "loss": 1.3244,
      "step": 785
    },
    {
      "epoch": 5.103896103896104,
      "grad_norm": 2.231184482574463,
      "learning_rate": 9.402552244821182e-05,
      "loss": 1.0768,
      "step": 786
    },
    {
      "epoch": 5.1103896103896105,
      "grad_norm": 2.422355890274048,
      "learning_rate": 9.401062176969442e-05,
      "loss": 1.1453,
      "step": 787
    },
    {
      "epoch": 5.116883116883117,
      "grad_norm": 2.3468523025512695,
      "learning_rate": 9.399570371648052e-05,
      "loss": 1.1517,
      "step": 788
    },
    {
      "epoch": 5.123376623376624,
      "grad_norm": 2.1444785594940186,
      "learning_rate": 9.398076829445958e-05,
      "loss": 1.0645,
      "step": 789
    },
    {
      "epoch": 5.12987012987013,
      "grad_norm": 2.26538348197937,
      "learning_rate": 9.396581550952781e-05,
      "loss": 1.1867,
      "step": 790
    },
    {
      "epoch": 5.136363636363637,
      "grad_norm": 2.3012781143188477,
      "learning_rate": 9.395084536758838e-05,
      "loss": 1.1908,
      "step": 791
    },
    {
      "epoch": 5.142857142857143,
      "grad_norm": 2.353574514389038,
      "learning_rate": 9.393585787455124e-05,
      "loss": 1.1686,
      "step": 792
    },
    {
      "epoch": 5.14935064935065,
      "grad_norm": 2.434039354324341,
      "learning_rate": 9.392085303633323e-05,
      "loss": 1.1923,
      "step": 793
    },
    {
      "epoch": 5.1558441558441555,
      "grad_norm": 2.4139668941497803,
      "learning_rate": 9.3905830858858e-05,
      "loss": 1.2742,
      "step": 794
    },
    {
      "epoch": 5.162337662337662,
      "grad_norm": 2.5167605876922607,
      "learning_rate": 9.389079134805609e-05,
      "loss": 1.2763,
      "step": 795
    },
    {
      "epoch": 5.1688311688311686,
      "grad_norm": 2.29848313331604,
      "learning_rate": 9.387573450986484e-05,
      "loss": 1.2501,
      "step": 796
    },
    {
      "epoch": 5.175324675324675,
      "grad_norm": 2.4731240272521973,
      "learning_rate": 9.386066035022848e-05,
      "loss": 1.4086,
      "step": 797
    },
    {
      "epoch": 5.181818181818182,
      "grad_norm": 2.226640224456787,
      "learning_rate": 9.384556887509802e-05,
      "loss": 1.1452,
      "step": 798
    },
    {
      "epoch": 5.188311688311688,
      "grad_norm": 2.1757423877716064,
      "learning_rate": 9.383046009043134e-05,
      "loss": 1.2051,
      "step": 799
    },
    {
      "epoch": 5.194805194805195,
      "grad_norm": 2.5131468772888184,
      "learning_rate": 9.381533400219318e-05,
      "loss": 1.3115,
      "step": 800
    },
    {
      "epoch": 5.201298701298701,
      "grad_norm": 2.236072063446045,
      "learning_rate": 9.380019061635506e-05,
      "loss": 1.1337,
      "step": 801
    },
    {
      "epoch": 5.207792207792208,
      "grad_norm": 2.25288987159729,
      "learning_rate": 9.378502993889533e-05,
      "loss": 1.2222,
      "step": 802
    },
    {
      "epoch": 5.214285714285714,
      "grad_norm": 2.3511617183685303,
      "learning_rate": 9.37698519757992e-05,
      "loss": 1.228,
      "step": 803
    },
    {
      "epoch": 5.220779220779221,
      "grad_norm": 2.281393051147461,
      "learning_rate": 9.375465673305869e-05,
      "loss": 1.1854,
      "step": 804
    },
    {
      "epoch": 5.2272727272727275,
      "grad_norm": 2.451622247695923,
      "learning_rate": 9.373944421667265e-05,
      "loss": 1.2707,
      "step": 805
    },
    {
      "epoch": 5.233766233766234,
      "grad_norm": 2.3030169010162354,
      "learning_rate": 9.372421443264671e-05,
      "loss": 1.0945,
      "step": 806
    },
    {
      "epoch": 5.240259740259741,
      "grad_norm": 2.3412272930145264,
      "learning_rate": 9.370896738699339e-05,
      "loss": 1.1891,
      "step": 807
    },
    {
      "epoch": 5.246753246753247,
      "grad_norm": 2.457958936691284,
      "learning_rate": 9.369370308573198e-05,
      "loss": 1.2034,
      "step": 808
    },
    {
      "epoch": 5.253246753246753,
      "grad_norm": 2.3870041370391846,
      "learning_rate": 9.367842153488854e-05,
      "loss": 1.1308,
      "step": 809
    },
    {
      "epoch": 5.259740259740259,
      "grad_norm": 2.373983860015869,
      "learning_rate": 9.366312274049602e-05,
      "loss": 1.2246,
      "step": 810
    },
    {
      "epoch": 5.266233766233766,
      "grad_norm": 2.238525152206421,
      "learning_rate": 9.364780670859412e-05,
      "loss": 1.1794,
      "step": 811
    },
    {
      "epoch": 5.2727272727272725,
      "grad_norm": 2.2688422203063965,
      "learning_rate": 9.363247344522939e-05,
      "loss": 1.3021,
      "step": 812
    },
    {
      "epoch": 5.279220779220779,
      "grad_norm": 2.4562861919403076,
      "learning_rate": 9.361712295645515e-05,
      "loss": 1.2968,
      "step": 813
    },
    {
      "epoch": 5.285714285714286,
      "grad_norm": 1.8892732858657837,
      "learning_rate": 9.360175524833153e-05,
      "loss": 0.895,
      "step": 814
    },
    {
      "epoch": 5.292207792207792,
      "grad_norm": 2.4016454219818115,
      "learning_rate": 9.358637032692545e-05,
      "loss": 1.1788,
      "step": 815
    },
    {
      "epoch": 5.298701298701299,
      "grad_norm": 2.198923349380493,
      "learning_rate": 9.357096819831064e-05,
      "loss": 1.1777,
      "step": 816
    },
    {
      "epoch": 5.305194805194805,
      "grad_norm": 2.5778071880340576,
      "learning_rate": 9.355554886856762e-05,
      "loss": 1.2441,
      "step": 817
    },
    {
      "epoch": 5.311688311688312,
      "grad_norm": 2.578562021255493,
      "learning_rate": 9.354011234378369e-05,
      "loss": 1.3336,
      "step": 818
    },
    {
      "epoch": 5.318181818181818,
      "grad_norm": 2.29508376121521,
      "learning_rate": 9.352465863005296e-05,
      "loss": 1.3143,
      "step": 819
    },
    {
      "epoch": 5.324675324675325,
      "grad_norm": 2.3751208782196045,
      "learning_rate": 9.35091877334763e-05,
      "loss": 1.3194,
      "step": 820
    },
    {
      "epoch": 5.3311688311688314,
      "grad_norm": 2.456490993499756,
      "learning_rate": 9.349369966016134e-05,
      "loss": 1.2553,
      "step": 821
    },
    {
      "epoch": 5.337662337662338,
      "grad_norm": 1.9089747667312622,
      "learning_rate": 9.347819441622261e-05,
      "loss": 0.9778,
      "step": 822
    },
    {
      "epoch": 5.3441558441558445,
      "grad_norm": 2.301745891571045,
      "learning_rate": 9.346267200778126e-05,
      "loss": 1.1925,
      "step": 823
    },
    {
      "epoch": 5.35064935064935,
      "grad_norm": 2.3186118602752686,
      "learning_rate": 9.344713244096533e-05,
      "loss": 1.2568,
      "step": 824
    },
    {
      "epoch": 5.357142857142857,
      "grad_norm": 2.4224462509155273,
      "learning_rate": 9.343157572190957e-05,
      "loss": 1.2227,
      "step": 825
    },
    {
      "epoch": 5.363636363636363,
      "grad_norm": 2.4902286529541016,
      "learning_rate": 9.341600185675554e-05,
      "loss": 1.2466,
      "step": 826
    },
    {
      "epoch": 5.37012987012987,
      "grad_norm": 2.3658058643341064,
      "learning_rate": 9.340041085165155e-05,
      "loss": 1.288,
      "step": 827
    },
    {
      "epoch": 5.376623376623376,
      "grad_norm": 2.256941318511963,
      "learning_rate": 9.33848027127527e-05,
      "loss": 1.2251,
      "step": 828
    },
    {
      "epoch": 5.383116883116883,
      "grad_norm": 2.4632041454315186,
      "learning_rate": 9.336917744622081e-05,
      "loss": 1.3159,
      "step": 829
    },
    {
      "epoch": 5.3896103896103895,
      "grad_norm": 2.1642262935638428,
      "learning_rate": 9.33535350582245e-05,
      "loss": 1.1923,
      "step": 830
    },
    {
      "epoch": 5.396103896103896,
      "grad_norm": 2.154273748397827,
      "learning_rate": 9.333787555493914e-05,
      "loss": 1.1474,
      "step": 831
    },
    {
      "epoch": 5.402597402597403,
      "grad_norm": 2.319180965423584,
      "learning_rate": 9.332219894254686e-05,
      "loss": 1.2664,
      "step": 832
    },
    {
      "epoch": 5.409090909090909,
      "grad_norm": 2.0859668254852295,
      "learning_rate": 9.330650522723652e-05,
      "loss": 1.0003,
      "step": 833
    },
    {
      "epoch": 5.415584415584416,
      "grad_norm": 2.426114082336426,
      "learning_rate": 9.329079441520377e-05,
      "loss": 1.249,
      "step": 834
    },
    {
      "epoch": 5.422077922077922,
      "grad_norm": 2.3975894451141357,
      "learning_rate": 9.327506651265095e-05,
      "loss": 1.2747,
      "step": 835
    },
    {
      "epoch": 5.428571428571429,
      "grad_norm": 2.561692476272583,
      "learning_rate": 9.325932152578725e-05,
      "loss": 1.1971,
      "step": 836
    },
    {
      "epoch": 5.435064935064935,
      "grad_norm": 2.311870813369751,
      "learning_rate": 9.324355946082848e-05,
      "loss": 1.1388,
      "step": 837
    },
    {
      "epoch": 5.441558441558442,
      "grad_norm": 2.2985804080963135,
      "learning_rate": 9.322778032399728e-05,
      "loss": 1.1915,
      "step": 838
    },
    {
      "epoch": 5.448051948051948,
      "grad_norm": 2.5095090866088867,
      "learning_rate": 9.321198412152301e-05,
      "loss": 1.2289,
      "step": 839
    },
    {
      "epoch": 5.454545454545454,
      "grad_norm": 2.4977407455444336,
      "learning_rate": 9.319617085964176e-05,
      "loss": 1.1548,
      "step": 840
    },
    {
      "epoch": 5.461038961038961,
      "grad_norm": 2.479334831237793,
      "learning_rate": 9.318034054459637e-05,
      "loss": 1.2881,
      "step": 841
    },
    {
      "epoch": 5.467532467532467,
      "grad_norm": 2.316788911819458,
      "learning_rate": 9.316449318263635e-05,
      "loss": 1.0807,
      "step": 842
    },
    {
      "epoch": 5.474025974025974,
      "grad_norm": 2.3556737899780273,
      "learning_rate": 9.314862878001803e-05,
      "loss": 1.3983,
      "step": 843
    },
    {
      "epoch": 5.48051948051948,
      "grad_norm": 2.27823805809021,
      "learning_rate": 9.313274734300439e-05,
      "loss": 1.2266,
      "step": 844
    },
    {
      "epoch": 5.487012987012987,
      "grad_norm": 2.213878870010376,
      "learning_rate": 9.31168488778652e-05,
      "loss": 1.1607,
      "step": 845
    },
    {
      "epoch": 5.4935064935064934,
      "grad_norm": 2.411067485809326,
      "learning_rate": 9.310093339087692e-05,
      "loss": 1.3101,
      "step": 846
    },
    {
      "epoch": 5.5,
      "grad_norm": 2.3228533267974854,
      "learning_rate": 9.308500088832272e-05,
      "loss": 1.1342,
      "step": 847
    },
    {
      "epoch": 5.5064935064935066,
      "grad_norm": 2.511704683303833,
      "learning_rate": 9.30690513764925e-05,
      "loss": 1.3019,
      "step": 848
    },
    {
      "epoch": 5.512987012987013,
      "grad_norm": 2.267284393310547,
      "learning_rate": 9.305308486168288e-05,
      "loss": 1.2113,
      "step": 849
    },
    {
      "epoch": 5.51948051948052,
      "grad_norm": 2.3928427696228027,
      "learning_rate": 9.30371013501972e-05,
      "loss": 1.2584,
      "step": 850
    },
    {
      "epoch": 5.525974025974026,
      "grad_norm": 2.4797680377960205,
      "learning_rate": 9.302110084834545e-05,
      "loss": 1.2511,
      "step": 851
    },
    {
      "epoch": 5.532467532467533,
      "grad_norm": 2.280144214630127,
      "learning_rate": 9.300508336244444e-05,
      "loss": 1.3214,
      "step": 852
    },
    {
      "epoch": 5.538961038961039,
      "grad_norm": 2.3917574882507324,
      "learning_rate": 9.298904889881757e-05,
      "loss": 1.3104,
      "step": 853
    },
    {
      "epoch": 5.545454545454545,
      "grad_norm": 1.8771497011184692,
      "learning_rate": 9.297299746379502e-05,
      "loss": 1.0212,
      "step": 854
    },
    {
      "epoch": 5.551948051948052,
      "grad_norm": 2.3260936737060547,
      "learning_rate": 9.295692906371363e-05,
      "loss": 1.2671,
      "step": 855
    },
    {
      "epoch": 5.558441558441558,
      "grad_norm": 2.335958957672119,
      "learning_rate": 9.294084370491694e-05,
      "loss": 1.2515,
      "step": 856
    },
    {
      "epoch": 5.564935064935065,
      "grad_norm": 2.451331853866577,
      "learning_rate": 9.292474139375522e-05,
      "loss": 1.0896,
      "step": 857
    },
    {
      "epoch": 5.571428571428571,
      "grad_norm": 2.263322353363037,
      "learning_rate": 9.29086221365854e-05,
      "loss": 1.247,
      "step": 858
    },
    {
      "epoch": 5.577922077922078,
      "grad_norm": 2.4265804290771484,
      "learning_rate": 9.289248593977109e-05,
      "loss": 1.2482,
      "step": 859
    },
    {
      "epoch": 5.584415584415584,
      "grad_norm": 2.4868714809417725,
      "learning_rate": 9.287633280968261e-05,
      "loss": 1.2864,
      "step": 860
    },
    {
      "epoch": 5.590909090909091,
      "grad_norm": 2.502161979675293,
      "learning_rate": 9.286016275269698e-05,
      "loss": 1.2699,
      "step": 861
    },
    {
      "epoch": 5.597402597402597,
      "grad_norm": 2.301053762435913,
      "learning_rate": 9.284397577519788e-05,
      "loss": 1.2422,
      "step": 862
    },
    {
      "epoch": 5.603896103896104,
      "grad_norm": 2.7383460998535156,
      "learning_rate": 9.282777188357565e-05,
      "loss": 1.3225,
      "step": 863
    },
    {
      "epoch": 5.6103896103896105,
      "grad_norm": 2.5754849910736084,
      "learning_rate": 9.281155108422733e-05,
      "loss": 1.3382,
      "step": 864
    },
    {
      "epoch": 5.616883116883117,
      "grad_norm": 2.30465030670166,
      "learning_rate": 9.279531338355666e-05,
      "loss": 1.2114,
      "step": 865
    },
    {
      "epoch": 5.623376623376624,
      "grad_norm": 2.1572651863098145,
      "learning_rate": 9.2779058787974e-05,
      "loss": 1.2831,
      "step": 866
    },
    {
      "epoch": 5.62987012987013,
      "grad_norm": 2.3018293380737305,
      "learning_rate": 9.276278730389642e-05,
      "loss": 1.1312,
      "step": 867
    },
    {
      "epoch": 5.636363636363637,
      "grad_norm": 2.3740503787994385,
      "learning_rate": 9.274649893774767e-05,
      "loss": 1.2546,
      "step": 868
    },
    {
      "epoch": 5.642857142857143,
      "grad_norm": 2.359429121017456,
      "learning_rate": 9.273019369595809e-05,
      "loss": 1.2289,
      "step": 869
    },
    {
      "epoch": 5.64935064935065,
      "grad_norm": 2.7115275859832764,
      "learning_rate": 9.271387158496476e-05,
      "loss": 1.3554,
      "step": 870
    },
    {
      "epoch": 5.6558441558441555,
      "grad_norm": 1.951027274131775,
      "learning_rate": 9.269753261121138e-05,
      "loss": 1.1395,
      "step": 871
    },
    {
      "epoch": 5.662337662337662,
      "grad_norm": 2.2826426029205322,
      "learning_rate": 9.268117678114834e-05,
      "loss": 1.2125,
      "step": 872
    },
    {
      "epoch": 5.6688311688311686,
      "grad_norm": 2.1922526359558105,
      "learning_rate": 9.266480410123264e-05,
      "loss": 1.0959,
      "step": 873
    },
    {
      "epoch": 5.675324675324675,
      "grad_norm": 2.3663859367370605,
      "learning_rate": 9.264841457792795e-05,
      "loss": 1.2634,
      "step": 874
    },
    {
      "epoch": 5.681818181818182,
      "grad_norm": 2.077533483505249,
      "learning_rate": 9.263200821770461e-05,
      "loss": 1.1805,
      "step": 875
    },
    {
      "epoch": 5.688311688311688,
      "grad_norm": 2.255629062652588,
      "learning_rate": 9.26155850270396e-05,
      "loss": 1.3269,
      "step": 876
    },
    {
      "epoch": 5.694805194805195,
      "grad_norm": 2.387958288192749,
      "learning_rate": 9.259914501241652e-05,
      "loss": 1.2001,
      "step": 877
    },
    {
      "epoch": 5.701298701298701,
      "grad_norm": 2.3133974075317383,
      "learning_rate": 9.258268818032561e-05,
      "loss": 1.2889,
      "step": 878
    },
    {
      "epoch": 5.707792207792208,
      "grad_norm": 2.17250919342041,
      "learning_rate": 9.256621453726379e-05,
      "loss": 1.2761,
      "step": 879
    },
    {
      "epoch": 5.714285714285714,
      "grad_norm": 2.288163900375366,
      "learning_rate": 9.254972408973461e-05,
      "loss": 1.2545,
      "step": 880
    },
    {
      "epoch": 5.720779220779221,
      "grad_norm": 2.2603843212127686,
      "learning_rate": 9.25332168442482e-05,
      "loss": 1.2539,
      "step": 881
    },
    {
      "epoch": 5.7272727272727275,
      "grad_norm": 2.111243963241577,
      "learning_rate": 9.251669280732137e-05,
      "loss": 1.2284,
      "step": 882
    },
    {
      "epoch": 5.733766233766234,
      "grad_norm": 9.176555633544922,
      "learning_rate": 9.250015198547757e-05,
      "loss": 1.3237,
      "step": 883
    },
    {
      "epoch": 5.740259740259741,
      "grad_norm": 2.652176856994629,
      "learning_rate": 9.248359438524683e-05,
      "loss": 1.3154,
      "step": 884
    },
    {
      "epoch": 5.746753246753247,
      "grad_norm": 2.4932117462158203,
      "learning_rate": 9.246702001316583e-05,
      "loss": 1.2712,
      "step": 885
    },
    {
      "epoch": 5.753246753246753,
      "grad_norm": 2.2778451442718506,
      "learning_rate": 9.245042887577788e-05,
      "loss": 1.3088,
      "step": 886
    },
    {
      "epoch": 5.759740259740259,
      "grad_norm": 2.240194082260132,
      "learning_rate": 9.243382097963291e-05,
      "loss": 1.1532,
      "step": 887
    },
    {
      "epoch": 5.766233766233766,
      "grad_norm": 2.4618711471557617,
      "learning_rate": 9.241719633128743e-05,
      "loss": 1.4605,
      "step": 888
    },
    {
      "epoch": 5.7727272727272725,
      "grad_norm": 2.5479254722595215,
      "learning_rate": 9.24005549373046e-05,
      "loss": 1.3666,
      "step": 889
    },
    {
      "epoch": 5.779220779220779,
      "grad_norm": 2.277982473373413,
      "learning_rate": 9.238389680425416e-05,
      "loss": 1.3209,
      "step": 890
    },
    {
      "epoch": 5.785714285714286,
      "grad_norm": 2.3099091053009033,
      "learning_rate": 9.236722193871252e-05,
      "loss": 1.2911,
      "step": 891
    },
    {
      "epoch": 5.792207792207792,
      "grad_norm": 2.1005160808563232,
      "learning_rate": 9.23505303472626e-05,
      "loss": 1.1609,
      "step": 892
    },
    {
      "epoch": 5.798701298701299,
      "grad_norm": 2.325089693069458,
      "learning_rate": 9.233382203649401e-05,
      "loss": 1.2268,
      "step": 893
    },
    {
      "epoch": 5.805194805194805,
      "grad_norm": 2.2837464809417725,
      "learning_rate": 9.231709701300293e-05,
      "loss": 1.2928,
      "step": 894
    },
    {
      "epoch": 5.811688311688312,
      "grad_norm": 2.1417577266693115,
      "learning_rate": 9.230035528339211e-05,
      "loss": 1.2102,
      "step": 895
    },
    {
      "epoch": 5.818181818181818,
      "grad_norm": 2.3167195320129395,
      "learning_rate": 9.228359685427095e-05,
      "loss": 1.2982,
      "step": 896
    },
    {
      "epoch": 5.824675324675325,
      "grad_norm": 2.281616449356079,
      "learning_rate": 9.226682173225537e-05,
      "loss": 1.2442,
      "step": 897
    },
    {
      "epoch": 5.8311688311688314,
      "grad_norm": 2.245262622833252,
      "learning_rate": 9.225002992396796e-05,
      "loss": 1.2757,
      "step": 898
    },
    {
      "epoch": 5.837662337662338,
      "grad_norm": 2.2756450176239014,
      "learning_rate": 9.223322143603785e-05,
      "loss": 1.1611,
      "step": 899
    },
    {
      "epoch": 5.8441558441558445,
      "grad_norm": 2.2296695709228516,
      "learning_rate": 9.221639627510076e-05,
      "loss": 1.2486,
      "step": 900
    },
    {
      "epoch": 5.85064935064935,
      "grad_norm": 2.456972599029541,
      "learning_rate": 9.2199554447799e-05,
      "loss": 1.2553,
      "step": 901
    },
    {
      "epoch": 5.857142857142857,
      "grad_norm": 2.2772555351257324,
      "learning_rate": 9.218269596078146e-05,
      "loss": 1.1621,
      "step": 902
    },
    {
      "epoch": 5.863636363636363,
      "grad_norm": 2.2198710441589355,
      "learning_rate": 9.216582082070358e-05,
      "loss": 1.1448,
      "step": 903
    },
    {
      "epoch": 5.87012987012987,
      "grad_norm": 2.201732873916626,
      "learning_rate": 9.214892903422744e-05,
      "loss": 1.1754,
      "step": 904
    },
    {
      "epoch": 5.876623376623376,
      "grad_norm": 2.263493061065674,
      "learning_rate": 9.213202060802161e-05,
      "loss": 1.3105,
      "step": 905
    },
    {
      "epoch": 5.883116883116883,
      "grad_norm": 2.3806960582733154,
      "learning_rate": 9.21150955487613e-05,
      "loss": 1.2363,
      "step": 906
    },
    {
      "epoch": 5.8896103896103895,
      "grad_norm": 2.324129104614258,
      "learning_rate": 9.209815386312824e-05,
      "loss": 1.3469,
      "step": 907
    },
    {
      "epoch": 5.896103896103896,
      "grad_norm": 2.4612674713134766,
      "learning_rate": 9.208119555781074e-05,
      "loss": 1.3424,
      "step": 908
    },
    {
      "epoch": 5.902597402597403,
      "grad_norm": 2.341952323913574,
      "learning_rate": 9.206422063950367e-05,
      "loss": 1.2235,
      "step": 909
    },
    {
      "epoch": 5.909090909090909,
      "grad_norm": 2.305021047592163,
      "learning_rate": 9.204722911490846e-05,
      "loss": 1.3399,
      "step": 910
    },
    {
      "epoch": 5.915584415584416,
      "grad_norm": 2.174239158630371,
      "learning_rate": 9.203022099073309e-05,
      "loss": 1.106,
      "step": 911
    },
    {
      "epoch": 5.922077922077922,
      "grad_norm": 2.136263847351074,
      "learning_rate": 9.201319627369211e-05,
      "loss": 1.185,
      "step": 912
    },
    {
      "epoch": 5.928571428571429,
      "grad_norm": 2.0439507961273193,
      "learning_rate": 9.199615497050659e-05,
      "loss": 1.1936,
      "step": 913
    },
    {
      "epoch": 5.935064935064935,
      "grad_norm": 2.331047773361206,
      "learning_rate": 9.19790970879042e-05,
      "loss": 1.2955,
      "step": 914
    },
    {
      "epoch": 5.941558441558442,
      "grad_norm": 2.1538991928100586,
      "learning_rate": 9.19620226326191e-05,
      "loss": 1.2146,
      "step": 915
    },
    {
      "epoch": 5.948051948051948,
      "grad_norm": 2.235663414001465,
      "learning_rate": 9.194493161139199e-05,
      "loss": 1.2987,
      "step": 916
    },
    {
      "epoch": 5.954545454545455,
      "grad_norm": 2.266671657562256,
      "learning_rate": 9.192782403097018e-05,
      "loss": 1.2618,
      "step": 917
    },
    {
      "epoch": 5.961038961038961,
      "grad_norm": 2.287224292755127,
      "learning_rate": 9.191069989810744e-05,
      "loss": 1.2801,
      "step": 918
    },
    {
      "epoch": 5.967532467532467,
      "grad_norm": 2.454397678375244,
      "learning_rate": 9.189355921956412e-05,
      "loss": 1.3529,
      "step": 919
    },
    {
      "epoch": 5.974025974025974,
      "grad_norm": 2.4021215438842773,
      "learning_rate": 9.187640200210708e-05,
      "loss": 1.2859,
      "step": 920
    },
    {
      "epoch": 5.98051948051948,
      "grad_norm": 2.4222583770751953,
      "learning_rate": 9.185922825250974e-05,
      "loss": 1.3745,
      "step": 921
    },
    {
      "epoch": 5.987012987012987,
      "grad_norm": 1.9028486013412476,
      "learning_rate": 9.1842037977552e-05,
      "loss": 1.0603,
      "step": 922
    },
    {
      "epoch": 5.9935064935064934,
      "grad_norm": 2.0316622257232666,
      "learning_rate": 9.182483118402033e-05,
      "loss": 1.1392,
      "step": 923
    },
    {
      "epoch": 6.0,
      "grad_norm": 264.6945495605469,
      "learning_rate": 9.180760787870765e-05,
      "loss": 1.1341,
      "step": 924
    },
    {
      "epoch": 6.0064935064935066,
      "grad_norm": 1.8904297351837158,
      "learning_rate": 9.179036806841352e-05,
      "loss": 0.9749,
      "step": 925
    },
    {
      "epoch": 6.012987012987013,
      "grad_norm": 1.7428091764450073,
      "learning_rate": 9.17731117599439e-05,
      "loss": 0.857,
      "step": 926
    },
    {
      "epoch": 6.01948051948052,
      "grad_norm": 2.1916327476501465,
      "learning_rate": 9.175583896011131e-05,
      "loss": 1.1159,
      "step": 927
    },
    {
      "epoch": 6.025974025974026,
      "grad_norm": 2.009938955307007,
      "learning_rate": 9.173854967573479e-05,
      "loss": 0.985,
      "step": 928
    },
    {
      "epoch": 6.032467532467533,
      "grad_norm": 2.4173030853271484,
      "learning_rate": 9.172124391363985e-05,
      "loss": 1.1468,
      "step": 929
    },
    {
      "epoch": 6.038961038961039,
      "grad_norm": 2.157717227935791,
      "learning_rate": 9.170392168065857e-05,
      "loss": 1.0836,
      "step": 930
    },
    {
      "epoch": 6.045454545454546,
      "grad_norm": 1.9934571981430054,
      "learning_rate": 9.168658298362946e-05,
      "loss": 1.0066,
      "step": 931
    },
    {
      "epoch": 6.0519480519480515,
      "grad_norm": 2.120309352874756,
      "learning_rate": 9.166922782939758e-05,
      "loss": 1.1275,
      "step": 932
    },
    {
      "epoch": 6.058441558441558,
      "grad_norm": 2.1822173595428467,
      "learning_rate": 9.165185622481447e-05,
      "loss": 1.1083,
      "step": 933
    },
    {
      "epoch": 6.064935064935065,
      "grad_norm": 2.1950483322143555,
      "learning_rate": 9.163446817673817e-05,
      "loss": 1.0423,
      "step": 934
    },
    {
      "epoch": 6.071428571428571,
      "grad_norm": 2.231881856918335,
      "learning_rate": 9.161706369203317e-05,
      "loss": 1.086,
      "step": 935
    },
    {
      "epoch": 6.077922077922078,
      "grad_norm": 2.369540214538574,
      "learning_rate": 9.159964277757054e-05,
      "loss": 0.9978,
      "step": 936
    },
    {
      "epoch": 6.084415584415584,
      "grad_norm": 2.1250226497650146,
      "learning_rate": 9.158220544022773e-05,
      "loss": 0.9875,
      "step": 937
    },
    {
      "epoch": 6.090909090909091,
      "grad_norm": 2.21665096282959,
      "learning_rate": 9.156475168688877e-05,
      "loss": 1.0885,
      "step": 938
    },
    {
      "epoch": 6.097402597402597,
      "grad_norm": 1.7771512269973755,
      "learning_rate": 9.154728152444408e-05,
      "loss": 0.789,
      "step": 939
    },
    {
      "epoch": 6.103896103896104,
      "grad_norm": 2.3178699016571045,
      "learning_rate": 9.152979495979063e-05,
      "loss": 1.0956,
      "step": 940
    },
    {
      "epoch": 6.1103896103896105,
      "grad_norm": 2.534238576889038,
      "learning_rate": 9.151229199983184e-05,
      "loss": 1.092,
      "step": 941
    },
    {
      "epoch": 6.116883116883117,
      "grad_norm": 2.4788708686828613,
      "learning_rate": 9.14947726514776e-05,
      "loss": 1.1408,
      "step": 942
    },
    {
      "epoch": 6.123376623376624,
      "grad_norm": 2.188908576965332,
      "learning_rate": 9.147723692164427e-05,
      "loss": 1.0523,
      "step": 943
    },
    {
      "epoch": 6.12987012987013,
      "grad_norm": 1.940743327140808,
      "learning_rate": 9.145968481725467e-05,
      "loss": 0.8914,
      "step": 944
    },
    {
      "epoch": 6.136363636363637,
      "grad_norm": 2.2595057487487793,
      "learning_rate": 9.14421163452381e-05,
      "loss": 1.064,
      "step": 945
    },
    {
      "epoch": 6.142857142857143,
      "grad_norm": 2.3622570037841797,
      "learning_rate": 9.142453151253032e-05,
      "loss": 1.0703,
      "step": 946
    },
    {
      "epoch": 6.14935064935065,
      "grad_norm": 2.097458600997925,
      "learning_rate": 9.140693032607353e-05,
      "loss": 0.9451,
      "step": 947
    },
    {
      "epoch": 6.1558441558441555,
      "grad_norm": 2.064142942428589,
      "learning_rate": 9.138931279281639e-05,
      "loss": 0.9752,
      "step": 948
    },
    {
      "epoch": 6.162337662337662,
      "grad_norm": 2.3236732482910156,
      "learning_rate": 9.137167891971407e-05,
      "loss": 1.0357,
      "step": 949
    },
    {
      "epoch": 6.1688311688311686,
      "grad_norm": 2.375629186630249,
      "learning_rate": 9.135402871372808e-05,
      "loss": 1.1145,
      "step": 950
    },
    {
      "epoch": 6.175324675324675,
      "grad_norm": 2.575246572494507,
      "learning_rate": 9.13363621818265e-05,
      "loss": 1.2448,
      "step": 951
    },
    {
      "epoch": 6.181818181818182,
      "grad_norm": 2.221675395965576,
      "learning_rate": 9.131867933098378e-05,
      "loss": 1.0203,
      "step": 952
    },
    {
      "epoch": 6.188311688311688,
      "grad_norm": 2.4579875469207764,
      "learning_rate": 9.13009801681808e-05,
      "loss": 1.1714,
      "step": 953
    },
    {
      "epoch": 6.194805194805195,
      "grad_norm": 2.4114532470703125,
      "learning_rate": 9.128326470040495e-05,
      "loss": 1.0732,
      "step": 954
    },
    {
      "epoch": 6.201298701298701,
      "grad_norm": 2.374720335006714,
      "learning_rate": 9.126553293464998e-05,
      "loss": 1.0896,
      "step": 955
    },
    {
      "epoch": 6.207792207792208,
      "grad_norm": 2.4675614833831787,
      "learning_rate": 9.124778487791615e-05,
      "loss": 1.1355,
      "step": 956
    },
    {
      "epoch": 6.214285714285714,
      "grad_norm": 2.0805985927581787,
      "learning_rate": 9.123002053721005e-05,
      "loss": 1.0721,
      "step": 957
    },
    {
      "epoch": 6.220779220779221,
      "grad_norm": 2.3115074634552,
      "learning_rate": 9.121223991954484e-05,
      "loss": 1.0833,
      "step": 958
    },
    {
      "epoch": 6.2272727272727275,
      "grad_norm": 2.4905924797058105,
      "learning_rate": 9.119444303193996e-05,
      "loss": 1.0787,
      "step": 959
    },
    {
      "epoch": 6.233766233766234,
      "grad_norm": 2.3526439666748047,
      "learning_rate": 9.117662988142138e-05,
      "loss": 1.1503,
      "step": 960
    },
    {
      "epoch": 6.240259740259741,
      "grad_norm": 2.615757703781128,
      "learning_rate": 9.115880047502142e-05,
      "loss": 1.1554,
      "step": 961
    },
    {
      "epoch": 6.246753246753247,
      "grad_norm": 2.563284158706665,
      "learning_rate": 9.114095481977888e-05,
      "loss": 1.3076,
      "step": 962
    },
    {
      "epoch": 6.253246753246753,
      "grad_norm": 2.29913592338562,
      "learning_rate": 9.112309292273891e-05,
      "loss": 1.1206,
      "step": 963
    },
    {
      "epoch": 6.259740259740259,
      "grad_norm": 2.1598641872406006,
      "learning_rate": 9.110521479095312e-05,
      "loss": 1.0892,
      "step": 964
    },
    {
      "epoch": 6.266233766233766,
      "grad_norm": 2.0765597820281982,
      "learning_rate": 9.108732043147952e-05,
      "loss": 1.013,
      "step": 965
    },
    {
      "epoch": 6.2727272727272725,
      "grad_norm": 2.281052589416504,
      "learning_rate": 9.10694098513825e-05,
      "loss": 1.1901,
      "step": 966
    },
    {
      "epoch": 6.279220779220779,
      "grad_norm": 2.169802665710449,
      "learning_rate": 9.10514830577329e-05,
      "loss": 1.0905,
      "step": 967
    },
    {
      "epoch": 6.285714285714286,
      "grad_norm": 2.2257354259490967,
      "learning_rate": 9.103354005760791e-05,
      "loss": 1.0368,
      "step": 968
    },
    {
      "epoch": 6.292207792207792,
      "grad_norm": 2.169473648071289,
      "learning_rate": 9.101558085809114e-05,
      "loss": 1.0166,
      "step": 969
    },
    {
      "epoch": 6.298701298701299,
      "grad_norm": 2.37520694732666,
      "learning_rate": 9.099760546627261e-05,
      "loss": 1.1822,
      "step": 970
    },
    {
      "epoch": 6.305194805194805,
      "grad_norm": 2.1943604946136475,
      "learning_rate": 9.097961388924873e-05,
      "loss": 1.1304,
      "step": 971
    },
    {
      "epoch": 6.311688311688312,
      "grad_norm": 2.553886651992798,
      "learning_rate": 9.096160613412228e-05,
      "loss": 1.3013,
      "step": 972
    },
    {
      "epoch": 6.318181818181818,
      "grad_norm": 2.3251588344573975,
      "learning_rate": 9.094358220800243e-05,
      "loss": 1.0648,
      "step": 973
    },
    {
      "epoch": 6.324675324675325,
      "grad_norm": 2.508411407470703,
      "learning_rate": 9.092554211800474e-05,
      "loss": 1.1993,
      "step": 974
    },
    {
      "epoch": 6.3311688311688314,
      "grad_norm": 2.10422682762146,
      "learning_rate": 9.090748587125118e-05,
      "loss": 1.1125,
      "step": 975
    },
    {
      "epoch": 6.337662337662338,
      "grad_norm": 2.246373176574707,
      "learning_rate": 9.088941347487003e-05,
      "loss": 1.1525,
      "step": 976
    },
    {
      "epoch": 6.3441558441558445,
      "grad_norm": 2.19040846824646,
      "learning_rate": 9.0871324935996e-05,
      "loss": 1.0981,
      "step": 977
    },
    {
      "epoch": 6.35064935064935,
      "grad_norm": 2.4398250579833984,
      "learning_rate": 9.085322026177017e-05,
      "loss": 1.2318,
      "step": 978
    },
    {
      "epoch": 6.357142857142857,
      "grad_norm": 2.36503267288208,
      "learning_rate": 9.083509945933997e-05,
      "loss": 1.1359,
      "step": 979
    },
    {
      "epoch": 6.363636363636363,
      "grad_norm": 2.545269012451172,
      "learning_rate": 9.081696253585921e-05,
      "loss": 1.0359,
      "step": 980
    },
    {
      "epoch": 6.37012987012987,
      "grad_norm": 2.456841468811035,
      "learning_rate": 9.079880949848805e-05,
      "loss": 1.1641,
      "step": 981
    },
    {
      "epoch": 6.376623376623376,
      "grad_norm": 2.2646665573120117,
      "learning_rate": 9.078064035439301e-05,
      "loss": 1.1425,
      "step": 982
    },
    {
      "epoch": 6.383116883116883,
      "grad_norm": 2.244145154953003,
      "learning_rate": 9.076245511074703e-05,
      "loss": 1.0961,
      "step": 983
    },
    {
      "epoch": 6.3896103896103895,
      "grad_norm": 2.3785531520843506,
      "learning_rate": 9.074425377472931e-05,
      "loss": 1.1008,
      "step": 984
    },
    {
      "epoch": 6.396103896103896,
      "grad_norm": 2.486354351043701,
      "learning_rate": 9.072603635352548e-05,
      "loss": 1.1167,
      "step": 985
    },
    {
      "epoch": 6.402597402597403,
      "grad_norm": 2.5050578117370605,
      "learning_rate": 9.070780285432745e-05,
      "loss": 1.1885,
      "step": 986
    },
    {
      "epoch": 6.409090909090909,
      "grad_norm": 2.260310649871826,
      "learning_rate": 9.068955328433355e-05,
      "loss": 1.1354,
      "step": 987
    },
    {
      "epoch": 6.415584415584416,
      "grad_norm": 2.1151485443115234,
      "learning_rate": 9.067128765074842e-05,
      "loss": 1.0162,
      "step": 988
    },
    {
      "epoch": 6.422077922077922,
      "grad_norm": 2.35497784614563,
      "learning_rate": 9.065300596078303e-05,
      "loss": 1.0928,
      "step": 989
    },
    {
      "epoch": 6.428571428571429,
      "grad_norm": 2.429023504257202,
      "learning_rate": 9.06347082216547e-05,
      "loss": 1.1322,
      "step": 990
    },
    {
      "epoch": 6.435064935064935,
      "grad_norm": 2.556748867034912,
      "learning_rate": 9.06163944405871e-05,
      "loss": 1.208,
      "step": 991
    },
    {
      "epoch": 6.441558441558442,
      "grad_norm": 2.451526403427124,
      "learning_rate": 9.059806462481023e-05,
      "loss": 1.0806,
      "step": 992
    },
    {
      "epoch": 6.448051948051948,
      "grad_norm": 2.4315357208251953,
      "learning_rate": 9.057971878156036e-05,
      "loss": 1.2842,
      "step": 993
    },
    {
      "epoch": 6.454545454545454,
      "grad_norm": 2.3418030738830566,
      "learning_rate": 9.056135691808019e-05,
      "loss": 1.1049,
      "step": 994
    },
    {
      "epoch": 6.461038961038961,
      "grad_norm": 2.454900026321411,
      "learning_rate": 9.054297904161868e-05,
      "loss": 1.2332,
      "step": 995
    },
    {
      "epoch": 6.467532467532467,
      "grad_norm": 2.3480823040008545,
      "learning_rate": 9.052458515943111e-05,
      "loss": 1.0631,
      "step": 996
    },
    {
      "epoch": 6.474025974025974,
      "grad_norm": 2.2208402156829834,
      "learning_rate": 9.050617527877911e-05,
      "loss": 1.1265,
      "step": 997
    },
    {
      "epoch": 6.48051948051948,
      "grad_norm": 2.285851240158081,
      "learning_rate": 9.048774940693062e-05,
      "loss": 1.0793,
      "step": 998
    },
    {
      "epoch": 6.487012987012987,
      "grad_norm": 1.9776757955551147,
      "learning_rate": 9.046930755115985e-05,
      "loss": 0.8993,
      "step": 999
    },
    {
      "epoch": 6.4935064935064934,
      "grad_norm": 2.425506591796875,
      "learning_rate": 9.045084971874738e-05,
      "loss": 1.183,
      "step": 1000
    },
    {
      "epoch": 6.5,
      "grad_norm": 2.349262237548828,
      "learning_rate": 9.043237591698004e-05,
      "loss": 1.1151,
      "step": 1001
    },
    {
      "epoch": 6.5064935064935066,
      "grad_norm": 2.294227123260498,
      "learning_rate": 9.041388615315102e-05,
      "loss": 1.1007,
      "step": 1002
    },
    {
      "epoch": 6.512987012987013,
      "grad_norm": 2.1493442058563232,
      "learning_rate": 9.03953804345598e-05,
      "loss": 1.0778,
      "step": 1003
    },
    {
      "epoch": 6.51948051948052,
      "grad_norm": 2.619725227355957,
      "learning_rate": 9.03768587685121e-05,
      "loss": 1.2408,
      "step": 1004
    },
    {
      "epoch": 6.525974025974026,
      "grad_norm": 2.156489372253418,
      "learning_rate": 9.035832116232001e-05,
      "loss": 1.0415,
      "step": 1005
    },
    {
      "epoch": 6.532467532467533,
      "grad_norm": 2.3598475456237793,
      "learning_rate": 9.03397676233019e-05,
      "loss": 1.1308,
      "step": 1006
    },
    {
      "epoch": 6.538961038961039,
      "grad_norm": 2.3119537830352783,
      "learning_rate": 9.032119815878236e-05,
      "loss": 1.1402,
      "step": 1007
    },
    {
      "epoch": 6.545454545454545,
      "grad_norm": 2.459968328475952,
      "learning_rate": 9.030261277609236e-05,
      "loss": 1.2053,
      "step": 1008
    },
    {
      "epoch": 6.551948051948052,
      "grad_norm": 2.4004056453704834,
      "learning_rate": 9.02840114825691e-05,
      "loss": 1.2789,
      "step": 1009
    },
    {
      "epoch": 6.558441558441558,
      "grad_norm": 2.6905951499938965,
      "learning_rate": 9.02653942855561e-05,
      "loss": 1.1713,
      "step": 1010
    },
    {
      "epoch": 6.564935064935065,
      "grad_norm": 2.306800365447998,
      "learning_rate": 9.024676119240311e-05,
      "loss": 1.1418,
      "step": 1011
    },
    {
      "epoch": 6.571428571428571,
      "grad_norm": 2.355865001678467,
      "learning_rate": 9.022811221046618e-05,
      "loss": 1.1373,
      "step": 1012
    },
    {
      "epoch": 6.577922077922078,
      "grad_norm": 2.5856380462646484,
      "learning_rate": 9.020944734710766e-05,
      "loss": 1.2152,
      "step": 1013
    },
    {
      "epoch": 6.584415584415584,
      "grad_norm": 2.4231441020965576,
      "learning_rate": 9.01907666096961e-05,
      "loss": 1.0314,
      "step": 1014
    },
    {
      "epoch": 6.590909090909091,
      "grad_norm": 2.426198720932007,
      "learning_rate": 9.017207000560638e-05,
      "loss": 1.3291,
      "step": 1015
    },
    {
      "epoch": 6.597402597402597,
      "grad_norm": 2.3727035522460938,
      "learning_rate": 9.015335754221964e-05,
      "loss": 1.2275,
      "step": 1016
    },
    {
      "epoch": 6.603896103896104,
      "grad_norm": 2.151705265045166,
      "learning_rate": 9.013462922692324e-05,
      "loss": 1.0354,
      "step": 1017
    },
    {
      "epoch": 6.6103896103896105,
      "grad_norm": 2.015474319458008,
      "learning_rate": 9.011588506711083e-05,
      "loss": 0.9932,
      "step": 1018
    },
    {
      "epoch": 6.616883116883117,
      "grad_norm": 2.2131268978118896,
      "learning_rate": 9.009712507018231e-05,
      "loss": 1.1619,
      "step": 1019
    },
    {
      "epoch": 6.623376623376624,
      "grad_norm": 2.402639865875244,
      "learning_rate": 9.007834924354383e-05,
      "loss": 1.2321,
      "step": 1020
    },
    {
      "epoch": 6.62987012987013,
      "grad_norm": 2.2011077404022217,
      "learning_rate": 9.005955759460779e-05,
      "loss": 1.0755,
      "step": 1021
    },
    {
      "epoch": 6.636363636363637,
      "grad_norm": 2.4937143325805664,
      "learning_rate": 9.004075013079283e-05,
      "loss": 1.1842,
      "step": 1022
    },
    {
      "epoch": 6.642857142857143,
      "grad_norm": 2.269645929336548,
      "learning_rate": 9.002192685952385e-05,
      "loss": 1.0566,
      "step": 1023
    },
    {
      "epoch": 6.64935064935065,
      "grad_norm": 2.230419874191284,
      "learning_rate": 9.000308778823195e-05,
      "loss": 1.0918,
      "step": 1024
    },
    {
      "epoch": 6.6558441558441555,
      "grad_norm": 2.2211496829986572,
      "learning_rate": 8.998423292435454e-05,
      "loss": 1.0644,
      "step": 1025
    },
    {
      "epoch": 6.662337662337662,
      "grad_norm": 2.4004454612731934,
      "learning_rate": 8.996536227533519e-05,
      "loss": 1.21,
      "step": 1026
    },
    {
      "epoch": 6.6688311688311686,
      "grad_norm": 2.172208070755005,
      "learning_rate": 8.994647584862374e-05,
      "loss": 1.1403,
      "step": 1027
    },
    {
      "epoch": 6.675324675324675,
      "grad_norm": 2.2280516624450684,
      "learning_rate": 8.992757365167626e-05,
      "loss": 1.1776,
      "step": 1028
    },
    {
      "epoch": 6.681818181818182,
      "grad_norm": 2.2168164253234863,
      "learning_rate": 8.990865569195502e-05,
      "loss": 1.0141,
      "step": 1029
    },
    {
      "epoch": 6.688311688311688,
      "grad_norm": 2.4428412914276123,
      "learning_rate": 8.988972197692855e-05,
      "loss": 1.2419,
      "step": 1030
    },
    {
      "epoch": 6.694805194805195,
      "grad_norm": 2.5520572662353516,
      "learning_rate": 8.987077251407158e-05,
      "loss": 1.2397,
      "step": 1031
    },
    {
      "epoch": 6.701298701298701,
      "grad_norm": 2.211339235305786,
      "learning_rate": 8.985180731086505e-05,
      "loss": 1.0824,
      "step": 1032
    },
    {
      "epoch": 6.707792207792208,
      "grad_norm": 1.9707832336425781,
      "learning_rate": 8.983282637479614e-05,
      "loss": 1.0771,
      "step": 1033
    },
    {
      "epoch": 6.714285714285714,
      "grad_norm": 2.4689748287200928,
      "learning_rate": 8.981382971335819e-05,
      "loss": 1.1662,
      "step": 1034
    },
    {
      "epoch": 6.720779220779221,
      "grad_norm": 2.417175769805908,
      "learning_rate": 8.97948173340508e-05,
      "loss": 1.1183,
      "step": 1035
    },
    {
      "epoch": 6.7272727272727275,
      "grad_norm": 2.415498733520508,
      "learning_rate": 8.977578924437974e-05,
      "loss": 1.1773,
      "step": 1036
    },
    {
      "epoch": 6.733766233766234,
      "grad_norm": 2.1574041843414307,
      "learning_rate": 8.975674545185703e-05,
      "loss": 1.035,
      "step": 1037
    },
    {
      "epoch": 6.740259740259741,
      "grad_norm": 2.415621042251587,
      "learning_rate": 8.973768596400085e-05,
      "loss": 1.2409,
      "step": 1038
    },
    {
      "epoch": 6.746753246753247,
      "grad_norm": 2.438495635986328,
      "learning_rate": 8.971861078833557e-05,
      "loss": 1.1134,
      "step": 1039
    },
    {
      "epoch": 6.753246753246753,
      "grad_norm": 2.4584386348724365,
      "learning_rate": 8.969951993239177e-05,
      "loss": 1.105,
      "step": 1040
    },
    {
      "epoch": 6.759740259740259,
      "grad_norm": 2.286853313446045,
      "learning_rate": 8.968041340370621e-05,
      "loss": 1.1215,
      "step": 1041
    },
    {
      "epoch": 6.766233766233766,
      "grad_norm": 2.5318830013275146,
      "learning_rate": 8.966129120982188e-05,
      "loss": 1.2174,
      "step": 1042
    },
    {
      "epoch": 6.7727272727272725,
      "grad_norm": 2.14926815032959,
      "learning_rate": 8.964215335828787e-05,
      "loss": 1.0082,
      "step": 1043
    },
    {
      "epoch": 6.779220779220779,
      "grad_norm": 2.270693302154541,
      "learning_rate": 8.962299985665953e-05,
      "loss": 1.1778,
      "step": 1044
    },
    {
      "epoch": 6.785714285714286,
      "grad_norm": 2.420628547668457,
      "learning_rate": 8.960383071249836e-05,
      "loss": 1.1542,
      "step": 1045
    },
    {
      "epoch": 6.792207792207792,
      "grad_norm": 2.379152297973633,
      "learning_rate": 8.958464593337202e-05,
      "loss": 1.2034,
      "step": 1046
    },
    {
      "epoch": 6.798701298701299,
      "grad_norm": 2.2058722972869873,
      "learning_rate": 8.956544552685437e-05,
      "loss": 1.0938,
      "step": 1047
    },
    {
      "epoch": 6.805194805194805,
      "grad_norm": 2.075653553009033,
      "learning_rate": 8.954622950052542e-05,
      "loss": 1.0615,
      "step": 1048
    },
    {
      "epoch": 6.811688311688312,
      "grad_norm": 2.1014020442962646,
      "learning_rate": 8.952699786197137e-05,
      "loss": 1.0083,
      "step": 1049
    },
    {
      "epoch": 6.818181818181818,
      "grad_norm": 2.262732982635498,
      "learning_rate": 8.950775061878453e-05,
      "loss": 1.1361,
      "step": 1050
    },
    {
      "epoch": 6.824675324675325,
      "grad_norm": 2.087364435195923,
      "learning_rate": 8.948848777856343e-05,
      "loss": 1.0599,
      "step": 1051
    },
    {
      "epoch": 6.8311688311688314,
      "grad_norm": 2.4589810371398926,
      "learning_rate": 8.946920934891274e-05,
      "loss": 1.1639,
      "step": 1052
    },
    {
      "epoch": 6.837662337662338,
      "grad_norm": 2.136878490447998,
      "learning_rate": 8.944991533744326e-05,
      "loss": 1.0882,
      "step": 1053
    },
    {
      "epoch": 6.8441558441558445,
      "grad_norm": 2.4780335426330566,
      "learning_rate": 8.943060575177197e-05,
      "loss": 1.2518,
      "step": 1054
    },
    {
      "epoch": 6.85064935064935,
      "grad_norm": 2.233759641647339,
      "learning_rate": 8.941128059952201e-05,
      "loss": 1.1386,
      "step": 1055
    },
    {
      "epoch": 6.857142857142857,
      "grad_norm": 2.1583468914031982,
      "learning_rate": 8.93919398883226e-05,
      "loss": 1.0189,
      "step": 1056
    },
    {
      "epoch": 6.863636363636363,
      "grad_norm": 2.2025363445281982,
      "learning_rate": 8.937258362580919e-05,
      "loss": 1.0769,
      "step": 1057
    },
    {
      "epoch": 6.87012987012987,
      "grad_norm": 2.1733157634735107,
      "learning_rate": 8.93532118196233e-05,
      "loss": 1.0378,
      "step": 1058
    },
    {
      "epoch": 6.876623376623376,
      "grad_norm": 2.4087839126586914,
      "learning_rate": 8.93338244774126e-05,
      "loss": 1.1531,
      "step": 1059
    },
    {
      "epoch": 6.883116883116883,
      "grad_norm": 2.359661340713501,
      "learning_rate": 8.931442160683094e-05,
      "loss": 1.0614,
      "step": 1060
    },
    {
      "epoch": 6.8896103896103895,
      "grad_norm": 2.222322940826416,
      "learning_rate": 8.929500321553826e-05,
      "loss": 1.0687,
      "step": 1061
    },
    {
      "epoch": 6.896103896103896,
      "grad_norm": 2.2607553005218506,
      "learning_rate": 8.92755693112006e-05,
      "loss": 1.2072,
      "step": 1062
    },
    {
      "epoch": 6.902597402597403,
      "grad_norm": 2.3558382987976074,
      "learning_rate": 8.925611990149021e-05,
      "loss": 1.1749,
      "step": 1063
    },
    {
      "epoch": 6.909090909090909,
      "grad_norm": 2.3526875972747803,
      "learning_rate": 8.923665499408536e-05,
      "loss": 1.193,
      "step": 1064
    },
    {
      "epoch": 6.915584415584416,
      "grad_norm": 2.508218288421631,
      "learning_rate": 8.92171745966705e-05,
      "loss": 1.3347,
      "step": 1065
    },
    {
      "epoch": 6.922077922077922,
      "grad_norm": 1.8409452438354492,
      "learning_rate": 8.91976787169362e-05,
      "loss": 0.8723,
      "step": 1066
    },
    {
      "epoch": 6.928571428571429,
      "grad_norm": 2.0151336193084717,
      "learning_rate": 8.917816736257912e-05,
      "loss": 0.952,
      "step": 1067
    },
    {
      "epoch": 6.935064935064935,
      "grad_norm": 1.8350368738174438,
      "learning_rate": 8.915864054130204e-05,
      "loss": 0.8401,
      "step": 1068
    },
    {
      "epoch": 6.941558441558442,
      "grad_norm": 2.169405698776245,
      "learning_rate": 8.91390982608138e-05,
      "loss": 1.1113,
      "step": 1069
    },
    {
      "epoch": 6.948051948051948,
      "grad_norm": 2.3240628242492676,
      "learning_rate": 8.91195405288294e-05,
      "loss": 1.1703,
      "step": 1070
    },
    {
      "epoch": 6.954545454545455,
      "grad_norm": 2.274879217147827,
      "learning_rate": 8.909996735306996e-05,
      "loss": 1.1955,
      "step": 1071
    },
    {
      "epoch": 6.961038961038961,
      "grad_norm": 2.3772335052490234,
      "learning_rate": 8.908037874126263e-05,
      "loss": 1.2593,
      "step": 1072
    },
    {
      "epoch": 6.967532467532467,
      "grad_norm": 2.1066665649414062,
      "learning_rate": 8.906077470114069e-05,
      "loss": 1.0618,
      "step": 1073
    },
    {
      "epoch": 6.974025974025974,
      "grad_norm": 2.411353826522827,
      "learning_rate": 8.904115524044348e-05,
      "loss": 1.2221,
      "step": 1074
    },
    {
      "epoch": 6.98051948051948,
      "grad_norm": 2.37351131439209,
      "learning_rate": 8.90215203669165e-05,
      "loss": 1.1621,
      "step": 1075
    },
    {
      "epoch": 6.987012987012987,
      "grad_norm": 2.0856990814208984,
      "learning_rate": 8.900187008831125e-05,
      "loss": 1.0735,
      "step": 1076
    },
    {
      "epoch": 6.9935064935064934,
      "grad_norm": 2.174851179122925,
      "learning_rate": 8.898220441238534e-05,
      "loss": 1.1343,
      "step": 1077
    },
    {
      "epoch": 7.0,
      "grad_norm": 17985.041015625,
      "learning_rate": 8.896252334690251e-05,
      "loss": 1.1035,
      "step": 1078
    },
    {
      "epoch": 7.0064935064935066,
      "grad_norm": 1.939652442932129,
      "learning_rate": 8.894282689963251e-05,
      "loss": 0.8732,
      "step": 1079
    },
    {
      "epoch": 7.012987012987013,
      "grad_norm": 2.128493547439575,
      "learning_rate": 8.892311507835119e-05,
      "loss": 0.9418,
      "step": 1080
    },
    {
      "epoch": 7.01948051948052,
      "grad_norm": 2.4166202545166016,
      "learning_rate": 8.890338789084045e-05,
      "loss": 0.9604,
      "step": 1081
    },
    {
      "epoch": 7.025974025974026,
      "grad_norm": 1.8071622848510742,
      "learning_rate": 8.888364534488827e-05,
      "loss": 0.8387,
      "step": 1082
    },
    {
      "epoch": 7.032467532467533,
      "grad_norm": 2.023660182952881,
      "learning_rate": 8.886388744828872e-05,
      "loss": 0.9013,
      "step": 1083
    },
    {
      "epoch": 7.038961038961039,
      "grad_norm": 2.2689034938812256,
      "learning_rate": 8.884411420884189e-05,
      "loss": 0.986,
      "step": 1084
    },
    {
      "epoch": 7.045454545454546,
      "grad_norm": 2.1169872283935547,
      "learning_rate": 8.882432563435393e-05,
      "loss": 0.9093,
      "step": 1085
    },
    {
      "epoch": 7.0519480519480515,
      "grad_norm": 2.1221537590026855,
      "learning_rate": 8.880452173263709e-05,
      "loss": 0.9582,
      "step": 1086
    },
    {
      "epoch": 7.058441558441558,
      "grad_norm": 2.1945886611938477,
      "learning_rate": 8.87847025115096e-05,
      "loss": 1.022,
      "step": 1087
    },
    {
      "epoch": 7.064935064935065,
      "grad_norm": 2.1672163009643555,
      "learning_rate": 8.876486797879579e-05,
      "loss": 0.9379,
      "step": 1088
    },
    {
      "epoch": 7.071428571428571,
      "grad_norm": 2.1163387298583984,
      "learning_rate": 8.874501814232603e-05,
      "loss": 0.9828,
      "step": 1089
    },
    {
      "epoch": 7.077922077922078,
      "grad_norm": 1.9327890872955322,
      "learning_rate": 8.872515300993669e-05,
      "loss": 0.8449,
      "step": 1090
    },
    {
      "epoch": 7.084415584415584,
      "grad_norm": 2.355165958404541,
      "learning_rate": 8.870527258947024e-05,
      "loss": 1.016,
      "step": 1091
    },
    {
      "epoch": 7.090909090909091,
      "grad_norm": 2.0632739067077637,
      "learning_rate": 8.868537688877516e-05,
      "loss": 0.8259,
      "step": 1092
    },
    {
      "epoch": 7.097402597402597,
      "grad_norm": 2.1498799324035645,
      "learning_rate": 8.866546591570592e-05,
      "loss": 0.8963,
      "step": 1093
    },
    {
      "epoch": 7.103896103896104,
      "grad_norm": 2.1312801837921143,
      "learning_rate": 8.864553967812309e-05,
      "loss": 0.8811,
      "step": 1094
    },
    {
      "epoch": 7.1103896103896105,
      "grad_norm": 2.255749464035034,
      "learning_rate": 8.862559818389322e-05,
      "loss": 0.9756,
      "step": 1095
    },
    {
      "epoch": 7.116883116883117,
      "grad_norm": 2.2630326747894287,
      "learning_rate": 8.860564144088891e-05,
      "loss": 0.9027,
      "step": 1096
    },
    {
      "epoch": 7.123376623376624,
      "grad_norm": 2.323734998703003,
      "learning_rate": 8.858566945698874e-05,
      "loss": 1.078,
      "step": 1097
    },
    {
      "epoch": 7.12987012987013,
      "grad_norm": 2.2903385162353516,
      "learning_rate": 8.856568224007736e-05,
      "loss": 0.8821,
      "step": 1098
    },
    {
      "epoch": 7.136363636363637,
      "grad_norm": 2.2442169189453125,
      "learning_rate": 8.854567979804538e-05,
      "loss": 0.9962,
      "step": 1099
    },
    {
      "epoch": 7.142857142857143,
      "grad_norm": 2.57180118560791,
      "learning_rate": 8.852566213878947e-05,
      "loss": 1.104,
      "step": 1100
    },
    {
      "epoch": 7.14935064935065,
      "grad_norm": 2.2986552715301514,
      "learning_rate": 8.850562927021227e-05,
      "loss": 0.9048,
      "step": 1101
    },
    {
      "epoch": 7.1558441558441555,
      "grad_norm": 2.231067419052124,
      "learning_rate": 8.848558120022246e-05,
      "loss": 0.9655,
      "step": 1102
    },
    {
      "epoch": 7.162337662337662,
      "grad_norm": 5.123160362243652,
      "learning_rate": 8.846551793673467e-05,
      "loss": 1.0098,
      "step": 1103
    },
    {
      "epoch": 7.1688311688311686,
      "grad_norm": 2.8940987586975098,
      "learning_rate": 8.844543948766958e-05,
      "loss": 0.8868,
      "step": 1104
    },
    {
      "epoch": 7.175324675324675,
      "grad_norm": 2.3510234355926514,
      "learning_rate": 8.842534586095383e-05,
      "loss": 0.9608,
      "step": 1105
    },
    {
      "epoch": 7.181818181818182,
      "grad_norm": 2.365297317504883,
      "learning_rate": 8.840523706452009e-05,
      "loss": 1.0061,
      "step": 1106
    },
    {
      "epoch": 7.188311688311688,
      "grad_norm": 2.305508852005005,
      "learning_rate": 8.838511310630697e-05,
      "loss": 0.9354,
      "step": 1107
    },
    {
      "epoch": 7.194805194805195,
      "grad_norm": 2.5390188694000244,
      "learning_rate": 8.83649739942591e-05,
      "loss": 1.0172,
      "step": 1108
    },
    {
      "epoch": 7.201298701298701,
      "grad_norm": 2.373601198196411,
      "learning_rate": 8.834481973632708e-05,
      "loss": 1.0023,
      "step": 1109
    },
    {
      "epoch": 7.207792207792208,
      "grad_norm": 2.274458408355713,
      "learning_rate": 8.832465034046749e-05,
      "loss": 0.9513,
      "step": 1110
    },
    {
      "epoch": 7.214285714285714,
      "grad_norm": 1.957666039466858,
      "learning_rate": 8.83044658146429e-05,
      "loss": 0.7818,
      "step": 1111
    },
    {
      "epoch": 7.220779220779221,
      "grad_norm": 2.1820261478424072,
      "learning_rate": 8.828426616682183e-05,
      "loss": 0.842,
      "step": 1112
    },
    {
      "epoch": 7.2272727272727275,
      "grad_norm": 2.6345229148864746,
      "learning_rate": 8.826405140497878e-05,
      "loss": 1.0009,
      "step": 1113
    },
    {
      "epoch": 7.233766233766234,
      "grad_norm": 2.6103906631469727,
      "learning_rate": 8.824382153709421e-05,
      "loss": 0.9493,
      "step": 1114
    },
    {
      "epoch": 7.240259740259741,
      "grad_norm": 2.243248701095581,
      "learning_rate": 8.822357657115459e-05,
      "loss": 0.9307,
      "step": 1115
    },
    {
      "epoch": 7.246753246753247,
      "grad_norm": 2.402841091156006,
      "learning_rate": 8.820331651515226e-05,
      "loss": 1.0666,
      "step": 1116
    },
    {
      "epoch": 7.253246753246753,
      "grad_norm": 2.3419620990753174,
      "learning_rate": 8.81830413770856e-05,
      "loss": 0.9186,
      "step": 1117
    },
    {
      "epoch": 7.259740259740259,
      "grad_norm": 2.1303787231445312,
      "learning_rate": 8.816275116495892e-05,
      "loss": 0.9052,
      "step": 1118
    },
    {
      "epoch": 7.266233766233766,
      "grad_norm": 2.5103187561035156,
      "learning_rate": 8.814244588678245e-05,
      "loss": 1.0406,
      "step": 1119
    },
    {
      "epoch": 7.2727272727272725,
      "grad_norm": 2.2306487560272217,
      "learning_rate": 8.81221255505724e-05,
      "loss": 0.8912,
      "step": 1120
    },
    {
      "epoch": 7.279220779220779,
      "grad_norm": 2.2726263999938965,
      "learning_rate": 8.810179016435092e-05,
      "loss": 0.9709,
      "step": 1121
    },
    {
      "epoch": 7.285714285714286,
      "grad_norm": 2.5551421642303467,
      "learning_rate": 8.808143973614611e-05,
      "loss": 1.0897,
      "step": 1122
    },
    {
      "epoch": 7.292207792207792,
      "grad_norm": 2.1128101348876953,
      "learning_rate": 8.806107427399197e-05,
      "loss": 0.8943,
      "step": 1123
    },
    {
      "epoch": 7.298701298701299,
      "grad_norm": 2.3904340267181396,
      "learning_rate": 8.804069378592849e-05,
      "loss": 0.9773,
      "step": 1124
    },
    {
      "epoch": 7.305194805194805,
      "grad_norm": 1.9445910453796387,
      "learning_rate": 8.802029828000156e-05,
      "loss": 0.7842,
      "step": 1125
    },
    {
      "epoch": 7.311688311688312,
      "grad_norm": 2.3209996223449707,
      "learning_rate": 8.799988776426298e-05,
      "loss": 0.989,
      "step": 1126
    },
    {
      "epoch": 7.318181818181818,
      "grad_norm": 2.2647881507873535,
      "learning_rate": 8.797946224677052e-05,
      "loss": 0.9334,
      "step": 1127
    },
    {
      "epoch": 7.324675324675325,
      "grad_norm": 2.4576051235198975,
      "learning_rate": 8.795902173558783e-05,
      "loss": 1.0929,
      "step": 1128
    },
    {
      "epoch": 7.3311688311688314,
      "grad_norm": 2.433586597442627,
      "learning_rate": 8.793856623878452e-05,
      "loss": 1.0337,
      "step": 1129
    },
    {
      "epoch": 7.337662337662338,
      "grad_norm": 2.304011821746826,
      "learning_rate": 8.79180957644361e-05,
      "loss": 0.9771,
      "step": 1130
    },
    {
      "epoch": 7.3441558441558445,
      "grad_norm": 2.250781536102295,
      "learning_rate": 8.789761032062397e-05,
      "loss": 1.0171,
      "step": 1131
    },
    {
      "epoch": 7.35064935064935,
      "grad_norm": 2.306729555130005,
      "learning_rate": 8.787710991543548e-05,
      "loss": 0.8904,
      "step": 1132
    },
    {
      "epoch": 7.357142857142857,
      "grad_norm": 2.406986951828003,
      "learning_rate": 8.785659455696384e-05,
      "loss": 1.0551,
      "step": 1133
    },
    {
      "epoch": 7.363636363636363,
      "grad_norm": 2.177974224090576,
      "learning_rate": 8.783606425330819e-05,
      "loss": 0.9901,
      "step": 1134
    },
    {
      "epoch": 7.37012987012987,
      "grad_norm": 2.277726650238037,
      "learning_rate": 8.78155190125736e-05,
      "loss": 1.0153,
      "step": 1135
    },
    {
      "epoch": 7.376623376623376,
      "grad_norm": 2.368788480758667,
      "learning_rate": 8.7794958842871e-05,
      "loss": 0.9903,
      "step": 1136
    },
    {
      "epoch": 7.383116883116883,
      "grad_norm": 2.5523831844329834,
      "learning_rate": 8.777438375231717e-05,
      "loss": 1.0406,
      "step": 1137
    },
    {
      "epoch": 7.3896103896103895,
      "grad_norm": 2.339541435241699,
      "learning_rate": 8.775379374903487e-05,
      "loss": 1.0202,
      "step": 1138
    },
    {
      "epoch": 7.396103896103896,
      "grad_norm": 2.353381633758545,
      "learning_rate": 8.773318884115273e-05,
      "loss": 0.9538,
      "step": 1139
    },
    {
      "epoch": 7.402597402597403,
      "grad_norm": 2.490757703781128,
      "learning_rate": 8.771256903680519e-05,
      "loss": 1.0989,
      "step": 1140
    },
    {
      "epoch": 7.409090909090909,
      "grad_norm": 2.4964358806610107,
      "learning_rate": 8.769193434413266e-05,
      "loss": 1.0535,
      "step": 1141
    },
    {
      "epoch": 7.415584415584416,
      "grad_norm": 2.534123182296753,
      "learning_rate": 8.767128477128137e-05,
      "loss": 1.008,
      "step": 1142
    },
    {
      "epoch": 7.422077922077922,
      "grad_norm": 2.3385679721832275,
      "learning_rate": 8.765062032640346e-05,
      "loss": 1.0306,
      "step": 1143
    },
    {
      "epoch": 7.428571428571429,
      "grad_norm": 2.352762222290039,
      "learning_rate": 8.76299410176569e-05,
      "loss": 1.037,
      "step": 1144
    },
    {
      "epoch": 7.435064935064935,
      "grad_norm": 2.549403429031372,
      "learning_rate": 8.760924685320557e-05,
      "loss": 1.0304,
      "step": 1145
    },
    {
      "epoch": 7.441558441558442,
      "grad_norm": 2.2719662189483643,
      "learning_rate": 8.758853784121921e-05,
      "loss": 0.9224,
      "step": 1146
    },
    {
      "epoch": 7.448051948051948,
      "grad_norm": 2.1252450942993164,
      "learning_rate": 8.75678139898734e-05,
      "loss": 0.8933,
      "step": 1147
    },
    {
      "epoch": 7.454545454545454,
      "grad_norm": 2.4617741107940674,
      "learning_rate": 8.754707530734957e-05,
      "loss": 1.1581,
      "step": 1148
    },
    {
      "epoch": 7.461038961038961,
      "grad_norm": 2.141832113265991,
      "learning_rate": 8.752632180183505e-05,
      "loss": 0.9494,
      "step": 1149
    },
    {
      "epoch": 7.467532467532467,
      "grad_norm": 2.639061212539673,
      "learning_rate": 8.750555348152298e-05,
      "loss": 1.0582,
      "step": 1150
    },
    {
      "epoch": 7.474025974025974,
      "grad_norm": 1.9012668132781982,
      "learning_rate": 8.748477035461238e-05,
      "loss": 0.8111,
      "step": 1151
    },
    {
      "epoch": 7.48051948051948,
      "grad_norm": 2.2771406173706055,
      "learning_rate": 8.746397242930807e-05,
      "loss": 0.9934,
      "step": 1152
    },
    {
      "epoch": 7.487012987012987,
      "grad_norm": 2.5009937286376953,
      "learning_rate": 8.744315971382078e-05,
      "loss": 1.0519,
      "step": 1153
    },
    {
      "epoch": 7.4935064935064934,
      "grad_norm": 2.192047595977783,
      "learning_rate": 8.7422332216367e-05,
      "loss": 0.9727,
      "step": 1154
    },
    {
      "epoch": 7.5,
      "grad_norm": 2.1458213329315186,
      "learning_rate": 8.740148994516912e-05,
      "loss": 0.9318,
      "step": 1155
    },
    {
      "epoch": 7.5064935064935066,
      "grad_norm": 2.565033435821533,
      "learning_rate": 8.738063290845535e-05,
      "loss": 1.0677,
      "step": 1156
    },
    {
      "epoch": 7.512987012987013,
      "grad_norm": 2.0971720218658447,
      "learning_rate": 8.73597611144597e-05,
      "loss": 0.954,
      "step": 1157
    },
    {
      "epoch": 7.51948051948052,
      "grad_norm": 2.1589765548706055,
      "learning_rate": 8.733887457142202e-05,
      "loss": 1.0059,
      "step": 1158
    },
    {
      "epoch": 7.525974025974026,
      "grad_norm": 2.1708037853240967,
      "learning_rate": 8.7317973287588e-05,
      "loss": 1.0171,
      "step": 1159
    },
    {
      "epoch": 7.532467532467533,
      "grad_norm": 2.249652624130249,
      "learning_rate": 8.729705727120911e-05,
      "loss": 1.043,
      "step": 1160
    },
    {
      "epoch": 7.538961038961039,
      "grad_norm": 2.2765090465545654,
      "learning_rate": 8.727612653054269e-05,
      "loss": 0.9805,
      "step": 1161
    },
    {
      "epoch": 7.545454545454545,
      "grad_norm": 2.298006057739258,
      "learning_rate": 8.725518107385187e-05,
      "loss": 1.0361,
      "step": 1162
    },
    {
      "epoch": 7.551948051948052,
      "grad_norm": 2.2330729961395264,
      "learning_rate": 8.723422090940555e-05,
      "loss": 0.9749,
      "step": 1163
    },
    {
      "epoch": 7.558441558441558,
      "grad_norm": 2.1578798294067383,
      "learning_rate": 8.72132460454785e-05,
      "loss": 1.029,
      "step": 1164
    },
    {
      "epoch": 7.564935064935065,
      "grad_norm": 1.9553289413452148,
      "learning_rate": 8.719225649035126e-05,
      "loss": 0.7988,
      "step": 1165
    },
    {
      "epoch": 7.571428571428571,
      "grad_norm": 2.2159762382507324,
      "learning_rate": 8.717125225231017e-05,
      "loss": 1.062,
      "step": 1166
    },
    {
      "epoch": 7.577922077922078,
      "grad_norm": 1.8637006282806396,
      "learning_rate": 8.715023333964736e-05,
      "loss": 0.8109,
      "step": 1167
    },
    {
      "epoch": 7.584415584415584,
      "grad_norm": 2.314709186553955,
      "learning_rate": 8.712919976066077e-05,
      "loss": 0.9619,
      "step": 1168
    },
    {
      "epoch": 7.590909090909091,
      "grad_norm": 2.3416078090667725,
      "learning_rate": 8.710815152365415e-05,
      "loss": 1.0572,
      "step": 1169
    },
    {
      "epoch": 7.597402597402597,
      "grad_norm": 2.2990005016326904,
      "learning_rate": 8.708708863693697e-05,
      "loss": 0.9534,
      "step": 1170
    },
    {
      "epoch": 7.603896103896104,
      "grad_norm": 2.4119932651519775,
      "learning_rate": 8.706601110882454e-05,
      "loss": 1.0283,
      "step": 1171
    },
    {
      "epoch": 7.6103896103896105,
      "grad_norm": 2.183415174484253,
      "learning_rate": 8.704491894763794e-05,
      "loss": 0.9208,
      "step": 1172
    },
    {
      "epoch": 7.616883116883117,
      "grad_norm": 2.2732317447662354,
      "learning_rate": 8.702381216170403e-05,
      "loss": 0.9305,
      "step": 1173
    },
    {
      "epoch": 7.623376623376624,
      "grad_norm": 2.3167757987976074,
      "learning_rate": 8.700269075935541e-05,
      "loss": 0.9895,
      "step": 1174
    },
    {
      "epoch": 7.62987012987013,
      "grad_norm": 2.2215218544006348,
      "learning_rate": 8.69815547489305e-05,
      "loss": 1.0183,
      "step": 1175
    },
    {
      "epoch": 7.636363636363637,
      "grad_norm": 2.470303773880005,
      "learning_rate": 8.696040413877344e-05,
      "loss": 1.0394,
      "step": 1176
    },
    {
      "epoch": 7.642857142857143,
      "grad_norm": 2.379303216934204,
      "learning_rate": 8.693923893723416e-05,
      "loss": 0.9527,
      "step": 1177
    },
    {
      "epoch": 7.64935064935065,
      "grad_norm": 2.383970022201538,
      "learning_rate": 8.691805915266836e-05,
      "loss": 0.959,
      "step": 1178
    },
    {
      "epoch": 7.6558441558441555,
      "grad_norm": 2.301018238067627,
      "learning_rate": 8.689686479343747e-05,
      "loss": 0.963,
      "step": 1179
    },
    {
      "epoch": 7.662337662337662,
      "grad_norm": 2.6751325130462646,
      "learning_rate": 8.68756558679087e-05,
      "loss": 1.1706,
      "step": 1180
    },
    {
      "epoch": 7.6688311688311686,
      "grad_norm": 2.561551094055176,
      "learning_rate": 8.685443238445499e-05,
      "loss": 1.0545,
      "step": 1181
    },
    {
      "epoch": 7.675324675324675,
      "grad_norm": 2.3014729022979736,
      "learning_rate": 8.683319435145503e-05,
      "loss": 0.917,
      "step": 1182
    },
    {
      "epoch": 7.681818181818182,
      "grad_norm": 2.1952528953552246,
      "learning_rate": 8.681194177729327e-05,
      "loss": 1.0108,
      "step": 1183
    },
    {
      "epoch": 7.688311688311688,
      "grad_norm": 2.403540849685669,
      "learning_rate": 8.679067467035989e-05,
      "loss": 1.0096,
      "step": 1184
    },
    {
      "epoch": 7.694805194805195,
      "grad_norm": 2.3405985832214355,
      "learning_rate": 8.67693930390508e-05,
      "loss": 1.0406,
      "step": 1185
    },
    {
      "epoch": 7.701298701298701,
      "grad_norm": 2.394473075866699,
      "learning_rate": 8.674809689176764e-05,
      "loss": 1.0998,
      "step": 1186
    },
    {
      "epoch": 7.707792207792208,
      "grad_norm": 1.9719188213348389,
      "learning_rate": 8.672678623691783e-05,
      "loss": 0.8589,
      "step": 1187
    },
    {
      "epoch": 7.714285714285714,
      "grad_norm": 2.511230230331421,
      "learning_rate": 8.670546108291443e-05,
      "loss": 1.0682,
      "step": 1188
    },
    {
      "epoch": 7.720779220779221,
      "grad_norm": 2.1016743183135986,
      "learning_rate": 8.668412143817631e-05,
      "loss": 0.9302,
      "step": 1189
    },
    {
      "epoch": 7.7272727272727275,
      "grad_norm": 2.0212161540985107,
      "learning_rate": 8.6662767311128e-05,
      "loss": 1.0233,
      "step": 1190
    },
    {
      "epoch": 7.733766233766234,
      "grad_norm": 2.402103900909424,
      "learning_rate": 8.66413987101998e-05,
      "loss": 1.0854,
      "step": 1191
    },
    {
      "epoch": 7.740259740259741,
      "grad_norm": 2.3115615844726562,
      "learning_rate": 8.662001564382767e-05,
      "loss": 0.9475,
      "step": 1192
    },
    {
      "epoch": 7.746753246753247,
      "grad_norm": 2.4030799865722656,
      "learning_rate": 8.65986181204533e-05,
      "loss": 1.0273,
      "step": 1193
    },
    {
      "epoch": 7.753246753246753,
      "grad_norm": 2.3997647762298584,
      "learning_rate": 8.657720614852411e-05,
      "loss": 1.0452,
      "step": 1194
    },
    {
      "epoch": 7.759740259740259,
      "grad_norm": 2.138275146484375,
      "learning_rate": 8.655577973649321e-05,
      "loss": 0.9578,
      "step": 1195
    },
    {
      "epoch": 7.766233766233766,
      "grad_norm": 2.3623483180999756,
      "learning_rate": 8.65343388928194e-05,
      "loss": 1.028,
      "step": 1196
    },
    {
      "epoch": 7.7727272727272725,
      "grad_norm": 2.2765283584594727,
      "learning_rate": 8.651288362596718e-05,
      "loss": 0.975,
      "step": 1197
    },
    {
      "epoch": 7.779220779220779,
      "grad_norm": 2.779798746109009,
      "learning_rate": 8.649141394440677e-05,
      "loss": 1.0598,
      "step": 1198
    },
    {
      "epoch": 7.785714285714286,
      "grad_norm": 2.338444232940674,
      "learning_rate": 8.646992985661404e-05,
      "loss": 1.0226,
      "step": 1199
    },
    {
      "epoch": 7.792207792207792,
      "grad_norm": 2.3311634063720703,
      "learning_rate": 8.644843137107059e-05,
      "loss": 0.9551,
      "step": 1200
    },
    {
      "epoch": 7.798701298701299,
      "grad_norm": 2.024712324142456,
      "learning_rate": 8.642691849626364e-05,
      "loss": 0.8728,
      "step": 1201
    },
    {
      "epoch": 7.805194805194805,
      "grad_norm": 2.1652753353118896,
      "learning_rate": 8.640539124068617e-05,
      "loss": 0.9638,
      "step": 1202
    },
    {
      "epoch": 7.811688311688312,
      "grad_norm": 2.3495283126831055,
      "learning_rate": 8.638384961283679e-05,
      "loss": 1.0124,
      "step": 1203
    },
    {
      "epoch": 7.818181818181818,
      "grad_norm": 2.242274522781372,
      "learning_rate": 8.63622936212198e-05,
      "loss": 0.992,
      "step": 1204
    },
    {
      "epoch": 7.824675324675325,
      "grad_norm": 2.282271146774292,
      "learning_rate": 8.634072327434515e-05,
      "loss": 1.0438,
      "step": 1205
    },
    {
      "epoch": 7.8311688311688314,
      "grad_norm": 2.7067198753356934,
      "learning_rate": 8.631913858072846e-05,
      "loss": 1.1024,
      "step": 1206
    },
    {
      "epoch": 7.837662337662338,
      "grad_norm": 2.1303887367248535,
      "learning_rate": 8.629753954889107e-05,
      "loss": 1.0117,
      "step": 1207
    },
    {
      "epoch": 7.8441558441558445,
      "grad_norm": 2.1298983097076416,
      "learning_rate": 8.627592618735989e-05,
      "loss": 0.9347,
      "step": 1208
    },
    {
      "epoch": 7.85064935064935,
      "grad_norm": 2.4259676933288574,
      "learning_rate": 8.625429850466756e-05,
      "loss": 1.0624,
      "step": 1209
    },
    {
      "epoch": 7.857142857142857,
      "grad_norm": 2.1696083545684814,
      "learning_rate": 8.623265650935234e-05,
      "loss": 0.9395,
      "step": 1210
    },
    {
      "epoch": 7.863636363636363,
      "grad_norm": 2.418948173522949,
      "learning_rate": 8.621100020995814e-05,
      "loss": 1.1141,
      "step": 1211
    },
    {
      "epoch": 7.87012987012987,
      "grad_norm": 2.3008179664611816,
      "learning_rate": 8.618932961503452e-05,
      "loss": 1.0913,
      "step": 1212
    },
    {
      "epoch": 7.876623376623376,
      "grad_norm": 2.23539662361145,
      "learning_rate": 8.616764473313671e-05,
      "loss": 1.0262,
      "step": 1213
    },
    {
      "epoch": 7.883116883116883,
      "grad_norm": 2.30684232711792,
      "learning_rate": 8.614594557282553e-05,
      "loss": 1.0216,
      "step": 1214
    },
    {
      "epoch": 7.8896103896103895,
      "grad_norm": 2.328866481781006,
      "learning_rate": 8.612423214266749e-05,
      "loss": 1.0333,
      "step": 1215
    },
    {
      "epoch": 7.896103896103896,
      "grad_norm": 2.5381362438201904,
      "learning_rate": 8.61025044512347e-05,
      "loss": 1.0486,
      "step": 1216
    },
    {
      "epoch": 7.902597402597403,
      "grad_norm": 2.353156328201294,
      "learning_rate": 8.60807625071049e-05,
      "loss": 1.0335,
      "step": 1217
    },
    {
      "epoch": 7.909090909090909,
      "grad_norm": 2.1642353534698486,
      "learning_rate": 8.605900631886147e-05,
      "loss": 0.998,
      "step": 1218
    },
    {
      "epoch": 7.915584415584416,
      "grad_norm": 2.3284685611724854,
      "learning_rate": 8.603723589509343e-05,
      "loss": 1.1231,
      "step": 1219
    },
    {
      "epoch": 7.922077922077922,
      "grad_norm": 2.2811686992645264,
      "learning_rate": 8.601545124439535e-05,
      "loss": 1.016,
      "step": 1220
    },
    {
      "epoch": 7.928571428571429,
      "grad_norm": 2.224337339401245,
      "learning_rate": 8.59936523753675e-05,
      "loss": 0.9823,
      "step": 1221
    },
    {
      "epoch": 7.935064935064935,
      "grad_norm": 1.7927359342575073,
      "learning_rate": 8.597183929661573e-05,
      "loss": 0.7452,
      "step": 1222
    },
    {
      "epoch": 7.941558441558442,
      "grad_norm": 2.179928779602051,
      "learning_rate": 8.595001201675147e-05,
      "loss": 1.0548,
      "step": 1223
    },
    {
      "epoch": 7.948051948051948,
      "grad_norm": 2.3533084392547607,
      "learning_rate": 8.592817054439184e-05,
      "loss": 1.0012,
      "step": 1224
    },
    {
      "epoch": 7.954545454545455,
      "grad_norm": 2.246802806854248,
      "learning_rate": 8.590631488815944e-05,
      "loss": 1.0731,
      "step": 1225
    },
    {
      "epoch": 7.961038961038961,
      "grad_norm": 2.3199892044067383,
      "learning_rate": 8.588444505668258e-05,
      "loss": 1.1082,
      "step": 1226
    },
    {
      "epoch": 7.967532467532467,
      "grad_norm": 2.17795991897583,
      "learning_rate": 8.586256105859512e-05,
      "loss": 0.9684,
      "step": 1227
    },
    {
      "epoch": 7.974025974025974,
      "grad_norm": 2.3997833728790283,
      "learning_rate": 8.58406629025365e-05,
      "loss": 1.1021,
      "step": 1228
    },
    {
      "epoch": 7.98051948051948,
      "grad_norm": 2.203303098678589,
      "learning_rate": 8.581875059715176e-05,
      "loss": 1.0425,
      "step": 1229
    },
    {
      "epoch": 7.987012987012987,
      "grad_norm": 2.3763298988342285,
      "learning_rate": 8.579682415109156e-05,
      "loss": 1.0595,
      "step": 1230
    },
    {
      "epoch": 7.9935064935064934,
      "grad_norm": 2.3501322269439697,
      "learning_rate": 8.57748835730121e-05,
      "loss": 1.0559,
      "step": 1231
    },
    {
      "epoch": 8.0,
      "grad_norm": 8490.2294921875,
      "learning_rate": 8.575292887157516e-05,
      "loss": 0.9554,
      "step": 1232
    },
    {
      "epoch": 8.006493506493506,
      "grad_norm": 2.023545742034912,
      "learning_rate": 8.573096005544811e-05,
      "loss": 0.8104,
      "step": 1233
    },
    {
      "epoch": 8.012987012987013,
      "grad_norm": 2.183959484100342,
      "learning_rate": 8.570897713330393e-05,
      "loss": 0.8353,
      "step": 1234
    },
    {
      "epoch": 8.019480519480519,
      "grad_norm": 2.0647823810577393,
      "learning_rate": 8.568698011382107e-05,
      "loss": 0.8269,
      "step": 1235
    },
    {
      "epoch": 8.025974025974026,
      "grad_norm": 2.166212797164917,
      "learning_rate": 8.566496900568363e-05,
      "loss": 0.8408,
      "step": 1236
    },
    {
      "epoch": 8.032467532467532,
      "grad_norm": 2.1798791885375977,
      "learning_rate": 8.564294381758128e-05,
      "loss": 0.9087,
      "step": 1237
    },
    {
      "epoch": 8.03896103896104,
      "grad_norm": 1.9627519845962524,
      "learning_rate": 8.562090455820918e-05,
      "loss": 0.7727,
      "step": 1238
    },
    {
      "epoch": 8.045454545454545,
      "grad_norm": 2.255664587020874,
      "learning_rate": 8.559885123626807e-05,
      "loss": 0.8939,
      "step": 1239
    },
    {
      "epoch": 8.051948051948052,
      "grad_norm": 2.0216453075408936,
      "learning_rate": 8.557678386046427e-05,
      "loss": 0.805,
      "step": 1240
    },
    {
      "epoch": 8.058441558441558,
      "grad_norm": 2.289820432662964,
      "learning_rate": 8.555470243950964e-05,
      "loss": 0.9322,
      "step": 1241
    },
    {
      "epoch": 8.064935064935066,
      "grad_norm": 2.1566131114959717,
      "learning_rate": 8.553260698212155e-05,
      "loss": 0.8221,
      "step": 1242
    },
    {
      "epoch": 8.071428571428571,
      "grad_norm": 2.0849456787109375,
      "learning_rate": 8.551049749702297e-05,
      "loss": 0.8451,
      "step": 1243
    },
    {
      "epoch": 8.077922077922079,
      "grad_norm": 2.2331180572509766,
      "learning_rate": 8.548837399294235e-05,
      "loss": 0.8813,
      "step": 1244
    },
    {
      "epoch": 8.084415584415584,
      "grad_norm": 2.1426122188568115,
      "learning_rate": 8.546623647861371e-05,
      "loss": 0.9262,
      "step": 1245
    },
    {
      "epoch": 8.090909090909092,
      "grad_norm": 2.1544063091278076,
      "learning_rate": 8.544408496277656e-05,
      "loss": 0.8672,
      "step": 1246
    },
    {
      "epoch": 8.097402597402597,
      "grad_norm": 2.375364065170288,
      "learning_rate": 8.542191945417601e-05,
      "loss": 0.8861,
      "step": 1247
    },
    {
      "epoch": 8.103896103896103,
      "grad_norm": 2.2304739952087402,
      "learning_rate": 8.539973996156265e-05,
      "loss": 0.8394,
      "step": 1248
    },
    {
      "epoch": 8.11038961038961,
      "grad_norm": 2.1960930824279785,
      "learning_rate": 8.537754649369255e-05,
      "loss": 0.7973,
      "step": 1249
    },
    {
      "epoch": 8.116883116883116,
      "grad_norm": 2.1149661540985107,
      "learning_rate": 8.535533905932738e-05,
      "loss": 0.8099,
      "step": 1250
    },
    {
      "epoch": 8.123376623376624,
      "grad_norm": 2.2761709690093994,
      "learning_rate": 8.533311766723428e-05,
      "loss": 0.9352,
      "step": 1251
    },
    {
      "epoch": 8.12987012987013,
      "grad_norm": 2.1831071376800537,
      "learning_rate": 8.531088232618588e-05,
      "loss": 0.8521,
      "step": 1252
    },
    {
      "epoch": 8.136363636363637,
      "grad_norm": 2.340667724609375,
      "learning_rate": 8.528863304496035e-05,
      "loss": 0.8881,
      "step": 1253
    },
    {
      "epoch": 8.142857142857142,
      "grad_norm": 2.133786201477051,
      "learning_rate": 8.526636983234135e-05,
      "loss": 0.8583,
      "step": 1254
    },
    {
      "epoch": 8.14935064935065,
      "grad_norm": 2.3610661029815674,
      "learning_rate": 8.524409269711809e-05,
      "loss": 0.9045,
      "step": 1255
    },
    {
      "epoch": 8.155844155844155,
      "grad_norm": 2.1793646812438965,
      "learning_rate": 8.522180164808516e-05,
      "loss": 0.8446,
      "step": 1256
    },
    {
      "epoch": 8.162337662337663,
      "grad_norm": 2.218177080154419,
      "learning_rate": 8.519949669404274e-05,
      "loss": 0.8123,
      "step": 1257
    },
    {
      "epoch": 8.168831168831169,
      "grad_norm": 2.30928111076355,
      "learning_rate": 8.51771778437965e-05,
      "loss": 0.8365,
      "step": 1258
    },
    {
      "epoch": 8.175324675324676,
      "grad_norm": 2.2388057708740234,
      "learning_rate": 8.515484510615753e-05,
      "loss": 0.9075,
      "step": 1259
    },
    {
      "epoch": 8.181818181818182,
      "grad_norm": 2.3225274085998535,
      "learning_rate": 8.513249848994246e-05,
      "loss": 0.9796,
      "step": 1260
    },
    {
      "epoch": 8.188311688311689,
      "grad_norm": 2.455937623977661,
      "learning_rate": 8.511013800397338e-05,
      "loss": 0.9009,
      "step": 1261
    },
    {
      "epoch": 8.194805194805195,
      "grad_norm": 2.265684127807617,
      "learning_rate": 8.508776365707787e-05,
      "loss": 0.8867,
      "step": 1262
    },
    {
      "epoch": 8.2012987012987,
      "grad_norm": 2.2409677505493164,
      "learning_rate": 8.506537545808892e-05,
      "loss": 0.8623,
      "step": 1263
    },
    {
      "epoch": 8.207792207792208,
      "grad_norm": 2.3265388011932373,
      "learning_rate": 8.504297341584508e-05,
      "loss": 0.8919,
      "step": 1264
    },
    {
      "epoch": 8.214285714285714,
      "grad_norm": 2.317533016204834,
      "learning_rate": 8.502055753919032e-05,
      "loss": 0.8108,
      "step": 1265
    },
    {
      "epoch": 8.220779220779221,
      "grad_norm": 2.1003658771514893,
      "learning_rate": 8.499812783697407e-05,
      "loss": 0.7612,
      "step": 1266
    },
    {
      "epoch": 8.227272727272727,
      "grad_norm": 2.0778393745422363,
      "learning_rate": 8.497568431805119e-05,
      "loss": 0.7407,
      "step": 1267
    },
    {
      "epoch": 8.233766233766234,
      "grad_norm": 2.2582011222839355,
      "learning_rate": 8.495322699128205e-05,
      "loss": 0.8541,
      "step": 1268
    },
    {
      "epoch": 8.24025974025974,
      "grad_norm": 2.1436638832092285,
      "learning_rate": 8.493075586553245e-05,
      "loss": 0.8317,
      "step": 1269
    },
    {
      "epoch": 8.246753246753247,
      "grad_norm": 2.144747257232666,
      "learning_rate": 8.490827094967363e-05,
      "loss": 0.8715,
      "step": 1270
    },
    {
      "epoch": 8.253246753246753,
      "grad_norm": 2.3335506916046143,
      "learning_rate": 8.48857722525823e-05,
      "loss": 0.8549,
      "step": 1271
    },
    {
      "epoch": 8.25974025974026,
      "grad_norm": 2.3309953212738037,
      "learning_rate": 8.486325978314055e-05,
      "loss": 0.892,
      "step": 1272
    },
    {
      "epoch": 8.266233766233766,
      "grad_norm": 2.2063148021698,
      "learning_rate": 8.484073355023596e-05,
      "loss": 0.8658,
      "step": 1273
    },
    {
      "epoch": 8.272727272727273,
      "grad_norm": 2.223764419555664,
      "learning_rate": 8.481819356276155e-05,
      "loss": 0.8397,
      "step": 1274
    },
    {
      "epoch": 8.279220779220779,
      "grad_norm": 2.1882009506225586,
      "learning_rate": 8.479563982961573e-05,
      "loss": 0.8788,
      "step": 1275
    },
    {
      "epoch": 8.285714285714286,
      "grad_norm": 2.291719675064087,
      "learning_rate": 8.477307235970236e-05,
      "loss": 0.8885,
      "step": 1276
    },
    {
      "epoch": 8.292207792207792,
      "grad_norm": 2.111360788345337,
      "learning_rate": 8.475049116193071e-05,
      "loss": 0.8755,
      "step": 1277
    },
    {
      "epoch": 8.2987012987013,
      "grad_norm": 2.1871907711029053,
      "learning_rate": 8.472789624521551e-05,
      "loss": 0.7544,
      "step": 1278
    },
    {
      "epoch": 8.305194805194805,
      "grad_norm": 2.198638439178467,
      "learning_rate": 8.470528761847684e-05,
      "loss": 0.8808,
      "step": 1279
    },
    {
      "epoch": 8.311688311688311,
      "grad_norm": 2.108152151107788,
      "learning_rate": 8.468266529064026e-05,
      "loss": 0.8238,
      "step": 1280
    },
    {
      "epoch": 8.318181818181818,
      "grad_norm": 2.1417620182037354,
      "learning_rate": 8.466002927063667e-05,
      "loss": 0.8019,
      "step": 1281
    },
    {
      "epoch": 8.324675324675324,
      "grad_norm": 2.3144867420196533,
      "learning_rate": 8.463737956740245e-05,
      "loss": 0.8671,
      "step": 1282
    },
    {
      "epoch": 8.331168831168831,
      "grad_norm": 1.9930377006530762,
      "learning_rate": 8.461471618987933e-05,
      "loss": 0.7778,
      "step": 1283
    },
    {
      "epoch": 8.337662337662337,
      "grad_norm": 2.3630878925323486,
      "learning_rate": 8.459203914701444e-05,
      "loss": 0.8404,
      "step": 1284
    },
    {
      "epoch": 8.344155844155845,
      "grad_norm": 2.4153730869293213,
      "learning_rate": 8.456934844776032e-05,
      "loss": 0.9375,
      "step": 1285
    },
    {
      "epoch": 8.35064935064935,
      "grad_norm": 2.2773656845092773,
      "learning_rate": 8.454664410107493e-05,
      "loss": 0.9166,
      "step": 1286
    },
    {
      "epoch": 8.357142857142858,
      "grad_norm": 2.1964240074157715,
      "learning_rate": 8.452392611592153e-05,
      "loss": 0.8462,
      "step": 1287
    },
    {
      "epoch": 8.363636363636363,
      "grad_norm": 2.1902146339416504,
      "learning_rate": 8.450119450126887e-05,
      "loss": 0.8701,
      "step": 1288
    },
    {
      "epoch": 8.37012987012987,
      "grad_norm": 2.152127742767334,
      "learning_rate": 8.447844926609103e-05,
      "loss": 0.847,
      "step": 1289
    },
    {
      "epoch": 8.376623376623376,
      "grad_norm": 2.184800386428833,
      "learning_rate": 8.445569041936743e-05,
      "loss": 0.9207,
      "step": 1290
    },
    {
      "epoch": 8.383116883116884,
      "grad_norm": 2.275554895401001,
      "learning_rate": 8.443291797008293e-05,
      "loss": 0.98,
      "step": 1291
    },
    {
      "epoch": 8.38961038961039,
      "grad_norm": 2.2402615547180176,
      "learning_rate": 8.441013192722773e-05,
      "loss": 0.9529,
      "step": 1292
    },
    {
      "epoch": 8.396103896103897,
      "grad_norm": 2.243180513381958,
      "learning_rate": 8.438733229979741e-05,
      "loss": 0.8537,
      "step": 1293
    },
    {
      "epoch": 8.402597402597403,
      "grad_norm": 2.4076995849609375,
      "learning_rate": 8.436451909679287e-05,
      "loss": 0.9193,
      "step": 1294
    },
    {
      "epoch": 8.409090909090908,
      "grad_norm": 2.4312427043914795,
      "learning_rate": 8.434169232722043e-05,
      "loss": 0.8835,
      "step": 1295
    },
    {
      "epoch": 8.415584415584416,
      "grad_norm": 2.348674774169922,
      "learning_rate": 8.431885200009171e-05,
      "loss": 0.9248,
      "step": 1296
    },
    {
      "epoch": 8.422077922077921,
      "grad_norm": 2.3464605808258057,
      "learning_rate": 8.429599812442373e-05,
      "loss": 0.9194,
      "step": 1297
    },
    {
      "epoch": 8.428571428571429,
      "grad_norm": 2.3472964763641357,
      "learning_rate": 8.427313070923885e-05,
      "loss": 0.9237,
      "step": 1298
    },
    {
      "epoch": 8.435064935064934,
      "grad_norm": 2.479236364364624,
      "learning_rate": 8.425024976356474e-05,
      "loss": 1.0045,
      "step": 1299
    },
    {
      "epoch": 8.441558441558442,
      "grad_norm": 2.499112367630005,
      "learning_rate": 8.422735529643444e-05,
      "loss": 0.9375,
      "step": 1300
    },
    {
      "epoch": 8.448051948051948,
      "grad_norm": 2.4231646060943604,
      "learning_rate": 8.420444731688633e-05,
      "loss": 0.9719,
      "step": 1301
    },
    {
      "epoch": 8.454545454545455,
      "grad_norm": 2.3807730674743652,
      "learning_rate": 8.41815258339641e-05,
      "loss": 0.8821,
      "step": 1302
    },
    {
      "epoch": 8.46103896103896,
      "grad_norm": 2.4089229106903076,
      "learning_rate": 8.415859085671683e-05,
      "loss": 0.8724,
      "step": 1303
    },
    {
      "epoch": 8.467532467532468,
      "grad_norm": 2.2512381076812744,
      "learning_rate": 8.413564239419884e-05,
      "loss": 0.8726,
      "step": 1304
    },
    {
      "epoch": 8.474025974025974,
      "grad_norm": 2.4043405055999756,
      "learning_rate": 8.411268045546983e-05,
      "loss": 0.9624,
      "step": 1305
    },
    {
      "epoch": 8.480519480519481,
      "grad_norm": 2.2187204360961914,
      "learning_rate": 8.408970504959483e-05,
      "loss": 0.9099,
      "step": 1306
    },
    {
      "epoch": 8.487012987012987,
      "grad_norm": 2.2931110858917236,
      "learning_rate": 8.406671618564415e-05,
      "loss": 1.0083,
      "step": 1307
    },
    {
      "epoch": 8.493506493506494,
      "grad_norm": 2.2975666522979736,
      "learning_rate": 8.404371387269342e-05,
      "loss": 0.8684,
      "step": 1308
    },
    {
      "epoch": 8.5,
      "grad_norm": 2.3833606243133545,
      "learning_rate": 8.40206981198236e-05,
      "loss": 0.8932,
      "step": 1309
    },
    {
      "epoch": 8.506493506493506,
      "grad_norm": 2.406860589981079,
      "learning_rate": 8.399766893612096e-05,
      "loss": 0.9119,
      "step": 1310
    },
    {
      "epoch": 8.512987012987013,
      "grad_norm": 1.7942426204681396,
      "learning_rate": 8.397462633067705e-05,
      "loss": 0.6829,
      "step": 1311
    },
    {
      "epoch": 8.519480519480519,
      "grad_norm": 2.5286710262298584,
      "learning_rate": 8.395157031258871e-05,
      "loss": 0.9069,
      "step": 1312
    },
    {
      "epoch": 8.525974025974026,
      "grad_norm": 2.38651967048645,
      "learning_rate": 8.392850089095809e-05,
      "loss": 0.8739,
      "step": 1313
    },
    {
      "epoch": 8.532467532467532,
      "grad_norm": 2.5119848251342773,
      "learning_rate": 8.390541807489265e-05,
      "loss": 0.9988,
      "step": 1314
    },
    {
      "epoch": 8.53896103896104,
      "grad_norm": 2.1202709674835205,
      "learning_rate": 8.388232187350512e-05,
      "loss": 0.7731,
      "step": 1315
    },
    {
      "epoch": 8.545454545454545,
      "grad_norm": 2.3501884937286377,
      "learning_rate": 8.38592122959135e-05,
      "loss": 0.9621,
      "step": 1316
    },
    {
      "epoch": 8.551948051948052,
      "grad_norm": 2.2469823360443115,
      "learning_rate": 8.38360893512411e-05,
      "loss": 0.8462,
      "step": 1317
    },
    {
      "epoch": 8.558441558441558,
      "grad_norm": 2.3673346042633057,
      "learning_rate": 8.381295304861647e-05,
      "loss": 0.888,
      "step": 1318
    },
    {
      "epoch": 8.564935064935066,
      "grad_norm": 2.33341908454895,
      "learning_rate": 8.378980339717349e-05,
      "loss": 0.8749,
      "step": 1319
    },
    {
      "epoch": 8.571428571428571,
      "grad_norm": 2.006885290145874,
      "learning_rate": 8.376664040605122e-05,
      "loss": 0.8391,
      "step": 1320
    },
    {
      "epoch": 8.577922077922079,
      "grad_norm": 2.2687089443206787,
      "learning_rate": 8.374346408439411e-05,
      "loss": 0.8874,
      "step": 1321
    },
    {
      "epoch": 8.584415584415584,
      "grad_norm": 2.21208119392395,
      "learning_rate": 8.372027444135176e-05,
      "loss": 0.8589,
      "step": 1322
    },
    {
      "epoch": 8.590909090909092,
      "grad_norm": 1.9771323204040527,
      "learning_rate": 8.36970714860791e-05,
      "loss": 0.7858,
      "step": 1323
    },
    {
      "epoch": 8.597402597402597,
      "grad_norm": 1.8855808973312378,
      "learning_rate": 8.367385522773627e-05,
      "loss": 0.6732,
      "step": 1324
    },
    {
      "epoch": 8.603896103896105,
      "grad_norm": 2.59310245513916,
      "learning_rate": 8.365062567548867e-05,
      "loss": 0.9894,
      "step": 1325
    },
    {
      "epoch": 8.61038961038961,
      "grad_norm": 2.2608039379119873,
      "learning_rate": 8.3627382838507e-05,
      "loss": 0.8949,
      "step": 1326
    },
    {
      "epoch": 8.616883116883116,
      "grad_norm": 2.323054790496826,
      "learning_rate": 8.360412672596712e-05,
      "loss": 0.9194,
      "step": 1327
    },
    {
      "epoch": 8.623376623376624,
      "grad_norm": 1.8838788270950317,
      "learning_rate": 8.358085734705022e-05,
      "loss": 0.6433,
      "step": 1328
    },
    {
      "epoch": 8.62987012987013,
      "grad_norm": 2.2631521224975586,
      "learning_rate": 8.355757471094263e-05,
      "loss": 0.88,
      "step": 1329
    },
    {
      "epoch": 8.636363636363637,
      "grad_norm": 2.5464706420898438,
      "learning_rate": 8.3534278826836e-05,
      "loss": 0.9375,
      "step": 1330
    },
    {
      "epoch": 8.642857142857142,
      "grad_norm": 2.3076460361480713,
      "learning_rate": 8.351096970392717e-05,
      "loss": 0.9402,
      "step": 1331
    },
    {
      "epoch": 8.64935064935065,
      "grad_norm": 1.9361927509307861,
      "learning_rate": 8.348764735141823e-05,
      "loss": 0.6748,
      "step": 1332
    },
    {
      "epoch": 8.655844155844155,
      "grad_norm": 2.359254837036133,
      "learning_rate": 8.346431177851644e-05,
      "loss": 0.9341,
      "step": 1333
    },
    {
      "epoch": 8.662337662337663,
      "grad_norm": 2.4542675018310547,
      "learning_rate": 8.344096299443434e-05,
      "loss": 1.0064,
      "step": 1334
    },
    {
      "epoch": 8.668831168831169,
      "grad_norm": 2.462061882019043,
      "learning_rate": 8.341760100838965e-05,
      "loss": 0.943,
      "step": 1335
    },
    {
      "epoch": 8.675324675324676,
      "grad_norm": 2.2975666522979736,
      "learning_rate": 8.339422582960532e-05,
      "loss": 0.833,
      "step": 1336
    },
    {
      "epoch": 8.681818181818182,
      "grad_norm": 2.5934624671936035,
      "learning_rate": 8.33708374673095e-05,
      "loss": 1.0069,
      "step": 1337
    },
    {
      "epoch": 8.688311688311689,
      "grad_norm": 2.2661728858947754,
      "learning_rate": 8.334743593073553e-05,
      "loss": 0.9142,
      "step": 1338
    },
    {
      "epoch": 8.694805194805195,
      "grad_norm": 2.22139835357666,
      "learning_rate": 8.332402122912198e-05,
      "loss": 0.8557,
      "step": 1339
    },
    {
      "epoch": 8.7012987012987,
      "grad_norm": 2.3808929920196533,
      "learning_rate": 8.330059337171258e-05,
      "loss": 0.9882,
      "step": 1340
    },
    {
      "epoch": 8.707792207792208,
      "grad_norm": 2.285977840423584,
      "learning_rate": 8.327715236775633e-05,
      "loss": 0.9428,
      "step": 1341
    },
    {
      "epoch": 8.714285714285714,
      "grad_norm": 2.046553373336792,
      "learning_rate": 8.32536982265073e-05,
      "loss": 0.8864,
      "step": 1342
    },
    {
      "epoch": 8.720779220779221,
      "grad_norm": 2.305896520614624,
      "learning_rate": 8.323023095722486e-05,
      "loss": 0.9274,
      "step": 1343
    },
    {
      "epoch": 8.727272727272727,
      "grad_norm": 2.6005120277404785,
      "learning_rate": 8.320675056917352e-05,
      "loss": 0.9657,
      "step": 1344
    },
    {
      "epoch": 8.733766233766234,
      "grad_norm": 2.3337910175323486,
      "learning_rate": 8.318325707162293e-05,
      "loss": 0.9651,
      "step": 1345
    },
    {
      "epoch": 8.74025974025974,
      "grad_norm": 2.21706223487854,
      "learning_rate": 8.315975047384798e-05,
      "loss": 0.8629,
      "step": 1346
    },
    {
      "epoch": 8.746753246753247,
      "grad_norm": 2.2574703693389893,
      "learning_rate": 8.313623078512869e-05,
      "loss": 0.854,
      "step": 1347
    },
    {
      "epoch": 8.753246753246753,
      "grad_norm": 2.4945363998413086,
      "learning_rate": 8.311269801475026e-05,
      "loss": 0.923,
      "step": 1348
    },
    {
      "epoch": 8.75974025974026,
      "grad_norm": 2.107628107070923,
      "learning_rate": 8.308915217200306e-05,
      "loss": 0.8478,
      "step": 1349
    },
    {
      "epoch": 8.766233766233766,
      "grad_norm": 2.124215841293335,
      "learning_rate": 8.306559326618259e-05,
      "loss": 0.8439,
      "step": 1350
    },
    {
      "epoch": 8.772727272727273,
      "grad_norm": 2.057939052581787,
      "learning_rate": 8.304202130658959e-05,
      "loss": 0.8302,
      "step": 1351
    },
    {
      "epoch": 8.779220779220779,
      "grad_norm": 2.1522345542907715,
      "learning_rate": 8.301843630252985e-05,
      "loss": 0.8959,
      "step": 1352
    },
    {
      "epoch": 8.785714285714286,
      "grad_norm": 2.0224645137786865,
      "learning_rate": 8.299483826331437e-05,
      "loss": 0.7963,
      "step": 1353
    },
    {
      "epoch": 8.792207792207792,
      "grad_norm": 2.25050950050354,
      "learning_rate": 8.297122719825927e-05,
      "loss": 0.9035,
      "step": 1354
    },
    {
      "epoch": 8.7987012987013,
      "grad_norm": 2.3904287815093994,
      "learning_rate": 8.294760311668586e-05,
      "loss": 0.9492,
      "step": 1355
    },
    {
      "epoch": 8.805194805194805,
      "grad_norm": 2.241687536239624,
      "learning_rate": 8.29239660279205e-05,
      "loss": 0.8868,
      "step": 1356
    },
    {
      "epoch": 8.811688311688311,
      "grad_norm": 2.4004805088043213,
      "learning_rate": 8.29003159412948e-05,
      "loss": 0.9366,
      "step": 1357
    },
    {
      "epoch": 8.818181818181818,
      "grad_norm": 2.3381006717681885,
      "learning_rate": 8.287665286614538e-05,
      "loss": 0.9514,
      "step": 1358
    },
    {
      "epoch": 8.824675324675324,
      "grad_norm": 2.573312520980835,
      "learning_rate": 8.285297681181408e-05,
      "loss": 0.9617,
      "step": 1359
    },
    {
      "epoch": 8.831168831168831,
      "grad_norm": 2.1311933994293213,
      "learning_rate": 8.282928778764783e-05,
      "loss": 0.8366,
      "step": 1360
    },
    {
      "epoch": 8.837662337662337,
      "grad_norm": 2.3406474590301514,
      "learning_rate": 8.280558580299867e-05,
      "loss": 0.9838,
      "step": 1361
    },
    {
      "epoch": 8.844155844155845,
      "grad_norm": 2.5213472843170166,
      "learning_rate": 8.278187086722379e-05,
      "loss": 0.9613,
      "step": 1362
    },
    {
      "epoch": 8.85064935064935,
      "grad_norm": 2.03776216506958,
      "learning_rate": 8.275814298968544e-05,
      "loss": 0.8486,
      "step": 1363
    },
    {
      "epoch": 8.857142857142858,
      "grad_norm": 2.144857406616211,
      "learning_rate": 8.273440217975103e-05,
      "loss": 0.8526,
      "step": 1364
    },
    {
      "epoch": 8.863636363636363,
      "grad_norm": 2.279902935028076,
      "learning_rate": 8.271064844679306e-05,
      "loss": 0.918,
      "step": 1365
    },
    {
      "epoch": 8.87012987012987,
      "grad_norm": 2.033252477645874,
      "learning_rate": 8.268688180018912e-05,
      "loss": 0.8447,
      "step": 1366
    },
    {
      "epoch": 8.876623376623376,
      "grad_norm": 1.9368460178375244,
      "learning_rate": 8.26631022493219e-05,
      "loss": 0.8335,
      "step": 1367
    },
    {
      "epoch": 8.883116883116884,
      "grad_norm": 2.206130027770996,
      "learning_rate": 8.263930980357919e-05,
      "loss": 0.9042,
      "step": 1368
    },
    {
      "epoch": 8.88961038961039,
      "grad_norm": 2.219876289367676,
      "learning_rate": 8.261550447235389e-05,
      "loss": 0.9174,
      "step": 1369
    },
    {
      "epoch": 8.896103896103895,
      "grad_norm": 2.1504364013671875,
      "learning_rate": 8.259168626504395e-05,
      "loss": 0.8749,
      "step": 1370
    },
    {
      "epoch": 8.902597402597403,
      "grad_norm": 2.304591655731201,
      "learning_rate": 8.256785519105241e-05,
      "loss": 0.8961,
      "step": 1371
    },
    {
      "epoch": 8.909090909090908,
      "grad_norm": 2.122875690460205,
      "learning_rate": 8.254401125978743e-05,
      "loss": 0.7802,
      "step": 1372
    },
    {
      "epoch": 8.915584415584416,
      "grad_norm": 2.283535957336426,
      "learning_rate": 8.25201544806622e-05,
      "loss": 0.9022,
      "step": 1373
    },
    {
      "epoch": 8.922077922077921,
      "grad_norm": 2.3608994483947754,
      "learning_rate": 8.249628486309502e-05,
      "loss": 0.8685,
      "step": 1374
    },
    {
      "epoch": 8.928571428571429,
      "grad_norm": 2.3611888885498047,
      "learning_rate": 8.247240241650919e-05,
      "loss": 0.9948,
      "step": 1375
    },
    {
      "epoch": 8.935064935064934,
      "grad_norm": 2.3135151863098145,
      "learning_rate": 8.244850715033317e-05,
      "loss": 0.8179,
      "step": 1376
    },
    {
      "epoch": 8.941558441558442,
      "grad_norm": 2.3506641387939453,
      "learning_rate": 8.242459907400039e-05,
      "loss": 0.9665,
      "step": 1377
    },
    {
      "epoch": 8.948051948051948,
      "grad_norm": 2.306962251663208,
      "learning_rate": 8.240067819694942e-05,
      "loss": 0.9321,
      "step": 1378
    },
    {
      "epoch": 8.954545454545455,
      "grad_norm": 2.266268014907837,
      "learning_rate": 8.237674452862382e-05,
      "loss": 0.9511,
      "step": 1379
    },
    {
      "epoch": 8.96103896103896,
      "grad_norm": 2.541228771209717,
      "learning_rate": 8.235279807847223e-05,
      "loss": 0.9981,
      "step": 1380
    },
    {
      "epoch": 8.967532467532468,
      "grad_norm": 2.1593658924102783,
      "learning_rate": 8.232883885594832e-05,
      "loss": 0.8888,
      "step": 1381
    },
    {
      "epoch": 8.974025974025974,
      "grad_norm": 2.110729455947876,
      "learning_rate": 8.230486687051082e-05,
      "loss": 0.8165,
      "step": 1382
    },
    {
      "epoch": 8.980519480519481,
      "grad_norm": 2.523921489715576,
      "learning_rate": 8.22808821316235e-05,
      "loss": 0.9764,
      "step": 1383
    },
    {
      "epoch": 8.987012987012987,
      "grad_norm": 2.3435425758361816,
      "learning_rate": 8.225688464875514e-05,
      "loss": 0.9116,
      "step": 1384
    },
    {
      "epoch": 8.993506493506494,
      "grad_norm": 2.383795738220215,
      "learning_rate": 8.223287443137957e-05,
      "loss": 0.9476,
      "step": 1385
    },
    {
      "epoch": 9.0,
      "grad_norm": 2654.56103515625,
      "learning_rate": 8.220885148897565e-05,
      "loss": 1.0768,
      "step": 1386
    },
    {
      "epoch": 9.006493506493506,
      "grad_norm": 2.0752618312835693,
      "learning_rate": 8.218481583102726e-05,
      "loss": 0.7959,
      "step": 1387
    },
    {
      "epoch": 9.012987012987013,
      "grad_norm": 2.4385485649108887,
      "learning_rate": 8.216076746702327e-05,
      "loss": 0.8151,
      "step": 1388
    },
    {
      "epoch": 9.019480519480519,
      "grad_norm": 2.270139217376709,
      "learning_rate": 8.213670640645762e-05,
      "loss": 0.8186,
      "step": 1389
    },
    {
      "epoch": 9.025974025974026,
      "grad_norm": 2.112992763519287,
      "learning_rate": 8.211263265882923e-05,
      "loss": 0.7687,
      "step": 1390
    },
    {
      "epoch": 9.032467532467532,
      "grad_norm": 2.1796791553497314,
      "learning_rate": 8.208854623364202e-05,
      "loss": 0.7425,
      "step": 1391
    },
    {
      "epoch": 9.03896103896104,
      "grad_norm": 2.2237792015075684,
      "learning_rate": 8.206444714040495e-05,
      "loss": 0.7597,
      "step": 1392
    },
    {
      "epoch": 9.045454545454545,
      "grad_norm": 2.2954790592193604,
      "learning_rate": 8.204033538863197e-05,
      "loss": 0.7829,
      "step": 1393
    },
    {
      "epoch": 9.051948051948052,
      "grad_norm": 1.7754745483398438,
      "learning_rate": 8.201621098784198e-05,
      "loss": 0.6863,
      "step": 1394
    },
    {
      "epoch": 9.058441558441558,
      "grad_norm": 2.219863176345825,
      "learning_rate": 8.199207394755893e-05,
      "loss": 0.7826,
      "step": 1395
    },
    {
      "epoch": 9.064935064935066,
      "grad_norm": 2.1216280460357666,
      "learning_rate": 8.196792427731175e-05,
      "loss": 0.7987,
      "step": 1396
    },
    {
      "epoch": 9.071428571428571,
      "grad_norm": 2.249668598175049,
      "learning_rate": 8.194376198663434e-05,
      "loss": 0.825,
      "step": 1397
    },
    {
      "epoch": 9.077922077922079,
      "grad_norm": 2.171496629714966,
      "learning_rate": 8.191958708506558e-05,
      "loss": 0.7567,
      "step": 1398
    },
    {
      "epoch": 9.084415584415584,
      "grad_norm": 2.2283010482788086,
      "learning_rate": 8.189539958214935e-05,
      "loss": 0.7638,
      "step": 1399
    },
    {
      "epoch": 9.090909090909092,
      "grad_norm": 2.242703914642334,
      "learning_rate": 8.18711994874345e-05,
      "loss": 0.8351,
      "step": 1400
    },
    {
      "epoch": 9.097402597402597,
      "grad_norm": 2.1934618949890137,
      "learning_rate": 8.184698681047482e-05,
      "loss": 0.7873,
      "step": 1401
    },
    {
      "epoch": 9.103896103896103,
      "grad_norm": 2.263496160507202,
      "learning_rate": 8.18227615608291e-05,
      "loss": 0.7448,
      "step": 1402
    },
    {
      "epoch": 9.11038961038961,
      "grad_norm": 2.06032395362854,
      "learning_rate": 8.179852374806112e-05,
      "loss": 0.7561,
      "step": 1403
    },
    {
      "epoch": 9.116883116883116,
      "grad_norm": 1.9076995849609375,
      "learning_rate": 8.177427338173953e-05,
      "loss": 0.7034,
      "step": 1404
    },
    {
      "epoch": 9.123376623376624,
      "grad_norm": 2.372173547744751,
      "learning_rate": 8.175001047143804e-05,
      "loss": 0.8417,
      "step": 1405
    },
    {
      "epoch": 9.12987012987013,
      "grad_norm": 1.8342652320861816,
      "learning_rate": 8.172573502673523e-05,
      "loss": 0.6436,
      "step": 1406
    },
    {
      "epoch": 9.136363636363637,
      "grad_norm": 2.130004405975342,
      "learning_rate": 8.170144705721466e-05,
      "loss": 0.6889,
      "step": 1407
    },
    {
      "epoch": 9.142857142857142,
      "grad_norm": 2.177459478378296,
      "learning_rate": 8.167714657246486e-05,
      "loss": 0.7952,
      "step": 1408
    },
    {
      "epoch": 9.14935064935065,
      "grad_norm": 2.637460231781006,
      "learning_rate": 8.165283358207924e-05,
      "loss": 0.8493,
      "step": 1409
    },
    {
      "epoch": 9.155844155844155,
      "grad_norm": 2.201258897781372,
      "learning_rate": 8.162850809565623e-05,
      "loss": 0.8234,
      "step": 1410
    },
    {
      "epoch": 9.162337662337663,
      "grad_norm": 2.2699949741363525,
      "learning_rate": 8.160417012279911e-05,
      "loss": 0.8444,
      "step": 1411
    },
    {
      "epoch": 9.168831168831169,
      "grad_norm": 2.2601559162139893,
      "learning_rate": 8.157981967311614e-05,
      "loss": 0.757,
      "step": 1412
    },
    {
      "epoch": 9.175324675324676,
      "grad_norm": 2.369778871536255,
      "learning_rate": 8.15554567562205e-05,
      "loss": 0.8558,
      "step": 1413
    },
    {
      "epoch": 9.181818181818182,
      "grad_norm": 2.1335222721099854,
      "learning_rate": 8.153108138173027e-05,
      "loss": 0.7702,
      "step": 1414
    },
    {
      "epoch": 9.188311688311689,
      "grad_norm": 2.2661004066467285,
      "learning_rate": 8.150669355926846e-05,
      "loss": 0.771,
      "step": 1415
    },
    {
      "epoch": 9.194805194805195,
      "grad_norm": 2.3101413249969482,
      "learning_rate": 8.148229329846301e-05,
      "loss": 0.7875,
      "step": 1416
    },
    {
      "epoch": 9.2012987012987,
      "grad_norm": 2.1832332611083984,
      "learning_rate": 8.145788060894674e-05,
      "loss": 0.7662,
      "step": 1417
    },
    {
      "epoch": 9.207792207792208,
      "grad_norm": 2.008483409881592,
      "learning_rate": 8.143345550035741e-05,
      "loss": 0.7156,
      "step": 1418
    },
    {
      "epoch": 9.214285714285714,
      "grad_norm": 2.1562743186950684,
      "learning_rate": 8.140901798233767e-05,
      "loss": 0.7594,
      "step": 1419
    },
    {
      "epoch": 9.220779220779221,
      "grad_norm": 2.1420395374298096,
      "learning_rate": 8.138456806453503e-05,
      "loss": 0.7728,
      "step": 1420
    },
    {
      "epoch": 9.227272727272727,
      "grad_norm": 2.2543087005615234,
      "learning_rate": 8.136010575660196e-05,
      "loss": 0.7595,
      "step": 1421
    },
    {
      "epoch": 9.233766233766234,
      "grad_norm": 2.3767106533050537,
      "learning_rate": 8.13356310681958e-05,
      "loss": 0.8208,
      "step": 1422
    },
    {
      "epoch": 9.24025974025974,
      "grad_norm": 2.0140974521636963,
      "learning_rate": 8.131114400897874e-05,
      "loss": 0.7136,
      "step": 1423
    },
    {
      "epoch": 9.246753246753247,
      "grad_norm": 2.367492198944092,
      "learning_rate": 8.12866445886179e-05,
      "loss": 0.8795,
      "step": 1424
    },
    {
      "epoch": 9.253246753246753,
      "grad_norm": 2.0226759910583496,
      "learning_rate": 8.126213281678528e-05,
      "loss": 0.7512,
      "step": 1425
    },
    {
      "epoch": 9.25974025974026,
      "grad_norm": 2.336621046066284,
      "learning_rate": 8.123760870315768e-05,
      "loss": 0.8067,
      "step": 1426
    },
    {
      "epoch": 9.266233766233766,
      "grad_norm": 2.2011666297912598,
      "learning_rate": 8.12130722574169e-05,
      "loss": 0.7792,
      "step": 1427
    },
    {
      "epoch": 9.272727272727273,
      "grad_norm": 2.476090669631958,
      "learning_rate": 8.118852348924953e-05,
      "loss": 0.775,
      "step": 1428
    },
    {
      "epoch": 9.279220779220779,
      "grad_norm": 2.2978625297546387,
      "learning_rate": 8.116396240834698e-05,
      "loss": 0.7948,
      "step": 1429
    },
    {
      "epoch": 9.285714285714286,
      "grad_norm": 2.2786600589752197,
      "learning_rate": 8.113938902440564e-05,
      "loss": 0.7556,
      "step": 1430
    },
    {
      "epoch": 9.292207792207792,
      "grad_norm": 2.113959312438965,
      "learning_rate": 8.111480334712665e-05,
      "loss": 0.7449,
      "step": 1431
    },
    {
      "epoch": 9.2987012987013,
      "grad_norm": 2.136590003967285,
      "learning_rate": 8.109020538621606e-05,
      "loss": 0.7521,
      "step": 1432
    },
    {
      "epoch": 9.305194805194805,
      "grad_norm": 2.285116672515869,
      "learning_rate": 8.106559515138477e-05,
      "loss": 0.7869,
      "step": 1433
    },
    {
      "epoch": 9.311688311688311,
      "grad_norm": 2.320920467376709,
      "learning_rate": 8.104097265234848e-05,
      "loss": 0.8478,
      "step": 1434
    },
    {
      "epoch": 9.318181818181818,
      "grad_norm": 2.108393907546997,
      "learning_rate": 8.101633789882781e-05,
      "loss": 0.7533,
      "step": 1435
    },
    {
      "epoch": 9.324675324675324,
      "grad_norm": 2.2300140857696533,
      "learning_rate": 8.099169090054813e-05,
      "loss": 0.7973,
      "step": 1436
    },
    {
      "epoch": 9.331168831168831,
      "grad_norm": 2.217241048812866,
      "learning_rate": 8.096703166723968e-05,
      "loss": 0.784,
      "step": 1437
    },
    {
      "epoch": 9.337662337662337,
      "grad_norm": 2.127923011779785,
      "learning_rate": 8.094236020863757e-05,
      "loss": 0.7696,
      "step": 1438
    },
    {
      "epoch": 9.344155844155845,
      "grad_norm": 2.279656171798706,
      "learning_rate": 8.091767653448167e-05,
      "loss": 0.7762,
      "step": 1439
    },
    {
      "epoch": 9.35064935064935,
      "grad_norm": 2.1744582653045654,
      "learning_rate": 8.089298065451672e-05,
      "loss": 0.7161,
      "step": 1440
    },
    {
      "epoch": 9.357142857142858,
      "grad_norm": 2.01845121383667,
      "learning_rate": 8.086827257849226e-05,
      "loss": 0.6556,
      "step": 1441
    },
    {
      "epoch": 9.363636363636363,
      "grad_norm": 2.32757830619812,
      "learning_rate": 8.084355231616265e-05,
      "loss": 0.8107,
      "step": 1442
    },
    {
      "epoch": 9.37012987012987,
      "grad_norm": 2.1850666999816895,
      "learning_rate": 8.081881987728703e-05,
      "loss": 0.7491,
      "step": 1443
    },
    {
      "epoch": 9.376623376623376,
      "grad_norm": 2.2571535110473633,
      "learning_rate": 8.079407527162944e-05,
      "loss": 0.8356,
      "step": 1444
    },
    {
      "epoch": 9.383116883116884,
      "grad_norm": 2.1496357917785645,
      "learning_rate": 8.076931850895859e-05,
      "loss": 0.7408,
      "step": 1445
    },
    {
      "epoch": 9.38961038961039,
      "grad_norm": 2.1930642127990723,
      "learning_rate": 8.074454959904807e-05,
      "loss": 0.751,
      "step": 1446
    },
    {
      "epoch": 9.396103896103897,
      "grad_norm": 1.8324921131134033,
      "learning_rate": 8.071976855167629e-05,
      "loss": 0.6065,
      "step": 1447
    },
    {
      "epoch": 9.402597402597403,
      "grad_norm": 2.2916312217712402,
      "learning_rate": 8.069497537662639e-05,
      "loss": 0.7501,
      "step": 1448
    },
    {
      "epoch": 9.409090909090908,
      "grad_norm": 2.2877542972564697,
      "learning_rate": 8.067017008368632e-05,
      "loss": 0.8381,
      "step": 1449
    },
    {
      "epoch": 9.415584415584416,
      "grad_norm": 1.8830348253250122,
      "learning_rate": 8.064535268264883e-05,
      "loss": 0.6252,
      "step": 1450
    },
    {
      "epoch": 9.422077922077921,
      "grad_norm": 2.369856834411621,
      "learning_rate": 8.062052318331142e-05,
      "loss": 0.8398,
      "step": 1451
    },
    {
      "epoch": 9.428571428571429,
      "grad_norm": 2.233564615249634,
      "learning_rate": 8.059568159547641e-05,
      "loss": 0.7632,
      "step": 1452
    },
    {
      "epoch": 9.435064935064934,
      "grad_norm": 2.254913806915283,
      "learning_rate": 8.057082792895083e-05,
      "loss": 0.8431,
      "step": 1453
    },
    {
      "epoch": 9.441558441558442,
      "grad_norm": 1.880247712135315,
      "learning_rate": 8.054596219354654e-05,
      "loss": 0.6151,
      "step": 1454
    },
    {
      "epoch": 9.448051948051948,
      "grad_norm": 2.2374584674835205,
      "learning_rate": 8.052108439908013e-05,
      "loss": 0.8184,
      "step": 1455
    },
    {
      "epoch": 9.454545454545455,
      "grad_norm": 2.427372455596924,
      "learning_rate": 8.049619455537296e-05,
      "loss": 0.7961,
      "step": 1456
    },
    {
      "epoch": 9.46103896103896,
      "grad_norm": 1.6904860734939575,
      "learning_rate": 8.047129267225115e-05,
      "loss": 0.5814,
      "step": 1457
    },
    {
      "epoch": 9.467532467532468,
      "grad_norm": 2.3941128253936768,
      "learning_rate": 8.044637875954556e-05,
      "loss": 0.8157,
      "step": 1458
    },
    {
      "epoch": 9.474025974025974,
      "grad_norm": 2.4132771492004395,
      "learning_rate": 8.042145282709182e-05,
      "loss": 0.869,
      "step": 1459
    },
    {
      "epoch": 9.480519480519481,
      "grad_norm": 2.3084611892700195,
      "learning_rate": 8.039651488473028e-05,
      "loss": 0.7596,
      "step": 1460
    },
    {
      "epoch": 9.487012987012987,
      "grad_norm": 2.256843328475952,
      "learning_rate": 8.037156494230604e-05,
      "loss": 0.8579,
      "step": 1461
    },
    {
      "epoch": 9.493506493506494,
      "grad_norm": 2.3451285362243652,
      "learning_rate": 8.034660300966898e-05,
      "loss": 0.8552,
      "step": 1462
    },
    {
      "epoch": 9.5,
      "grad_norm": 2.4393768310546875,
      "learning_rate": 8.032162909667362e-05,
      "loss": 0.8209,
      "step": 1463
    },
    {
      "epoch": 9.506493506493506,
      "grad_norm": 2.1030781269073486,
      "learning_rate": 8.029664321317932e-05,
      "loss": 0.8008,
      "step": 1464
    },
    {
      "epoch": 9.512987012987013,
      "grad_norm": 2.3159143924713135,
      "learning_rate": 8.027164536905008e-05,
      "loss": 0.8277,
      "step": 1465
    },
    {
      "epoch": 9.519480519480519,
      "grad_norm": 2.08125638961792,
      "learning_rate": 8.024663557415464e-05,
      "loss": 0.7845,
      "step": 1466
    },
    {
      "epoch": 9.525974025974026,
      "grad_norm": 2.2004313468933105,
      "learning_rate": 8.022161383836652e-05,
      "loss": 0.8462,
      "step": 1467
    },
    {
      "epoch": 9.532467532467532,
      "grad_norm": 2.0208821296691895,
      "learning_rate": 8.019658017156386e-05,
      "loss": 0.703,
      "step": 1468
    },
    {
      "epoch": 9.53896103896104,
      "grad_norm": 2.086357355117798,
      "learning_rate": 8.017153458362956e-05,
      "loss": 0.7341,
      "step": 1469
    },
    {
      "epoch": 9.545454545454545,
      "grad_norm": 2.44262957572937,
      "learning_rate": 8.014647708445124e-05,
      "loss": 0.8811,
      "step": 1470
    },
    {
      "epoch": 9.551948051948052,
      "grad_norm": 2.260904550552368,
      "learning_rate": 8.01214076839212e-05,
      "loss": 0.7389,
      "step": 1471
    },
    {
      "epoch": 9.558441558441558,
      "grad_norm": 2.211005687713623,
      "learning_rate": 8.009632639193643e-05,
      "loss": 0.7509,
      "step": 1472
    },
    {
      "epoch": 9.564935064935066,
      "grad_norm": 2.503387212753296,
      "learning_rate": 8.007123321839864e-05,
      "loss": 0.8911,
      "step": 1473
    },
    {
      "epoch": 9.571428571428571,
      "grad_norm": 2.088785171508789,
      "learning_rate": 8.00461281732142e-05,
      "loss": 0.8056,
      "step": 1474
    },
    {
      "epoch": 9.577922077922079,
      "grad_norm": 2.1521730422973633,
      "learning_rate": 8.002101126629421e-05,
      "loss": 0.8136,
      "step": 1475
    },
    {
      "epoch": 9.584415584415584,
      "grad_norm": 2.2391040325164795,
      "learning_rate": 7.999588250755442e-05,
      "loss": 0.8104,
      "step": 1476
    },
    {
      "epoch": 9.590909090909092,
      "grad_norm": 2.1446774005889893,
      "learning_rate": 7.997074190691522e-05,
      "loss": 0.8641,
      "step": 1477
    },
    {
      "epoch": 9.597402597402597,
      "grad_norm": 2.460958957672119,
      "learning_rate": 7.994558947430179e-05,
      "loss": 0.8526,
      "step": 1478
    },
    {
      "epoch": 9.603896103896105,
      "grad_norm": 2.084301471710205,
      "learning_rate": 7.992042521964389e-05,
      "loss": 0.7874,
      "step": 1479
    },
    {
      "epoch": 9.61038961038961,
      "grad_norm": 2.258708953857422,
      "learning_rate": 7.989524915287595e-05,
      "loss": 0.8217,
      "step": 1480
    },
    {
      "epoch": 9.616883116883116,
      "grad_norm": 2.2106471061706543,
      "learning_rate": 7.987006128393709e-05,
      "loss": 0.8194,
      "step": 1481
    },
    {
      "epoch": 9.623376623376624,
      "grad_norm": 2.176764488220215,
      "learning_rate": 7.98448616227711e-05,
      "loss": 0.7876,
      "step": 1482
    },
    {
      "epoch": 9.62987012987013,
      "grad_norm": 2.358097791671753,
      "learning_rate": 7.981965017932638e-05,
      "loss": 0.8781,
      "step": 1483
    },
    {
      "epoch": 9.636363636363637,
      "grad_norm": 2.351985216140747,
      "learning_rate": 7.979442696355602e-05,
      "loss": 0.7921,
      "step": 1484
    },
    {
      "epoch": 9.642857142857142,
      "grad_norm": 2.128723382949829,
      "learning_rate": 7.976919198541776e-05,
      "loss": 0.7536,
      "step": 1485
    },
    {
      "epoch": 9.64935064935065,
      "grad_norm": 2.456420660018921,
      "learning_rate": 7.974394525487394e-05,
      "loss": 0.843,
      "step": 1486
    },
    {
      "epoch": 9.655844155844155,
      "grad_norm": 2.167208671569824,
      "learning_rate": 7.971868678189161e-05,
      "loss": 0.7295,
      "step": 1487
    },
    {
      "epoch": 9.662337662337663,
      "grad_norm": 2.31095814704895,
      "learning_rate": 7.969341657644237e-05,
      "loss": 0.7792,
      "step": 1488
    },
    {
      "epoch": 9.668831168831169,
      "grad_norm": 2.448061227798462,
      "learning_rate": 7.966813464850251e-05,
      "loss": 0.9162,
      "step": 1489
    },
    {
      "epoch": 9.675324675324676,
      "grad_norm": 2.2463135719299316,
      "learning_rate": 7.964284100805297e-05,
      "loss": 0.8312,
      "step": 1490
    },
    {
      "epoch": 9.681818181818182,
      "grad_norm": 2.4721384048461914,
      "learning_rate": 7.961753566507924e-05,
      "loss": 0.8892,
      "step": 1491
    },
    {
      "epoch": 9.688311688311689,
      "grad_norm": 2.483168840408325,
      "learning_rate": 7.959221862957148e-05,
      "loss": 0.8688,
      "step": 1492
    },
    {
      "epoch": 9.694805194805195,
      "grad_norm": 2.2723283767700195,
      "learning_rate": 7.956688991152445e-05,
      "loss": 0.8294,
      "step": 1493
    },
    {
      "epoch": 9.7012987012987,
      "grad_norm": 2.0583672523498535,
      "learning_rate": 7.954154952093756e-05,
      "loss": 0.769,
      "step": 1494
    },
    {
      "epoch": 9.707792207792208,
      "grad_norm": 2.319795846939087,
      "learning_rate": 7.951619746781474e-05,
      "loss": 0.8562,
      "step": 1495
    },
    {
      "epoch": 9.714285714285714,
      "grad_norm": 2.2907357215881348,
      "learning_rate": 7.94908337621646e-05,
      "loss": 0.8138,
      "step": 1496
    },
    {
      "epoch": 9.720779220779221,
      "grad_norm": 1.994308590888977,
      "learning_rate": 7.946545841400035e-05,
      "loss": 0.7778,
      "step": 1497
    },
    {
      "epoch": 9.727272727272727,
      "grad_norm": 2.314957618713379,
      "learning_rate": 7.944007143333975e-05,
      "loss": 0.8457,
      "step": 1498
    },
    {
      "epoch": 9.733766233766234,
      "grad_norm": 2.4140167236328125,
      "learning_rate": 7.94146728302052e-05,
      "loss": 0.8849,
      "step": 1499
    },
    {
      "epoch": 9.74025974025974,
      "grad_norm": 2.302248954772949,
      "learning_rate": 7.938926261462367e-05,
      "loss": 0.8035,
      "step": 1500
    },
    {
      "epoch": 9.746753246753247,
      "grad_norm": 2.2283337116241455,
      "learning_rate": 7.936384079662666e-05,
      "loss": 0.754,
      "step": 1501
    },
    {
      "epoch": 9.753246753246753,
      "grad_norm": 2.261359214782715,
      "learning_rate": 7.933840738625036e-05,
      "loss": 0.8227,
      "step": 1502
    },
    {
      "epoch": 9.75974025974026,
      "grad_norm": 2.03810453414917,
      "learning_rate": 7.931296239353546e-05,
      "loss": 0.7251,
      "step": 1503
    },
    {
      "epoch": 9.766233766233766,
      "grad_norm": 2.1554248332977295,
      "learning_rate": 7.928750582852722e-05,
      "loss": 0.7853,
      "step": 1504
    },
    {
      "epoch": 9.772727272727273,
      "grad_norm": 1.9828263521194458,
      "learning_rate": 7.926203770127552e-05,
      "loss": 0.7483,
      "step": 1505
    },
    {
      "epoch": 9.779220779220779,
      "grad_norm": 2.166712522506714,
      "learning_rate": 7.923655802183474e-05,
      "loss": 0.7987,
      "step": 1506
    },
    {
      "epoch": 9.785714285714286,
      "grad_norm": 2.025691270828247,
      "learning_rate": 7.921106680026387e-05,
      "loss": 0.7963,
      "step": 1507
    },
    {
      "epoch": 9.792207792207792,
      "grad_norm": 2.278630256652832,
      "learning_rate": 7.918556404662644e-05,
      "loss": 0.8426,
      "step": 1508
    },
    {
      "epoch": 9.7987012987013,
      "grad_norm": 2.3733408451080322,
      "learning_rate": 7.916004977099055e-05,
      "loss": 0.8032,
      "step": 1509
    },
    {
      "epoch": 9.805194805194805,
      "grad_norm": 2.116952896118164,
      "learning_rate": 7.913452398342881e-05,
      "loss": 0.7179,
      "step": 1510
    },
    {
      "epoch": 9.811688311688311,
      "grad_norm": 2.2615432739257812,
      "learning_rate": 7.91089866940184e-05,
      "loss": 0.8633,
      "step": 1511
    },
    {
      "epoch": 9.818181818181818,
      "grad_norm": 1.9311797618865967,
      "learning_rate": 7.908343791284105e-05,
      "loss": 0.6622,
      "step": 1512
    },
    {
      "epoch": 9.824675324675324,
      "grad_norm": 2.372143268585205,
      "learning_rate": 7.905787764998299e-05,
      "loss": 0.8368,
      "step": 1513
    },
    {
      "epoch": 9.831168831168831,
      "grad_norm": 2.188035488128662,
      "learning_rate": 7.903230591553504e-05,
      "loss": 0.7697,
      "step": 1514
    },
    {
      "epoch": 9.837662337662337,
      "grad_norm": 2.158198118209839,
      "learning_rate": 7.900672271959249e-05,
      "loss": 0.7336,
      "step": 1515
    },
    {
      "epoch": 9.844155844155845,
      "grad_norm": 1.9618550539016724,
      "learning_rate": 7.898112807225517e-05,
      "loss": 0.7297,
      "step": 1516
    },
    {
      "epoch": 9.85064935064935,
      "grad_norm": 2.266169548034668,
      "learning_rate": 7.895552198362748e-05,
      "loss": 0.8173,
      "step": 1517
    },
    {
      "epoch": 9.857142857142858,
      "grad_norm": 2.3293449878692627,
      "learning_rate": 7.892990446381827e-05,
      "loss": 0.8151,
      "step": 1518
    },
    {
      "epoch": 9.863636363636363,
      "grad_norm": 2.265075445175171,
      "learning_rate": 7.890427552294093e-05,
      "loss": 0.8879,
      "step": 1519
    },
    {
      "epoch": 9.87012987012987,
      "grad_norm": 2.1060047149658203,
      "learning_rate": 7.887863517111338e-05,
      "loss": 0.8157,
      "step": 1520
    },
    {
      "epoch": 9.876623376623376,
      "grad_norm": 2.202415704727173,
      "learning_rate": 7.885298341845802e-05,
      "loss": 0.7675,
      "step": 1521
    },
    {
      "epoch": 9.883116883116884,
      "grad_norm": 2.1517200469970703,
      "learning_rate": 7.882732027510174e-05,
      "loss": 0.811,
      "step": 1522
    },
    {
      "epoch": 9.88961038961039,
      "grad_norm": 2.144573211669922,
      "learning_rate": 7.880164575117597e-05,
      "loss": 0.7476,
      "step": 1523
    },
    {
      "epoch": 9.896103896103895,
      "grad_norm": 2.335124969482422,
      "learning_rate": 7.877595985681656e-05,
      "loss": 0.872,
      "step": 1524
    },
    {
      "epoch": 9.902597402597403,
      "grad_norm": 2.477221727371216,
      "learning_rate": 7.875026260216393e-05,
      "loss": 0.9285,
      "step": 1525
    },
    {
      "epoch": 9.909090909090908,
      "grad_norm": 1.9062142372131348,
      "learning_rate": 7.872455399736295e-05,
      "loss": 0.7905,
      "step": 1526
    },
    {
      "epoch": 9.915584415584416,
      "grad_norm": 2.1966278553009033,
      "learning_rate": 7.869883405256295e-05,
      "loss": 0.8347,
      "step": 1527
    },
    {
      "epoch": 9.922077922077921,
      "grad_norm": 2.1748650074005127,
      "learning_rate": 7.867310277791778e-05,
      "loss": 0.7988,
      "step": 1528
    },
    {
      "epoch": 9.928571428571429,
      "grad_norm": 2.3306756019592285,
      "learning_rate": 7.864736018358571e-05,
      "loss": 0.8781,
      "step": 1529
    },
    {
      "epoch": 9.935064935064934,
      "grad_norm": 2.344447135925293,
      "learning_rate": 7.862160627972955e-05,
      "loss": 0.8185,
      "step": 1530
    },
    {
      "epoch": 9.941558441558442,
      "grad_norm": 1.9969489574432373,
      "learning_rate": 7.85958410765165e-05,
      "loss": 0.719,
      "step": 1531
    },
    {
      "epoch": 9.948051948051948,
      "grad_norm": 2.3205368518829346,
      "learning_rate": 7.857006458411826e-05,
      "loss": 0.8242,
      "step": 1532
    },
    {
      "epoch": 9.954545454545455,
      "grad_norm": 2.0827131271362305,
      "learning_rate": 7.854427681271099e-05,
      "loss": 0.7715,
      "step": 1533
    },
    {
      "epoch": 9.96103896103896,
      "grad_norm": 2.0778398513793945,
      "learning_rate": 7.851847777247528e-05,
      "loss": 0.7781,
      "step": 1534
    },
    {
      "epoch": 9.967532467532468,
      "grad_norm": 2.2156105041503906,
      "learning_rate": 7.84926674735962e-05,
      "loss": 0.851,
      "step": 1535
    },
    {
      "epoch": 9.974025974025974,
      "grad_norm": 2.1992075443267822,
      "learning_rate": 7.846684592626323e-05,
      "loss": 0.8263,
      "step": 1536
    },
    {
      "epoch": 9.980519480519481,
      "grad_norm": 2.202113628387451,
      "learning_rate": 7.844101314067032e-05,
      "loss": 0.8584,
      "step": 1537
    },
    {
      "epoch": 9.987012987012987,
      "grad_norm": 2.097062349319458,
      "learning_rate": 7.841516912701585e-05,
      "loss": 0.784,
      "step": 1538
    },
    {
      "epoch": 9.993506493506494,
      "grad_norm": 2.44765305519104,
      "learning_rate": 7.838931389550259e-05,
      "loss": 0.8952,
      "step": 1539
    },
    {
      "epoch": 10.0,
      "grad_norm": 2912.900390625,
      "learning_rate": 7.836344745633783e-05,
      "loss": 0.8677,
      "step": 1540
    },
    {
      "epoch": 10.006493506493506,
      "grad_norm": 1.9397345781326294,
      "learning_rate": 7.83375698197332e-05,
      "loss": 0.6288,
      "step": 1541
    },
    {
      "epoch": 10.012987012987013,
      "grad_norm": 2.1335549354553223,
      "learning_rate": 7.831168099590478e-05,
      "loss": 0.7503,
      "step": 1542
    },
    {
      "epoch": 10.019480519480519,
      "grad_norm": 2.2006771564483643,
      "learning_rate": 7.828578099507308e-05,
      "loss": 0.6708,
      "step": 1543
    },
    {
      "epoch": 10.025974025974026,
      "grad_norm": 1.8945122957229614,
      "learning_rate": 7.8259869827463e-05,
      "loss": 0.6138,
      "step": 1544
    },
    {
      "epoch": 10.032467532467532,
      "grad_norm": 2.2254040241241455,
      "learning_rate": 7.823394750330387e-05,
      "loss": 0.7047,
      "step": 1545
    },
    {
      "epoch": 10.03896103896104,
      "grad_norm": 2.0241615772247314,
      "learning_rate": 7.820801403282939e-05,
      "loss": 0.6689,
      "step": 1546
    },
    {
      "epoch": 10.045454545454545,
      "grad_norm": 2.107599973678589,
      "learning_rate": 7.81820694262777e-05,
      "loss": 0.6813,
      "step": 1547
    },
    {
      "epoch": 10.051948051948052,
      "grad_norm": 2.0900888442993164,
      "learning_rate": 7.815611369389133e-05,
      "loss": 0.6593,
      "step": 1548
    },
    {
      "epoch": 10.058441558441558,
      "grad_norm": 2.0664241313934326,
      "learning_rate": 7.813014684591718e-05,
      "loss": 0.6814,
      "step": 1549
    },
    {
      "epoch": 10.064935064935066,
      "grad_norm": 2.1444430351257324,
      "learning_rate": 7.810416889260653e-05,
      "loss": 0.664,
      "step": 1550
    },
    {
      "epoch": 10.071428571428571,
      "grad_norm": 2.220968246459961,
      "learning_rate": 7.80781798442151e-05,
      "loss": 0.6684,
      "step": 1551
    },
    {
      "epoch": 10.077922077922079,
      "grad_norm": 2.3568577766418457,
      "learning_rate": 7.805217971100295e-05,
      "loss": 0.69,
      "step": 1552
    },
    {
      "epoch": 10.084415584415584,
      "grad_norm": 2.0757124423980713,
      "learning_rate": 7.802616850323449e-05,
      "loss": 0.6463,
      "step": 1553
    },
    {
      "epoch": 10.090909090909092,
      "grad_norm": 2.0199642181396484,
      "learning_rate": 7.800014623117857e-05,
      "loss": 0.639,
      "step": 1554
    },
    {
      "epoch": 10.097402597402597,
      "grad_norm": 2.1582894325256348,
      "learning_rate": 7.797411290510837e-05,
      "loss": 0.7168,
      "step": 1555
    },
    {
      "epoch": 10.103896103896103,
      "grad_norm": 2.1365647315979004,
      "learning_rate": 7.794806853530138e-05,
      "loss": 0.7253,
      "step": 1556
    },
    {
      "epoch": 10.11038961038961,
      "grad_norm": 2.1413962841033936,
      "learning_rate": 7.792201313203957e-05,
      "loss": 0.6625,
      "step": 1557
    },
    {
      "epoch": 10.116883116883116,
      "grad_norm": 2.0579428672790527,
      "learning_rate": 7.789594670560917e-05,
      "loss": 0.6876,
      "step": 1558
    },
    {
      "epoch": 10.123376623376624,
      "grad_norm": 1.8744693994522095,
      "learning_rate": 7.78698692663008e-05,
      "loss": 0.5653,
      "step": 1559
    },
    {
      "epoch": 10.12987012987013,
      "grad_norm": 2.3525032997131348,
      "learning_rate": 7.784378082440941e-05,
      "loss": 0.6893,
      "step": 1560
    },
    {
      "epoch": 10.136363636363637,
      "grad_norm": 2.229753017425537,
      "learning_rate": 7.78176813902343e-05,
      "loss": 0.7107,
      "step": 1561
    },
    {
      "epoch": 10.142857142857142,
      "grad_norm": 2.3776869773864746,
      "learning_rate": 7.779157097407915e-05,
      "loss": 0.7396,
      "step": 1562
    },
    {
      "epoch": 10.14935064935065,
      "grad_norm": 2.364971876144409,
      "learning_rate": 7.77654495862519e-05,
      "loss": 0.7148,
      "step": 1563
    },
    {
      "epoch": 10.155844155844155,
      "grad_norm": 2.2058353424072266,
      "learning_rate": 7.773931723706487e-05,
      "loss": 0.7041,
      "step": 1564
    },
    {
      "epoch": 10.162337662337663,
      "grad_norm": 2.3210866451263428,
      "learning_rate": 7.771317393683471e-05,
      "loss": 0.7472,
      "step": 1565
    },
    {
      "epoch": 10.168831168831169,
      "grad_norm": 2.044942855834961,
      "learning_rate": 7.768701969588237e-05,
      "loss": 0.654,
      "step": 1566
    },
    {
      "epoch": 10.175324675324676,
      "grad_norm": 2.2671778202056885,
      "learning_rate": 7.766085452453312e-05,
      "loss": 0.7198,
      "step": 1567
    },
    {
      "epoch": 10.181818181818182,
      "grad_norm": 2.2623441219329834,
      "learning_rate": 7.763467843311658e-05,
      "loss": 0.6947,
      "step": 1568
    },
    {
      "epoch": 10.188311688311689,
      "grad_norm": 2.348132610321045,
      "learning_rate": 7.760849143196665e-05,
      "loss": 0.7413,
      "step": 1569
    },
    {
      "epoch": 10.194805194805195,
      "grad_norm": 2.34828782081604,
      "learning_rate": 7.758229353142152e-05,
      "loss": 0.7909,
      "step": 1570
    },
    {
      "epoch": 10.2012987012987,
      "grad_norm": 2.1013669967651367,
      "learning_rate": 7.755608474182373e-05,
      "loss": 0.6481,
      "step": 1571
    },
    {
      "epoch": 10.207792207792208,
      "grad_norm": 1.9958415031433105,
      "learning_rate": 7.752986507352008e-05,
      "loss": 0.7038,
      "step": 1572
    },
    {
      "epoch": 10.214285714285714,
      "grad_norm": 2.4060864448547363,
      "learning_rate": 7.75036345368617e-05,
      "loss": 0.7174,
      "step": 1573
    },
    {
      "epoch": 10.220779220779221,
      "grad_norm": 2.2779603004455566,
      "learning_rate": 7.747739314220396e-05,
      "loss": 0.6656,
      "step": 1574
    },
    {
      "epoch": 10.227272727272727,
      "grad_norm": 2.348428249359131,
      "learning_rate": 7.74511408999066e-05,
      "loss": 0.753,
      "step": 1575
    },
    {
      "epoch": 10.233766233766234,
      "grad_norm": 2.427178382873535,
      "learning_rate": 7.742487782033354e-05,
      "loss": 0.7245,
      "step": 1576
    },
    {
      "epoch": 10.24025974025974,
      "grad_norm": 2.1331942081451416,
      "learning_rate": 7.739860391385303e-05,
      "loss": 0.6282,
      "step": 1577
    },
    {
      "epoch": 10.246753246753247,
      "grad_norm": 2.1701390743255615,
      "learning_rate": 7.737231919083761e-05,
      "loss": 0.675,
      "step": 1578
    },
    {
      "epoch": 10.253246753246753,
      "grad_norm": 2.2979564666748047,
      "learning_rate": 7.734602366166406e-05,
      "loss": 0.7126,
      "step": 1579
    },
    {
      "epoch": 10.25974025974026,
      "grad_norm": 2.368772268295288,
      "learning_rate": 7.731971733671346e-05,
      "loss": 0.72,
      "step": 1580
    },
    {
      "epoch": 10.266233766233766,
      "grad_norm": 2.1105165481567383,
      "learning_rate": 7.729340022637112e-05,
      "loss": 0.6476,
      "step": 1581
    },
    {
      "epoch": 10.272727272727273,
      "grad_norm": 2.5445973873138428,
      "learning_rate": 7.726707234102659e-05,
      "loss": 0.8015,
      "step": 1582
    },
    {
      "epoch": 10.279220779220779,
      "grad_norm": 2.3810203075408936,
      "learning_rate": 7.724073369107376e-05,
      "loss": 0.733,
      "step": 1583
    },
    {
      "epoch": 10.285714285714286,
      "grad_norm": 2.166701316833496,
      "learning_rate": 7.721438428691065e-05,
      "loss": 0.7144,
      "step": 1584
    },
    {
      "epoch": 10.292207792207792,
      "grad_norm": 1.981048345565796,
      "learning_rate": 7.718802413893963e-05,
      "loss": 0.6094,
      "step": 1585
    },
    {
      "epoch": 10.2987012987013,
      "grad_norm": 2.206669330596924,
      "learning_rate": 7.716165325756725e-05,
      "loss": 0.6947,
      "step": 1586
    },
    {
      "epoch": 10.305194805194805,
      "grad_norm": 2.164898157119751,
      "learning_rate": 7.713527165320434e-05,
      "loss": 0.7293,
      "step": 1587
    },
    {
      "epoch": 10.311688311688311,
      "grad_norm": 2.202178716659546,
      "learning_rate": 7.710887933626589e-05,
      "loss": 0.6746,
      "step": 1588
    },
    {
      "epoch": 10.318181818181818,
      "grad_norm": 2.296952247619629,
      "learning_rate": 7.708247631717122e-05,
      "loss": 0.7263,
      "step": 1589
    },
    {
      "epoch": 10.324675324675324,
      "grad_norm": 2.1997387409210205,
      "learning_rate": 7.705606260634379e-05,
      "loss": 0.6513,
      "step": 1590
    },
    {
      "epoch": 10.331168831168831,
      "grad_norm": 2.4745490550994873,
      "learning_rate": 7.702963821421133e-05,
      "loss": 0.7661,
      "step": 1591
    },
    {
      "epoch": 10.337662337662337,
      "grad_norm": 2.397670269012451,
      "learning_rate": 7.70032031512058e-05,
      "loss": 0.7387,
      "step": 1592
    },
    {
      "epoch": 10.344155844155845,
      "grad_norm": 2.434361696243286,
      "learning_rate": 7.69767574277633e-05,
      "loss": 0.7354,
      "step": 1593
    },
    {
      "epoch": 10.35064935064935,
      "grad_norm": 1.842675805091858,
      "learning_rate": 7.695030105432417e-05,
      "loss": 0.5535,
      "step": 1594
    },
    {
      "epoch": 10.357142857142858,
      "grad_norm": 2.0731892585754395,
      "learning_rate": 7.692383404133301e-05,
      "loss": 0.6616,
      "step": 1595
    },
    {
      "epoch": 10.363636363636363,
      "grad_norm": 2.039299488067627,
      "learning_rate": 7.689735639923858e-05,
      "loss": 0.6804,
      "step": 1596
    },
    {
      "epoch": 10.37012987012987,
      "grad_norm": 1.9815698862075806,
      "learning_rate": 7.687086813849378e-05,
      "loss": 0.6807,
      "step": 1597
    },
    {
      "epoch": 10.376623376623376,
      "grad_norm": 2.2085583209991455,
      "learning_rate": 7.684436926955582e-05,
      "loss": 0.6732,
      "step": 1598
    },
    {
      "epoch": 10.383116883116884,
      "grad_norm": 2.3801488876342773,
      "learning_rate": 7.6817859802886e-05,
      "loss": 0.7524,
      "step": 1599
    },
    {
      "epoch": 10.38961038961039,
      "grad_norm": 1.944014549255371,
      "learning_rate": 7.679133974894983e-05,
      "loss": 0.6785,
      "step": 1600
    },
    {
      "epoch": 10.396103896103897,
      "grad_norm": 1.7202798128128052,
      "learning_rate": 7.676480911821705e-05,
      "loss": 0.5849,
      "step": 1601
    },
    {
      "epoch": 10.402597402597403,
      "grad_norm": 2.336012601852417,
      "learning_rate": 7.673826792116146e-05,
      "loss": 0.6768,
      "step": 1602
    },
    {
      "epoch": 10.409090909090908,
      "grad_norm": 2.4163119792938232,
      "learning_rate": 7.671171616826119e-05,
      "loss": 0.8147,
      "step": 1603
    },
    {
      "epoch": 10.415584415584416,
      "grad_norm": 2.2862460613250732,
      "learning_rate": 7.668515386999837e-05,
      "loss": 0.7144,
      "step": 1604
    },
    {
      "epoch": 10.422077922077921,
      "grad_norm": 2.4034481048583984,
      "learning_rate": 7.665858103685944e-05,
      "loss": 0.7784,
      "step": 1605
    },
    {
      "epoch": 10.428571428571429,
      "grad_norm": 2.230703592300415,
      "learning_rate": 7.663199767933489e-05,
      "loss": 0.7181,
      "step": 1606
    },
    {
      "epoch": 10.435064935064934,
      "grad_norm": 2.1745998859405518,
      "learning_rate": 7.660540380791942e-05,
      "loss": 0.6556,
      "step": 1607
    },
    {
      "epoch": 10.441558441558442,
      "grad_norm": 2.2904982566833496,
      "learning_rate": 7.65787994331119e-05,
      "loss": 0.7285,
      "step": 1608
    },
    {
      "epoch": 10.448051948051948,
      "grad_norm": 2.110764503479004,
      "learning_rate": 7.655218456541529e-05,
      "loss": 0.642,
      "step": 1609
    },
    {
      "epoch": 10.454545454545455,
      "grad_norm": 2.170912027359009,
      "learning_rate": 7.65255592153367e-05,
      "loss": 0.6743,
      "step": 1610
    },
    {
      "epoch": 10.46103896103896,
      "grad_norm": 2.565704584121704,
      "learning_rate": 7.649892339338744e-05,
      "loss": 0.7778,
      "step": 1611
    },
    {
      "epoch": 10.467532467532468,
      "grad_norm": 2.2946760654449463,
      "learning_rate": 7.647227711008287e-05,
      "loss": 0.754,
      "step": 1612
    },
    {
      "epoch": 10.474025974025974,
      "grad_norm": 2.269430160522461,
      "learning_rate": 7.644562037594254e-05,
      "loss": 0.74,
      "step": 1613
    },
    {
      "epoch": 10.480519480519481,
      "grad_norm": 1.9184123277664185,
      "learning_rate": 7.64189532014901e-05,
      "loss": 0.5831,
      "step": 1614
    },
    {
      "epoch": 10.487012987012987,
      "grad_norm": 2.157374858856201,
      "learning_rate": 7.639227559725332e-05,
      "loss": 0.7455,
      "step": 1615
    },
    {
      "epoch": 10.493506493506494,
      "grad_norm": 2.176115036010742,
      "learning_rate": 7.636558757376413e-05,
      "loss": 0.7088,
      "step": 1616
    },
    {
      "epoch": 10.5,
      "grad_norm": 2.1193859577178955,
      "learning_rate": 7.633888914155852e-05,
      "loss": 0.6377,
      "step": 1617
    },
    {
      "epoch": 10.506493506493506,
      "grad_norm": 2.1090433597564697,
      "learning_rate": 7.631218031117658e-05,
      "loss": 0.6521,
      "step": 1618
    },
    {
      "epoch": 10.512987012987013,
      "grad_norm": 2.215869426727295,
      "learning_rate": 7.628546109316257e-05,
      "loss": 0.7566,
      "step": 1619
    },
    {
      "epoch": 10.519480519480519,
      "grad_norm": 2.28544282913208,
      "learning_rate": 7.62587314980648e-05,
      "loss": 0.7332,
      "step": 1620
    },
    {
      "epoch": 10.525974025974026,
      "grad_norm": 2.1070797443389893,
      "learning_rate": 7.623199153643569e-05,
      "loss": 0.7118,
      "step": 1621
    },
    {
      "epoch": 10.532467532467532,
      "grad_norm": 2.4239230155944824,
      "learning_rate": 7.620524121883174e-05,
      "loss": 0.768,
      "step": 1622
    },
    {
      "epoch": 10.53896103896104,
      "grad_norm": 2.3072590827941895,
      "learning_rate": 7.617848055581361e-05,
      "loss": 0.7294,
      "step": 1623
    },
    {
      "epoch": 10.545454545454545,
      "grad_norm": 2.4551761150360107,
      "learning_rate": 7.615170955794591e-05,
      "loss": 0.8272,
      "step": 1624
    },
    {
      "epoch": 10.551948051948052,
      "grad_norm": 2.317012310028076,
      "learning_rate": 7.612492823579745e-05,
      "loss": 0.7588,
      "step": 1625
    },
    {
      "epoch": 10.558441558441558,
      "grad_norm": 2.036930799484253,
      "learning_rate": 7.609813659994108e-05,
      "loss": 0.6712,
      "step": 1626
    },
    {
      "epoch": 10.564935064935066,
      "grad_norm": 2.1572816371917725,
      "learning_rate": 7.607133466095366e-05,
      "loss": 0.712,
      "step": 1627
    },
    {
      "epoch": 10.571428571428571,
      "grad_norm": 1.9979349374771118,
      "learning_rate": 7.604452242941622e-05,
      "loss": 0.6711,
      "step": 1628
    },
    {
      "epoch": 10.577922077922079,
      "grad_norm": 2.5580434799194336,
      "learning_rate": 7.60176999159138e-05,
      "loss": 0.7787,
      "step": 1629
    },
    {
      "epoch": 10.584415584415584,
      "grad_norm": 2.1501293182373047,
      "learning_rate": 7.599086713103548e-05,
      "loss": 0.7245,
      "step": 1630
    },
    {
      "epoch": 10.590909090909092,
      "grad_norm": 2.2791953086853027,
      "learning_rate": 7.596402408537443e-05,
      "loss": 0.7575,
      "step": 1631
    },
    {
      "epoch": 10.597402597402597,
      "grad_norm": 2.2916219234466553,
      "learning_rate": 7.593717078952788e-05,
      "loss": 0.7194,
      "step": 1632
    },
    {
      "epoch": 10.603896103896105,
      "grad_norm": 2.251642942428589,
      "learning_rate": 7.591030725409707e-05,
      "loss": 0.7858,
      "step": 1633
    },
    {
      "epoch": 10.61038961038961,
      "grad_norm": 2.1227355003356934,
      "learning_rate": 7.588343348968728e-05,
      "loss": 0.7496,
      "step": 1634
    },
    {
      "epoch": 10.616883116883116,
      "grad_norm": 1.8362125158309937,
      "learning_rate": 7.585654950690786e-05,
      "loss": 0.5487,
      "step": 1635
    },
    {
      "epoch": 10.623376623376624,
      "grad_norm": 2.174795150756836,
      "learning_rate": 7.582965531637221e-05,
      "loss": 0.7212,
      "step": 1636
    },
    {
      "epoch": 10.62987012987013,
      "grad_norm": 2.2185142040252686,
      "learning_rate": 7.580275092869766e-05,
      "loss": 0.7792,
      "step": 1637
    },
    {
      "epoch": 10.636363636363637,
      "grad_norm": 2.1538960933685303,
      "learning_rate": 7.577583635450571e-05,
      "loss": 0.7158,
      "step": 1638
    },
    {
      "epoch": 10.642857142857142,
      "grad_norm": 2.1951744556427,
      "learning_rate": 7.574891160442179e-05,
      "loss": 0.7092,
      "step": 1639
    },
    {
      "epoch": 10.64935064935065,
      "grad_norm": 1.9684754610061646,
      "learning_rate": 7.572197668907532e-05,
      "loss": 0.6546,
      "step": 1640
    },
    {
      "epoch": 10.655844155844155,
      "grad_norm": 2.06463885307312,
      "learning_rate": 7.569503161909983e-05,
      "loss": 0.7007,
      "step": 1641
    },
    {
      "epoch": 10.662337662337663,
      "grad_norm": 2.0905561447143555,
      "learning_rate": 7.566807640513278e-05,
      "loss": 0.7559,
      "step": 1642
    },
    {
      "epoch": 10.668831168831169,
      "grad_norm": 2.309191942214966,
      "learning_rate": 7.564111105781568e-05,
      "loss": 0.7564,
      "step": 1643
    },
    {
      "epoch": 10.675324675324676,
      "grad_norm": 2.3112220764160156,
      "learning_rate": 7.561413558779402e-05,
      "loss": 0.7723,
      "step": 1644
    },
    {
      "epoch": 10.681818181818182,
      "grad_norm": 2.321855068206787,
      "learning_rate": 7.558715000571726e-05,
      "loss": 0.7907,
      "step": 1645
    },
    {
      "epoch": 10.688311688311689,
      "grad_norm": 2.232640027999878,
      "learning_rate": 7.55601543222389e-05,
      "loss": 0.794,
      "step": 1646
    },
    {
      "epoch": 10.694805194805195,
      "grad_norm": 1.7704936265945435,
      "learning_rate": 7.553314854801641e-05,
      "loss": 0.545,
      "step": 1647
    },
    {
      "epoch": 10.7012987012987,
      "grad_norm": 2.0758585929870605,
      "learning_rate": 7.550613269371124e-05,
      "loss": 0.7549,
      "step": 1648
    },
    {
      "epoch": 10.707792207792208,
      "grad_norm": 2.173079252243042,
      "learning_rate": 7.547910676998883e-05,
      "loss": 0.7017,
      "step": 1649
    },
    {
      "epoch": 10.714285714285714,
      "grad_norm": 2.047727584838867,
      "learning_rate": 7.545207078751857e-05,
      "loss": 0.6592,
      "step": 1650
    },
    {
      "epoch": 10.720779220779221,
      "grad_norm": 2.197402000427246,
      "learning_rate": 7.542502475697385e-05,
      "loss": 0.6863,
      "step": 1651
    },
    {
      "epoch": 10.727272727272727,
      "grad_norm": 2.0295863151550293,
      "learning_rate": 7.5397968689032e-05,
      "loss": 0.6389,
      "step": 1652
    },
    {
      "epoch": 10.733766233766234,
      "grad_norm": 2.347686529159546,
      "learning_rate": 7.537090259437435e-05,
      "loss": 0.7571,
      "step": 1653
    },
    {
      "epoch": 10.74025974025974,
      "grad_norm": 2.224231719970703,
      "learning_rate": 7.534382648368616e-05,
      "loss": 0.6876,
      "step": 1654
    },
    {
      "epoch": 10.746753246753247,
      "grad_norm": 2.1998109817504883,
      "learning_rate": 7.531674036765664e-05,
      "loss": 0.7749,
      "step": 1655
    },
    {
      "epoch": 10.753246753246753,
      "grad_norm": 2.3216822147369385,
      "learning_rate": 7.528964425697896e-05,
      "loss": 0.7631,
      "step": 1656
    },
    {
      "epoch": 10.75974025974026,
      "grad_norm": 2.265143632888794,
      "learning_rate": 7.526253816235023e-05,
      "loss": 0.762,
      "step": 1657
    },
    {
      "epoch": 10.766233766233766,
      "grad_norm": 2.359100341796875,
      "learning_rate": 7.523542209447152e-05,
      "loss": 0.7892,
      "step": 1658
    },
    {
      "epoch": 10.772727272727273,
      "grad_norm": 2.244819164276123,
      "learning_rate": 7.52082960640478e-05,
      "loss": 0.7531,
      "step": 1659
    },
    {
      "epoch": 10.779220779220779,
      "grad_norm": 1.8464090824127197,
      "learning_rate": 7.518116008178805e-05,
      "loss": 0.6299,
      "step": 1660
    },
    {
      "epoch": 10.785714285714286,
      "grad_norm": 2.5119171142578125,
      "learning_rate": 7.515401415840509e-05,
      "loss": 0.7831,
      "step": 1661
    },
    {
      "epoch": 10.792207792207792,
      "grad_norm": 2.5271401405334473,
      "learning_rate": 7.512685830461568e-05,
      "loss": 0.8111,
      "step": 1662
    },
    {
      "epoch": 10.7987012987013,
      "grad_norm": 2.099280595779419,
      "learning_rate": 7.509969253114055e-05,
      "loss": 0.7015,
      "step": 1663
    },
    {
      "epoch": 10.805194805194805,
      "grad_norm": 2.3430895805358887,
      "learning_rate": 7.507251684870433e-05,
      "loss": 0.7549,
      "step": 1664
    },
    {
      "epoch": 10.811688311688311,
      "grad_norm": 2.0067107677459717,
      "learning_rate": 7.50453312680355e-05,
      "loss": 0.6117,
      "step": 1665
    },
    {
      "epoch": 10.818181818181818,
      "grad_norm": 2.2068874835968018,
      "learning_rate": 7.501813579986656e-05,
      "loss": 0.7337,
      "step": 1666
    },
    {
      "epoch": 10.824675324675324,
      "grad_norm": 2.0831525325775146,
      "learning_rate": 7.499093045493379e-05,
      "loss": 0.6972,
      "step": 1667
    },
    {
      "epoch": 10.831168831168831,
      "grad_norm": 2.1348683834075928,
      "learning_rate": 7.496371524397746e-05,
      "loss": 0.6775,
      "step": 1668
    },
    {
      "epoch": 10.837662337662337,
      "grad_norm": 2.137401580810547,
      "learning_rate": 7.493649017774171e-05,
      "loss": 0.7376,
      "step": 1669
    },
    {
      "epoch": 10.844155844155845,
      "grad_norm": 2.398810625076294,
      "learning_rate": 7.490925526697455e-05,
      "loss": 0.7626,
      "step": 1670
    },
    {
      "epoch": 10.85064935064935,
      "grad_norm": 1.9947322607040405,
      "learning_rate": 7.48820105224279e-05,
      "loss": 0.6033,
      "step": 1671
    },
    {
      "epoch": 10.857142857142858,
      "grad_norm": 2.4584691524505615,
      "learning_rate": 7.485475595485756e-05,
      "loss": 0.806,
      "step": 1672
    },
    {
      "epoch": 10.863636363636363,
      "grad_norm": 2.294435739517212,
      "learning_rate": 7.482749157502318e-05,
      "loss": 0.7604,
      "step": 1673
    },
    {
      "epoch": 10.87012987012987,
      "grad_norm": 2.5067639350891113,
      "learning_rate": 7.480021739368833e-05,
      "loss": 0.7548,
      "step": 1674
    },
    {
      "epoch": 10.876623376623376,
      "grad_norm": 1.932142972946167,
      "learning_rate": 7.477293342162039e-05,
      "loss": 0.6518,
      "step": 1675
    },
    {
      "epoch": 10.883116883116884,
      "grad_norm": 2.36539626121521,
      "learning_rate": 7.474563966959067e-05,
      "loss": 0.7516,
      "step": 1676
    },
    {
      "epoch": 10.88961038961039,
      "grad_norm": 2.2556207180023193,
      "learning_rate": 7.471833614837431e-05,
      "loss": 0.737,
      "step": 1677
    },
    {
      "epoch": 10.896103896103895,
      "grad_norm": 2.3657636642456055,
      "learning_rate": 7.469102286875029e-05,
      "loss": 0.7405,
      "step": 1678
    },
    {
      "epoch": 10.902597402597403,
      "grad_norm": 1.943218469619751,
      "learning_rate": 7.46636998415015e-05,
      "loss": 0.6355,
      "step": 1679
    },
    {
      "epoch": 10.909090909090908,
      "grad_norm": 2.4874868392944336,
      "learning_rate": 7.463636707741458e-05,
      "loss": 0.793,
      "step": 1680
    },
    {
      "epoch": 10.915584415584416,
      "grad_norm": 2.346367597579956,
      "learning_rate": 7.460902458728012e-05,
      "loss": 0.749,
      "step": 1681
    },
    {
      "epoch": 10.922077922077921,
      "grad_norm": 2.232698678970337,
      "learning_rate": 7.458167238189248e-05,
      "loss": 0.7215,
      "step": 1682
    },
    {
      "epoch": 10.928571428571429,
      "grad_norm": 2.218801259994507,
      "learning_rate": 7.455431047204988e-05,
      "loss": 0.782,
      "step": 1683
    },
    {
      "epoch": 10.935064935064934,
      "grad_norm": 2.312042713165283,
      "learning_rate": 7.452693886855438e-05,
      "loss": 0.7656,
      "step": 1684
    },
    {
      "epoch": 10.941558441558442,
      "grad_norm": 2.2011115550994873,
      "learning_rate": 7.449955758221183e-05,
      "loss": 0.7446,
      "step": 1685
    },
    {
      "epoch": 10.948051948051948,
      "grad_norm": 2.228243350982666,
      "learning_rate": 7.447216662383196e-05,
      "loss": 0.6661,
      "step": 1686
    },
    {
      "epoch": 10.954545454545455,
      "grad_norm": 2.1731884479522705,
      "learning_rate": 7.444476600422828e-05,
      "loss": 0.7674,
      "step": 1687
    },
    {
      "epoch": 10.96103896103896,
      "grad_norm": 2.086169481277466,
      "learning_rate": 7.441735573421809e-05,
      "loss": 0.7016,
      "step": 1688
    },
    {
      "epoch": 10.967532467532468,
      "grad_norm": 2.0726120471954346,
      "learning_rate": 7.438993582462256e-05,
      "loss": 0.6815,
      "step": 1689
    },
    {
      "epoch": 10.974025974025974,
      "grad_norm": 2.0607035160064697,
      "learning_rate": 7.436250628626662e-05,
      "loss": 0.7075,
      "step": 1690
    },
    {
      "epoch": 10.980519480519481,
      "grad_norm": 2.23492169380188,
      "learning_rate": 7.433506712997904e-05,
      "loss": 0.7345,
      "step": 1691
    },
    {
      "epoch": 10.987012987012987,
      "grad_norm": 2.3279659748077393,
      "learning_rate": 7.430761836659235e-05,
      "loss": 0.7348,
      "step": 1692
    },
    {
      "epoch": 10.993506493506494,
      "grad_norm": 2.2870876789093018,
      "learning_rate": 7.428016000694286e-05,
      "loss": 0.7223,
      "step": 1693
    },
    {
      "epoch": 11.0,
      "grad_norm": 1362.704833984375,
      "learning_rate": 7.425269206187075e-05,
      "loss": 0.7249,
      "step": 1694
    },
    {
      "epoch": 11.006493506493506,
      "grad_norm": 1.971757173538208,
      "learning_rate": 7.42252145422199e-05,
      "loss": 0.5723,
      "step": 1695
    },
    {
      "epoch": 11.012987012987013,
      "grad_norm": 2.09525203704834,
      "learning_rate": 7.4197727458838e-05,
      "loss": 0.6099,
      "step": 1696
    },
    {
      "epoch": 11.019480519480519,
      "grad_norm": 1.9457815885543823,
      "learning_rate": 7.417023082257652e-05,
      "loss": 0.5656,
      "step": 1697
    },
    {
      "epoch": 11.025974025974026,
      "grad_norm": 2.0404152870178223,
      "learning_rate": 7.414272464429068e-05,
      "loss": 0.6151,
      "step": 1698
    },
    {
      "epoch": 11.032467532467532,
      "grad_norm": 2.177133560180664,
      "learning_rate": 7.411520893483951e-05,
      "loss": 0.6507,
      "step": 1699
    },
    {
      "epoch": 11.03896103896104,
      "grad_norm": 2.029690980911255,
      "learning_rate": 7.408768370508576e-05,
      "loss": 0.6013,
      "step": 1700
    },
    {
      "epoch": 11.045454545454545,
      "grad_norm": 2.175691604614258,
      "learning_rate": 7.406014896589597e-05,
      "loss": 0.6494,
      "step": 1701
    },
    {
      "epoch": 11.051948051948052,
      "grad_norm": 1.9023364782333374,
      "learning_rate": 7.403260472814039e-05,
      "loss": 0.5249,
      "step": 1702
    },
    {
      "epoch": 11.058441558441558,
      "grad_norm": 1.9307712316513062,
      "learning_rate": 7.400505100269308e-05,
      "loss": 0.605,
      "step": 1703
    },
    {
      "epoch": 11.064935064935066,
      "grad_norm": 2.0047030448913574,
      "learning_rate": 7.397748780043179e-05,
      "loss": 0.593,
      "step": 1704
    },
    {
      "epoch": 11.071428571428571,
      "grad_norm": 2.054450750350952,
      "learning_rate": 7.394991513223806e-05,
      "loss": 0.6133,
      "step": 1705
    },
    {
      "epoch": 11.077922077922079,
      "grad_norm": 2.20436954498291,
      "learning_rate": 7.392233300899712e-05,
      "loss": 0.6176,
      "step": 1706
    },
    {
      "epoch": 11.084415584415584,
      "grad_norm": 2.08552622795105,
      "learning_rate": 7.389474144159796e-05,
      "loss": 0.5731,
      "step": 1707
    },
    {
      "epoch": 11.090909090909092,
      "grad_norm": 1.974770188331604,
      "learning_rate": 7.38671404409333e-05,
      "loss": 0.5748,
      "step": 1708
    },
    {
      "epoch": 11.097402597402597,
      "grad_norm": 2.1373484134674072,
      "learning_rate": 7.38395300178996e-05,
      "loss": 0.6429,
      "step": 1709
    },
    {
      "epoch": 11.103896103896103,
      "grad_norm": 2.1213269233703613,
      "learning_rate": 7.381191018339696e-05,
      "loss": 0.6421,
      "step": 1710
    },
    {
      "epoch": 11.11038961038961,
      "grad_norm": 2.0944926738739014,
      "learning_rate": 7.378428094832931e-05,
      "loss": 0.6206,
      "step": 1711
    },
    {
      "epoch": 11.116883116883116,
      "grad_norm": 2.2668039798736572,
      "learning_rate": 7.37566423236042e-05,
      "loss": 0.658,
      "step": 1712
    },
    {
      "epoch": 11.123376623376624,
      "grad_norm": 1.8874119520187378,
      "learning_rate": 7.372899432013294e-05,
      "loss": 0.5201,
      "step": 1713
    },
    {
      "epoch": 11.12987012987013,
      "grad_norm": 2.1221206188201904,
      "learning_rate": 7.370133694883051e-05,
      "loss": 0.6018,
      "step": 1714
    },
    {
      "epoch": 11.136363636363637,
      "grad_norm": 1.8658677339553833,
      "learning_rate": 7.36736702206156e-05,
      "loss": 0.5639,
      "step": 1715
    },
    {
      "epoch": 11.142857142857142,
      "grad_norm": 2.2183215618133545,
      "learning_rate": 7.364599414641064e-05,
      "loss": 0.667,
      "step": 1716
    },
    {
      "epoch": 11.14935064935065,
      "grad_norm": 2.2229409217834473,
      "learning_rate": 7.361830873714165e-05,
      "loss": 0.651,
      "step": 1717
    },
    {
      "epoch": 11.155844155844155,
      "grad_norm": 2.024958610534668,
      "learning_rate": 7.35906140037384e-05,
      "loss": 0.5851,
      "step": 1718
    },
    {
      "epoch": 11.162337662337663,
      "grad_norm": 2.61043643951416,
      "learning_rate": 7.356290995713437e-05,
      "loss": 0.7094,
      "step": 1719
    },
    {
      "epoch": 11.168831168831169,
      "grad_norm": 1.994360089302063,
      "learning_rate": 7.353519660826665e-05,
      "loss": 0.597,
      "step": 1720
    },
    {
      "epoch": 11.175324675324676,
      "grad_norm": 2.4047257900238037,
      "learning_rate": 7.350747396807601e-05,
      "loss": 0.6236,
      "step": 1721
    },
    {
      "epoch": 11.181818181818182,
      "grad_norm": 2.2192437648773193,
      "learning_rate": 7.347974204750696e-05,
      "loss": 0.6513,
      "step": 1722
    },
    {
      "epoch": 11.188311688311689,
      "grad_norm": 2.100661039352417,
      "learning_rate": 7.345200085750757e-05,
      "loss": 0.6196,
      "step": 1723
    },
    {
      "epoch": 11.194805194805195,
      "grad_norm": 1.9950624704360962,
      "learning_rate": 7.342425040902967e-05,
      "loss": 0.588,
      "step": 1724
    },
    {
      "epoch": 11.2012987012987,
      "grad_norm": 2.0585994720458984,
      "learning_rate": 7.339649071302867e-05,
      "loss": 0.6089,
      "step": 1725
    },
    {
      "epoch": 11.207792207792208,
      "grad_norm": 2.2247488498687744,
      "learning_rate": 7.336872178046368e-05,
      "loss": 0.702,
      "step": 1726
    },
    {
      "epoch": 11.214285714285714,
      "grad_norm": 2.0867512226104736,
      "learning_rate": 7.334094362229739e-05,
      "loss": 0.5719,
      "step": 1727
    },
    {
      "epoch": 11.220779220779221,
      "grad_norm": 2.1678314208984375,
      "learning_rate": 7.331315624949625e-05,
      "loss": 0.6033,
      "step": 1728
    },
    {
      "epoch": 11.227272727272727,
      "grad_norm": 1.8890186548233032,
      "learning_rate": 7.32853596730302e-05,
      "loss": 0.5229,
      "step": 1729
    },
    {
      "epoch": 11.233766233766234,
      "grad_norm": 1.9374103546142578,
      "learning_rate": 7.325755390387292e-05,
      "loss": 0.5743,
      "step": 1730
    },
    {
      "epoch": 11.24025974025974,
      "grad_norm": 2.0223710536956787,
      "learning_rate": 7.32297389530017e-05,
      "loss": 0.5544,
      "step": 1731
    },
    {
      "epoch": 11.246753246753247,
      "grad_norm": 1.9823747873306274,
      "learning_rate": 7.320191483139742e-05,
      "loss": 0.5758,
      "step": 1732
    },
    {
      "epoch": 11.253246753246753,
      "grad_norm": 2.0667495727539062,
      "learning_rate": 7.317408155004462e-05,
      "loss": 0.5883,
      "step": 1733
    },
    {
      "epoch": 11.25974025974026,
      "grad_norm": 2.946258306503296,
      "learning_rate": 7.314623911993142e-05,
      "loss": 0.6296,
      "step": 1734
    },
    {
      "epoch": 11.266233766233766,
      "grad_norm": 2.1733124256134033,
      "learning_rate": 7.311838755204959e-05,
      "loss": 0.6713,
      "step": 1735
    },
    {
      "epoch": 11.272727272727273,
      "grad_norm": 2.237227201461792,
      "learning_rate": 7.309052685739448e-05,
      "loss": 0.6404,
      "step": 1736
    },
    {
      "epoch": 11.279220779220779,
      "grad_norm": 2.2390263080596924,
      "learning_rate": 7.306265704696504e-05,
      "loss": 0.65,
      "step": 1737
    },
    {
      "epoch": 11.285714285714286,
      "grad_norm": 1.8883891105651855,
      "learning_rate": 7.303477813176385e-05,
      "loss": 0.5331,
      "step": 1738
    },
    {
      "epoch": 11.292207792207792,
      "grad_norm": 1.9942971467971802,
      "learning_rate": 7.300689012279706e-05,
      "loss": 0.5833,
      "step": 1739
    },
    {
      "epoch": 11.2987012987013,
      "grad_norm": 2.324103355407715,
      "learning_rate": 7.297899303107441e-05,
      "loss": 0.6633,
      "step": 1740
    },
    {
      "epoch": 11.305194805194805,
      "grad_norm": 2.0713562965393066,
      "learning_rate": 7.29510868676092e-05,
      "loss": 0.6185,
      "step": 1741
    },
    {
      "epoch": 11.311688311688311,
      "grad_norm": 2.018885612487793,
      "learning_rate": 7.29231716434184e-05,
      "loss": 0.6032,
      "step": 1742
    },
    {
      "epoch": 11.318181818181818,
      "grad_norm": 2.5611448287963867,
      "learning_rate": 7.289524736952245e-05,
      "loss": 0.7274,
      "step": 1743
    },
    {
      "epoch": 11.324675324675324,
      "grad_norm": 2.0800020694732666,
      "learning_rate": 7.286731405694544e-05,
      "loss": 0.6414,
      "step": 1744
    },
    {
      "epoch": 11.331168831168831,
      "grad_norm": 1.932077169418335,
      "learning_rate": 7.283937171671498e-05,
      "loss": 0.5886,
      "step": 1745
    },
    {
      "epoch": 11.337662337662337,
      "grad_norm": 2.0475950241088867,
      "learning_rate": 7.281142035986227e-05,
      "loss": 0.6354,
      "step": 1746
    },
    {
      "epoch": 11.344155844155845,
      "grad_norm": 2.118962287902832,
      "learning_rate": 7.278345999742208e-05,
      "loss": 0.6537,
      "step": 1747
    },
    {
      "epoch": 11.35064935064935,
      "grad_norm": 2.1595616340637207,
      "learning_rate": 7.275549064043268e-05,
      "loss": 0.6664,
      "step": 1748
    },
    {
      "epoch": 11.357142857142858,
      "grad_norm": 2.058976411819458,
      "learning_rate": 7.272751229993598e-05,
      "loss": 0.5731,
      "step": 1749
    },
    {
      "epoch": 11.363636363636363,
      "grad_norm": 2.0019469261169434,
      "learning_rate": 7.269952498697734e-05,
      "loss": 0.5478,
      "step": 1750
    },
    {
      "epoch": 11.37012987012987,
      "grad_norm": 1.9929718971252441,
      "learning_rate": 7.267152871260573e-05,
      "loss": 0.566,
      "step": 1751
    },
    {
      "epoch": 11.376623376623376,
      "grad_norm": 2.4305410385131836,
      "learning_rate": 7.264352348787364e-05,
      "loss": 0.6865,
      "step": 1752
    },
    {
      "epoch": 11.383116883116884,
      "grad_norm": 2.2349486351013184,
      "learning_rate": 7.261550932383707e-05,
      "loss": 0.5859,
      "step": 1753
    },
    {
      "epoch": 11.38961038961039,
      "grad_norm": 2.297055959701538,
      "learning_rate": 7.258748623155558e-05,
      "loss": 0.6629,
      "step": 1754
    },
    {
      "epoch": 11.396103896103897,
      "grad_norm": 2.401366949081421,
      "learning_rate": 7.255945422209227e-05,
      "loss": 0.7201,
      "step": 1755
    },
    {
      "epoch": 11.402597402597403,
      "grad_norm": 2.1199653148651123,
      "learning_rate": 7.253141330651368e-05,
      "loss": 0.6021,
      "step": 1756
    },
    {
      "epoch": 11.409090909090908,
      "grad_norm": 2.2622432708740234,
      "learning_rate": 7.250336349588994e-05,
      "loss": 0.6528,
      "step": 1757
    },
    {
      "epoch": 11.415584415584416,
      "grad_norm": 2.107171058654785,
      "learning_rate": 7.247530480129469e-05,
      "loss": 0.6177,
      "step": 1758
    },
    {
      "epoch": 11.422077922077921,
      "grad_norm": 2.6965367794036865,
      "learning_rate": 7.244723723380505e-05,
      "loss": 0.724,
      "step": 1759
    },
    {
      "epoch": 11.428571428571429,
      "grad_norm": 2.36151385307312,
      "learning_rate": 7.241916080450163e-05,
      "loss": 0.6813,
      "step": 1760
    },
    {
      "epoch": 11.435064935064934,
      "grad_norm": 2.007305860519409,
      "learning_rate": 7.239107552446857e-05,
      "loss": 0.6151,
      "step": 1761
    },
    {
      "epoch": 11.441558441558442,
      "grad_norm": 2.315720319747925,
      "learning_rate": 7.236298140479351e-05,
      "loss": 0.6959,
      "step": 1762
    },
    {
      "epoch": 11.448051948051948,
      "grad_norm": 2.2148401737213135,
      "learning_rate": 7.233487845656754e-05,
      "loss": 0.6723,
      "step": 1763
    },
    {
      "epoch": 11.454545454545455,
      "grad_norm": 2.250305414199829,
      "learning_rate": 7.23067666908853e-05,
      "loss": 0.6875,
      "step": 1764
    },
    {
      "epoch": 11.46103896103896,
      "grad_norm": 2.0240795612335205,
      "learning_rate": 7.227864611884483e-05,
      "loss": 0.6607,
      "step": 1765
    },
    {
      "epoch": 11.467532467532468,
      "grad_norm": 2.3691110610961914,
      "learning_rate": 7.225051675154767e-05,
      "loss": 0.7038,
      "step": 1766
    },
    {
      "epoch": 11.474025974025974,
      "grad_norm": 2.293552875518799,
      "learning_rate": 7.222237860009892e-05,
      "loss": 0.7139,
      "step": 1767
    },
    {
      "epoch": 11.480519480519481,
      "grad_norm": 2.2120654582977295,
      "learning_rate": 7.219423167560701e-05,
      "loss": 0.6508,
      "step": 1768
    },
    {
      "epoch": 11.487012987012987,
      "grad_norm": 1.8835054636001587,
      "learning_rate": 7.216607598918393e-05,
      "loss": 0.4962,
      "step": 1769
    },
    {
      "epoch": 11.493506493506494,
      "grad_norm": 2.0173633098602295,
      "learning_rate": 7.21379115519451e-05,
      "loss": 0.5846,
      "step": 1770
    },
    {
      "epoch": 11.5,
      "grad_norm": 2.0470130443573,
      "learning_rate": 7.210973837500937e-05,
      "loss": 0.6357,
      "step": 1771
    },
    {
      "epoch": 11.506493506493506,
      "grad_norm": 2.2059829235076904,
      "learning_rate": 7.208155646949908e-05,
      "loss": 0.6786,
      "step": 1772
    },
    {
      "epoch": 11.512987012987013,
      "grad_norm": 2.187005043029785,
      "learning_rate": 7.205336584653999e-05,
      "loss": 0.6637,
      "step": 1773
    },
    {
      "epoch": 11.519480519480519,
      "grad_norm": 2.1610119342803955,
      "learning_rate": 7.202516651726134e-05,
      "loss": 0.6523,
      "step": 1774
    },
    {
      "epoch": 11.525974025974026,
      "grad_norm": 2.28117036819458,
      "learning_rate": 7.199695849279576e-05,
      "loss": 0.6649,
      "step": 1775
    },
    {
      "epoch": 11.532467532467532,
      "grad_norm": 2.059276580810547,
      "learning_rate": 7.196874178427933e-05,
      "loss": 0.6081,
      "step": 1776
    },
    {
      "epoch": 11.53896103896104,
      "grad_norm": 2.2199676036834717,
      "learning_rate": 7.194051640285157e-05,
      "loss": 0.6859,
      "step": 1777
    },
    {
      "epoch": 11.545454545454545,
      "grad_norm": 2.2723867893218994,
      "learning_rate": 7.191228235965538e-05,
      "loss": 0.6717,
      "step": 1778
    },
    {
      "epoch": 11.551948051948052,
      "grad_norm": 2.3416223526000977,
      "learning_rate": 7.188403966583716e-05,
      "loss": 0.6552,
      "step": 1779
    },
    {
      "epoch": 11.558441558441558,
      "grad_norm": 2.1369590759277344,
      "learning_rate": 7.185578833254664e-05,
      "loss": 0.6509,
      "step": 1780
    },
    {
      "epoch": 11.564935064935066,
      "grad_norm": 2.2083840370178223,
      "learning_rate": 7.182752837093702e-05,
      "loss": 0.6168,
      "step": 1781
    },
    {
      "epoch": 11.571428571428571,
      "grad_norm": 1.9752053022384644,
      "learning_rate": 7.179925979216491e-05,
      "loss": 0.6355,
      "step": 1782
    },
    {
      "epoch": 11.577922077922079,
      "grad_norm": 2.114065408706665,
      "learning_rate": 7.177098260739024e-05,
      "loss": 0.6583,
      "step": 1783
    },
    {
      "epoch": 11.584415584415584,
      "grad_norm": 2.2280924320220947,
      "learning_rate": 7.174269682777641e-05,
      "loss": 0.665,
      "step": 1784
    },
    {
      "epoch": 11.590909090909092,
      "grad_norm": 2.106236696243286,
      "learning_rate": 7.171440246449024e-05,
      "loss": 0.6574,
      "step": 1785
    },
    {
      "epoch": 11.597402597402597,
      "grad_norm": 2.398568630218506,
      "learning_rate": 7.168609952870184e-05,
      "loss": 0.6788,
      "step": 1786
    },
    {
      "epoch": 11.603896103896105,
      "grad_norm": 2.0673725605010986,
      "learning_rate": 7.16577880315848e-05,
      "loss": 0.6203,
      "step": 1787
    },
    {
      "epoch": 11.61038961038961,
      "grad_norm": 2.2148146629333496,
      "learning_rate": 7.162946798431604e-05,
      "loss": 0.7175,
      "step": 1788
    },
    {
      "epoch": 11.616883116883116,
      "grad_norm": 2.2335145473480225,
      "learning_rate": 7.160113939807587e-05,
      "loss": 0.6621,
      "step": 1789
    },
    {
      "epoch": 11.623376623376624,
      "grad_norm": 2.2381131649017334,
      "learning_rate": 7.157280228404795e-05,
      "loss": 0.6576,
      "step": 1790
    },
    {
      "epoch": 11.62987012987013,
      "grad_norm": 2.1792495250701904,
      "learning_rate": 7.154445665341933e-05,
      "loss": 0.6911,
      "step": 1791
    },
    {
      "epoch": 11.636363636363637,
      "grad_norm": 2.3321566581726074,
      "learning_rate": 7.151610251738045e-05,
      "loss": 0.6945,
      "step": 1792
    },
    {
      "epoch": 11.642857142857142,
      "grad_norm": 2.0206103324890137,
      "learning_rate": 7.148773988712503e-05,
      "loss": 0.6102,
      "step": 1793
    },
    {
      "epoch": 11.64935064935065,
      "grad_norm": 2.0455446243286133,
      "learning_rate": 7.145936877385018e-05,
      "loss": 0.6816,
      "step": 1794
    },
    {
      "epoch": 11.655844155844155,
      "grad_norm": 2.3636724948883057,
      "learning_rate": 7.143098918875643e-05,
      "loss": 0.6943,
      "step": 1795
    },
    {
      "epoch": 11.662337662337663,
      "grad_norm": 2.1322073936462402,
      "learning_rate": 7.140260114304751e-05,
      "loss": 0.7023,
      "step": 1796
    },
    {
      "epoch": 11.668831168831169,
      "grad_norm": 2.060019016265869,
      "learning_rate": 7.137420464793063e-05,
      "loss": 0.5799,
      "step": 1797
    },
    {
      "epoch": 11.675324675324676,
      "grad_norm": 1.8431174755096436,
      "learning_rate": 7.134579971461627e-05,
      "loss": 0.5107,
      "step": 1798
    },
    {
      "epoch": 11.681818181818182,
      "grad_norm": 2.0622732639312744,
      "learning_rate": 7.131738635431822e-05,
      "loss": 0.6045,
      "step": 1799
    },
    {
      "epoch": 11.688311688311689,
      "grad_norm": 1.8862097263336182,
      "learning_rate": 7.128896457825364e-05,
      "loss": 0.6123,
      "step": 1800
    },
    {
      "epoch": 11.694805194805195,
      "grad_norm": 2.087017297744751,
      "learning_rate": 7.126053439764299e-05,
      "loss": 0.6562,
      "step": 1801
    },
    {
      "epoch": 11.7012987012987,
      "grad_norm": 1.8770846128463745,
      "learning_rate": 7.123209582371006e-05,
      "loss": 0.6068,
      "step": 1802
    },
    {
      "epoch": 11.707792207792208,
      "grad_norm": 2.0145421028137207,
      "learning_rate": 7.120364886768197e-05,
      "loss": 0.645,
      "step": 1803
    },
    {
      "epoch": 11.714285714285714,
      "grad_norm": 1.9685002565383911,
      "learning_rate": 7.11751935407891e-05,
      "loss": 0.6093,
      "step": 1804
    },
    {
      "epoch": 11.720779220779221,
      "grad_norm": 2.0800533294677734,
      "learning_rate": 7.114672985426516e-05,
      "loss": 0.6473,
      "step": 1805
    },
    {
      "epoch": 11.727272727272727,
      "grad_norm": 1.9733763933181763,
      "learning_rate": 7.111825781934719e-05,
      "loss": 0.637,
      "step": 1806
    },
    {
      "epoch": 11.733766233766234,
      "grad_norm": 2.161607265472412,
      "learning_rate": 7.108977744727547e-05,
      "loss": 0.6241,
      "step": 1807
    },
    {
      "epoch": 11.74025974025974,
      "grad_norm": 2.179612636566162,
      "learning_rate": 7.106128874929363e-05,
      "loss": 0.7628,
      "step": 1808
    },
    {
      "epoch": 11.746753246753247,
      "grad_norm": 2.0560779571533203,
      "learning_rate": 7.103279173664851e-05,
      "loss": 0.5887,
      "step": 1809
    },
    {
      "epoch": 11.753246753246753,
      "grad_norm": 2.240081548690796,
      "learning_rate": 7.100428642059033e-05,
      "loss": 0.733,
      "step": 1810
    },
    {
      "epoch": 11.75974025974026,
      "grad_norm": 2.139436960220337,
      "learning_rate": 7.097577281237249e-05,
      "loss": 0.6425,
      "step": 1811
    },
    {
      "epoch": 11.766233766233766,
      "grad_norm": 2.2671170234680176,
      "learning_rate": 7.094725092325176e-05,
      "loss": 0.6823,
      "step": 1812
    },
    {
      "epoch": 11.772727272727273,
      "grad_norm": 2.1800804138183594,
      "learning_rate": 7.09187207644881e-05,
      "loss": 0.7122,
      "step": 1813
    },
    {
      "epoch": 11.779220779220779,
      "grad_norm": 1.9480642080307007,
      "learning_rate": 7.089018234734476e-05,
      "loss": 0.556,
      "step": 1814
    },
    {
      "epoch": 11.785714285714286,
      "grad_norm": 1.8281680345535278,
      "learning_rate": 7.086163568308828e-05,
      "loss": 0.5396,
      "step": 1815
    },
    {
      "epoch": 11.792207792207792,
      "grad_norm": 2.3980553150177,
      "learning_rate": 7.08330807829884e-05,
      "loss": 0.7549,
      "step": 1816
    },
    {
      "epoch": 11.7987012987013,
      "grad_norm": 2.317260503768921,
      "learning_rate": 7.080451765831817e-05,
      "loss": 0.6985,
      "step": 1817
    },
    {
      "epoch": 11.805194805194805,
      "grad_norm": 2.4132368564605713,
      "learning_rate": 7.077594632035385e-05,
      "loss": 0.7135,
      "step": 1818
    },
    {
      "epoch": 11.811688311688311,
      "grad_norm": 2.272742509841919,
      "learning_rate": 7.074736678037494e-05,
      "loss": 0.668,
      "step": 1819
    },
    {
      "epoch": 11.818181818181818,
      "grad_norm": 2.309525728225708,
      "learning_rate": 7.071877904966423e-05,
      "loss": 0.7512,
      "step": 1820
    },
    {
      "epoch": 11.824675324675324,
      "grad_norm": 2.0604569911956787,
      "learning_rate": 7.069018313950763e-05,
      "loss": 0.6374,
      "step": 1821
    },
    {
      "epoch": 11.831168831168831,
      "grad_norm": 2.2944161891937256,
      "learning_rate": 7.066157906119441e-05,
      "loss": 0.7265,
      "step": 1822
    },
    {
      "epoch": 11.837662337662337,
      "grad_norm": 2.2441184520721436,
      "learning_rate": 7.063296682601702e-05,
      "loss": 0.6809,
      "step": 1823
    },
    {
      "epoch": 11.844155844155845,
      "grad_norm": 2.2128078937530518,
      "learning_rate": 7.060434644527105e-05,
      "loss": 0.6976,
      "step": 1824
    },
    {
      "epoch": 11.85064935064935,
      "grad_norm": 2.045984983444214,
      "learning_rate": 7.057571793025544e-05,
      "loss": 0.6242,
      "step": 1825
    },
    {
      "epoch": 11.857142857142858,
      "grad_norm": 2.0411674976348877,
      "learning_rate": 7.054708129227224e-05,
      "loss": 0.6031,
      "step": 1826
    },
    {
      "epoch": 11.863636363636363,
      "grad_norm": 2.0572099685668945,
      "learning_rate": 7.051843654262676e-05,
      "loss": 0.5997,
      "step": 1827
    },
    {
      "epoch": 11.87012987012987,
      "grad_norm": 2.142810821533203,
      "learning_rate": 7.048978369262747e-05,
      "loss": 0.5677,
      "step": 1828
    },
    {
      "epoch": 11.876623376623376,
      "grad_norm": 2.1098294258117676,
      "learning_rate": 7.046112275358608e-05,
      "loss": 0.6425,
      "step": 1829
    },
    {
      "epoch": 11.883116883116884,
      "grad_norm": 2.272458791732788,
      "learning_rate": 7.043245373681747e-05,
      "loss": 0.7284,
      "step": 1830
    },
    {
      "epoch": 11.88961038961039,
      "grad_norm": 2.2965171337127686,
      "learning_rate": 7.040377665363969e-05,
      "loss": 0.6809,
      "step": 1831
    },
    {
      "epoch": 11.896103896103895,
      "grad_norm": 2.2099809646606445,
      "learning_rate": 7.037509151537403e-05,
      "loss": 0.6389,
      "step": 1832
    },
    {
      "epoch": 11.902597402597403,
      "grad_norm": 2.201387882232666,
      "learning_rate": 7.034639833334494e-05,
      "loss": 0.6783,
      "step": 1833
    },
    {
      "epoch": 11.909090909090908,
      "grad_norm": 2.2194228172302246,
      "learning_rate": 7.031769711888e-05,
      "loss": 0.6866,
      "step": 1834
    },
    {
      "epoch": 11.915584415584416,
      "grad_norm": 2.3768062591552734,
      "learning_rate": 7.028898788331e-05,
      "loss": 0.7316,
      "step": 1835
    },
    {
      "epoch": 11.922077922077921,
      "grad_norm": 2.1263155937194824,
      "learning_rate": 7.026027063796891e-05,
      "loss": 0.6822,
      "step": 1836
    },
    {
      "epoch": 11.928571428571429,
      "grad_norm": 2.297542095184326,
      "learning_rate": 7.023154539419384e-05,
      "loss": 0.6611,
      "step": 1837
    },
    {
      "epoch": 11.935064935064934,
      "grad_norm": 2.1204769611358643,
      "learning_rate": 7.020281216332503e-05,
      "loss": 0.658,
      "step": 1838
    },
    {
      "epoch": 11.941558441558442,
      "grad_norm": 1.9579997062683105,
      "learning_rate": 7.017407095670593e-05,
      "loss": 0.5524,
      "step": 1839
    },
    {
      "epoch": 11.948051948051948,
      "grad_norm": 2.3692562580108643,
      "learning_rate": 7.014532178568314e-05,
      "loss": 0.6995,
      "step": 1840
    },
    {
      "epoch": 11.954545454545455,
      "grad_norm": 2.3002681732177734,
      "learning_rate": 7.011656466160632e-05,
      "loss": 0.7267,
      "step": 1841
    },
    {
      "epoch": 11.96103896103896,
      "grad_norm": 2.3371801376342773,
      "learning_rate": 7.008779959582837e-05,
      "loss": 0.7695,
      "step": 1842
    },
    {
      "epoch": 11.967532467532468,
      "grad_norm": 2.0627670288085938,
      "learning_rate": 7.005902659970528e-05,
      "loss": 0.635,
      "step": 1843
    },
    {
      "epoch": 11.974025974025974,
      "grad_norm": 2.0339272022247314,
      "learning_rate": 7.003024568459614e-05,
      "loss": 0.6353,
      "step": 1844
    },
    {
      "epoch": 11.980519480519481,
      "grad_norm": 2.1179451942443848,
      "learning_rate": 7.000145686186324e-05,
      "loss": 0.6786,
      "step": 1845
    },
    {
      "epoch": 11.987012987012987,
      "grad_norm": 2.1404683589935303,
      "learning_rate": 6.997266014287193e-05,
      "loss": 0.7124,
      "step": 1846
    },
    {
      "epoch": 11.993506493506494,
      "grad_norm": 2.2253522872924805,
      "learning_rate": 6.994385553899069e-05,
      "loss": 0.7233,
      "step": 1847
    },
    {
      "epoch": 12.0,
      "grad_norm": 1076.5802001953125,
      "learning_rate": 6.991504306159114e-05,
      "loss": 0.6579,
      "step": 1848
    },
    {
      "epoch": 12.006493506493506,
      "grad_norm": 1.9225164651870728,
      "learning_rate": 6.988622272204799e-05,
      "loss": 0.5623,
      "step": 1849
    },
    {
      "epoch": 12.012987012987013,
      "grad_norm": 1.8583208322525024,
      "learning_rate": 6.985739453173903e-05,
      "loss": 0.53,
      "step": 1850
    },
    {
      "epoch": 12.019480519480519,
      "grad_norm": 1.985059380531311,
      "learning_rate": 6.98285585020452e-05,
      "loss": 0.5346,
      "step": 1851
    },
    {
      "epoch": 12.025974025974026,
      "grad_norm": 2.0284557342529297,
      "learning_rate": 6.97997146443505e-05,
      "loss": 0.5592,
      "step": 1852
    },
    {
      "epoch": 12.032467532467532,
      "grad_norm": 2.00041127204895,
      "learning_rate": 6.977086297004202e-05,
      "loss": 0.5619,
      "step": 1853
    },
    {
      "epoch": 12.03896103896104,
      "grad_norm": 2.134187936782837,
      "learning_rate": 6.974200349050996e-05,
      "loss": 0.6185,
      "step": 1854
    },
    {
      "epoch": 12.045454545454545,
      "grad_norm": 1.9643243551254272,
      "learning_rate": 6.971313621714756e-05,
      "loss": 0.5066,
      "step": 1855
    },
    {
      "epoch": 12.051948051948052,
      "grad_norm": 2.0005311965942383,
      "learning_rate": 6.968426116135118e-05,
      "loss": 0.591,
      "step": 1856
    },
    {
      "epoch": 12.058441558441558,
      "grad_norm": 1.8810878992080688,
      "learning_rate": 6.965537833452024e-05,
      "loss": 0.4961,
      "step": 1857
    },
    {
      "epoch": 12.064935064935066,
      "grad_norm": 1.9006643295288086,
      "learning_rate": 6.96264877480572e-05,
      "loss": 0.5309,
      "step": 1858
    },
    {
      "epoch": 12.071428571428571,
      "grad_norm": 2.020428419113159,
      "learning_rate": 6.959758941336762e-05,
      "loss": 0.5334,
      "step": 1859
    },
    {
      "epoch": 12.077922077922079,
      "grad_norm": 1.9589020013809204,
      "learning_rate": 6.956868334186013e-05,
      "loss": 0.5894,
      "step": 1860
    },
    {
      "epoch": 12.084415584415584,
      "grad_norm": 1.9384912252426147,
      "learning_rate": 6.953976954494635e-05,
      "loss": 0.5343,
      "step": 1861
    },
    {
      "epoch": 12.090909090909092,
      "grad_norm": 2.0977766513824463,
      "learning_rate": 6.9510848034041e-05,
      "loss": 0.5726,
      "step": 1862
    },
    {
      "epoch": 12.097402597402597,
      "grad_norm": 2.350893497467041,
      "learning_rate": 6.948191882056185e-05,
      "loss": 0.5778,
      "step": 1863
    },
    {
      "epoch": 12.103896103896103,
      "grad_norm": 1.8108704090118408,
      "learning_rate": 6.945298191592967e-05,
      "loss": 0.4941,
      "step": 1864
    },
    {
      "epoch": 12.11038961038961,
      "grad_norm": 1.985229253768921,
      "learning_rate": 6.942403733156832e-05,
      "loss": 0.5245,
      "step": 1865
    },
    {
      "epoch": 12.116883116883116,
      "grad_norm": 2.0757787227630615,
      "learning_rate": 6.939508507890464e-05,
      "loss": 0.5347,
      "step": 1866
    },
    {
      "epoch": 12.123376623376624,
      "grad_norm": 2.1860249042510986,
      "learning_rate": 6.936612516936852e-05,
      "loss": 0.5951,
      "step": 1867
    },
    {
      "epoch": 12.12987012987013,
      "grad_norm": 2.0259830951690674,
      "learning_rate": 6.93371576143929e-05,
      "loss": 0.5316,
      "step": 1868
    },
    {
      "epoch": 12.136363636363637,
      "grad_norm": 1.9689854383468628,
      "learning_rate": 6.930818242541368e-05,
      "loss": 0.5568,
      "step": 1869
    },
    {
      "epoch": 12.142857142857142,
      "grad_norm": 2.054041862487793,
      "learning_rate": 6.927919961386984e-05,
      "loss": 0.5711,
      "step": 1870
    },
    {
      "epoch": 12.14935064935065,
      "grad_norm": 1.7612172365188599,
      "learning_rate": 6.92502091912033e-05,
      "loss": 0.4729,
      "step": 1871
    },
    {
      "epoch": 12.155844155844155,
      "grad_norm": 2.099085569381714,
      "learning_rate": 6.922121116885903e-05,
      "loss": 0.5802,
      "step": 1872
    },
    {
      "epoch": 12.162337662337663,
      "grad_norm": 2.230409860610962,
      "learning_rate": 6.919220555828502e-05,
      "loss": 0.5908,
      "step": 1873
    },
    {
      "epoch": 12.168831168831169,
      "grad_norm": 2.0448951721191406,
      "learning_rate": 6.916319237093219e-05,
      "loss": 0.5871,
      "step": 1874
    },
    {
      "epoch": 12.175324675324676,
      "grad_norm": 1.6726332902908325,
      "learning_rate": 6.91341716182545e-05,
      "loss": 0.4441,
      "step": 1875
    },
    {
      "epoch": 12.181818181818182,
      "grad_norm": 1.9495742321014404,
      "learning_rate": 6.910514331170888e-05,
      "loss": 0.5741,
      "step": 1876
    },
    {
      "epoch": 12.188311688311689,
      "grad_norm": 1.893748164176941,
      "learning_rate": 6.907610746275523e-05,
      "loss": 0.5568,
      "step": 1877
    },
    {
      "epoch": 12.194805194805195,
      "grad_norm": 2.1908624172210693,
      "learning_rate": 6.904706408285648e-05,
      "loss": 0.6201,
      "step": 1878
    },
    {
      "epoch": 12.2012987012987,
      "grad_norm": 2.14664363861084,
      "learning_rate": 6.901801318347847e-05,
      "loss": 0.6187,
      "step": 1879
    },
    {
      "epoch": 12.207792207792208,
      "grad_norm": 2.0984911918640137,
      "learning_rate": 6.898895477609007e-05,
      "loss": 0.608,
      "step": 1880
    },
    {
      "epoch": 12.214285714285714,
      "grad_norm": 2.1119325160980225,
      "learning_rate": 6.895988887216302e-05,
      "loss": 0.6473,
      "step": 1881
    },
    {
      "epoch": 12.220779220779221,
      "grad_norm": 2.1093862056732178,
      "learning_rate": 6.893081548317211e-05,
      "loss": 0.6126,
      "step": 1882
    },
    {
      "epoch": 12.227272727272727,
      "grad_norm": 2.203214645385742,
      "learning_rate": 6.890173462059506e-05,
      "loss": 0.6142,
      "step": 1883
    },
    {
      "epoch": 12.233766233766234,
      "grad_norm": 2.082893133163452,
      "learning_rate": 6.887264629591254e-05,
      "loss": 0.584,
      "step": 1884
    },
    {
      "epoch": 12.24025974025974,
      "grad_norm": 1.9005697965621948,
      "learning_rate": 6.884355052060814e-05,
      "loss": 0.494,
      "step": 1885
    },
    {
      "epoch": 12.246753246753247,
      "grad_norm": 2.1379759311676025,
      "learning_rate": 6.881444730616842e-05,
      "loss": 0.5557,
      "step": 1886
    },
    {
      "epoch": 12.253246753246753,
      "grad_norm": 1.7855029106140137,
      "learning_rate": 6.878533666408286e-05,
      "loss": 0.4827,
      "step": 1887
    },
    {
      "epoch": 12.25974025974026,
      "grad_norm": 2.110853433609009,
      "learning_rate": 6.87562186058439e-05,
      "loss": 0.5239,
      "step": 1888
    },
    {
      "epoch": 12.266233766233766,
      "grad_norm": 2.10565185546875,
      "learning_rate": 6.872709314294685e-05,
      "loss": 0.5986,
      "step": 1889
    },
    {
      "epoch": 12.272727272727273,
      "grad_norm": 2.247499465942383,
      "learning_rate": 6.869796028689001e-05,
      "loss": 0.5874,
      "step": 1890
    },
    {
      "epoch": 12.279220779220779,
      "grad_norm": 2.151932954788208,
      "learning_rate": 6.86688200491746e-05,
      "loss": 0.5771,
      "step": 1891
    },
    {
      "epoch": 12.285714285714286,
      "grad_norm": 1.9797194004058838,
      "learning_rate": 6.863967244130467e-05,
      "loss": 0.5452,
      "step": 1892
    },
    {
      "epoch": 12.292207792207792,
      "grad_norm": 2.0008790493011475,
      "learning_rate": 6.861051747478726e-05,
      "loss": 0.5515,
      "step": 1893
    },
    {
      "epoch": 12.2987012987013,
      "grad_norm": 2.157362937927246,
      "learning_rate": 6.858135516113226e-05,
      "loss": 0.5479,
      "step": 1894
    },
    {
      "epoch": 12.305194805194805,
      "grad_norm": 2.224337577819824,
      "learning_rate": 6.855218551185255e-05,
      "loss": 0.6381,
      "step": 1895
    },
    {
      "epoch": 12.311688311688311,
      "grad_norm": 1.938456416130066,
      "learning_rate": 6.852300853846381e-05,
      "loss": 0.5629,
      "step": 1896
    },
    {
      "epoch": 12.318181818181818,
      "grad_norm": 1.7481359243392944,
      "learning_rate": 6.849382425248464e-05,
      "loss": 0.4578,
      "step": 1897
    },
    {
      "epoch": 12.324675324675324,
      "grad_norm": 1.9388666152954102,
      "learning_rate": 6.84646326654365e-05,
      "loss": 0.5472,
      "step": 1898
    },
    {
      "epoch": 12.331168831168831,
      "grad_norm": 2.1877782344818115,
      "learning_rate": 6.843543378884387e-05,
      "loss": 0.6215,
      "step": 1899
    },
    {
      "epoch": 12.337662337662337,
      "grad_norm": 2.2899937629699707,
      "learning_rate": 6.840622763423391e-05,
      "loss": 0.6257,
      "step": 1900
    },
    {
      "epoch": 12.344155844155845,
      "grad_norm": 2.166365146636963,
      "learning_rate": 6.837701421313677e-05,
      "loss": 0.5647,
      "step": 1901
    },
    {
      "epoch": 12.35064935064935,
      "grad_norm": 2.299274444580078,
      "learning_rate": 6.834779353708548e-05,
      "loss": 0.635,
      "step": 1902
    },
    {
      "epoch": 12.357142857142858,
      "grad_norm": 2.1747353076934814,
      "learning_rate": 6.831856561761585e-05,
      "loss": 0.6053,
      "step": 1903
    },
    {
      "epoch": 12.363636363636363,
      "grad_norm": 2.0763461589813232,
      "learning_rate": 6.828933046626664e-05,
      "loss": 0.5943,
      "step": 1904
    },
    {
      "epoch": 12.37012987012987,
      "grad_norm": 2.133894920349121,
      "learning_rate": 6.82600880945794e-05,
      "loss": 0.5598,
      "step": 1905
    },
    {
      "epoch": 12.376623376623376,
      "grad_norm": 2.123075485229492,
      "learning_rate": 6.823083851409857e-05,
      "loss": 0.5706,
      "step": 1906
    },
    {
      "epoch": 12.383116883116884,
      "grad_norm": 2.181898832321167,
      "learning_rate": 6.820158173637142e-05,
      "loss": 0.5897,
      "step": 1907
    },
    {
      "epoch": 12.38961038961039,
      "grad_norm": 2.125399589538574,
      "learning_rate": 6.817231777294803e-05,
      "loss": 0.5686,
      "step": 1908
    },
    {
      "epoch": 12.396103896103897,
      "grad_norm": 2.233853578567505,
      "learning_rate": 6.814304663538142e-05,
      "loss": 0.6248,
      "step": 1909
    },
    {
      "epoch": 12.402597402597403,
      "grad_norm": 2.1631298065185547,
      "learning_rate": 6.811376833522729e-05,
      "loss": 0.591,
      "step": 1910
    },
    {
      "epoch": 12.409090909090908,
      "grad_norm": 2.2579503059387207,
      "learning_rate": 6.808448288404431e-05,
      "loss": 0.6415,
      "step": 1911
    },
    {
      "epoch": 12.415584415584416,
      "grad_norm": 2.1994619369506836,
      "learning_rate": 6.805519029339387e-05,
      "loss": 0.5854,
      "step": 1912
    },
    {
      "epoch": 12.422077922077921,
      "grad_norm": 1.9904730319976807,
      "learning_rate": 6.802589057484027e-05,
      "loss": 0.5822,
      "step": 1913
    },
    {
      "epoch": 12.428571428571429,
      "grad_norm": 2.0386736392974854,
      "learning_rate": 6.799658373995053e-05,
      "loss": 0.5614,
      "step": 1914
    },
    {
      "epoch": 12.435064935064934,
      "grad_norm": 2.15208101272583,
      "learning_rate": 6.796726980029454e-05,
      "loss": 0.5926,
      "step": 1915
    },
    {
      "epoch": 12.441558441558442,
      "grad_norm": 1.9997342824935913,
      "learning_rate": 6.793794876744499e-05,
      "loss": 0.5069,
      "step": 1916
    },
    {
      "epoch": 12.448051948051948,
      "grad_norm": 2.0709683895111084,
      "learning_rate": 6.790862065297732e-05,
      "loss": 0.5953,
      "step": 1917
    },
    {
      "epoch": 12.454545454545455,
      "grad_norm": 1.7772856950759888,
      "learning_rate": 6.787928546846987e-05,
      "loss": 0.5253,
      "step": 1918
    },
    {
      "epoch": 12.46103896103896,
      "grad_norm": 2.1581130027770996,
      "learning_rate": 6.784994322550365e-05,
      "loss": 0.5836,
      "step": 1919
    },
    {
      "epoch": 12.467532467532468,
      "grad_norm": 1.9727606773376465,
      "learning_rate": 6.782059393566253e-05,
      "loss": 0.5558,
      "step": 1920
    },
    {
      "epoch": 12.474025974025974,
      "grad_norm": 2.0561611652374268,
      "learning_rate": 6.779123761053317e-05,
      "loss": 0.5259,
      "step": 1921
    },
    {
      "epoch": 12.480519480519481,
      "grad_norm": 2.3199853897094727,
      "learning_rate": 6.776187426170493e-05,
      "loss": 0.6292,
      "step": 1922
    },
    {
      "epoch": 12.487012987012987,
      "grad_norm": 2.1790263652801514,
      "learning_rate": 6.773250390077006e-05,
      "loss": 0.5693,
      "step": 1923
    },
    {
      "epoch": 12.493506493506494,
      "grad_norm": 2.2785651683807373,
      "learning_rate": 6.770312653932345e-05,
      "loss": 0.6542,
      "step": 1924
    },
    {
      "epoch": 12.5,
      "grad_norm": 1.8668270111083984,
      "learning_rate": 6.767374218896286e-05,
      "loss": 0.5379,
      "step": 1925
    },
    {
      "epoch": 12.506493506493506,
      "grad_norm": 2.2081425189971924,
      "learning_rate": 6.764435086128876e-05,
      "loss": 0.5977,
      "step": 1926
    },
    {
      "epoch": 12.512987012987013,
      "grad_norm": 2.148101329803467,
      "learning_rate": 6.761495256790435e-05,
      "loss": 0.5996,
      "step": 1927
    },
    {
      "epoch": 12.519480519480519,
      "grad_norm": 1.9085726737976074,
      "learning_rate": 6.758554732041564e-05,
      "loss": 0.5474,
      "step": 1928
    },
    {
      "epoch": 12.525974025974026,
      "grad_norm": 1.6701866388320923,
      "learning_rate": 6.755613513043137e-05,
      "loss": 0.4075,
      "step": 1929
    },
    {
      "epoch": 12.532467532467532,
      "grad_norm": 2.0828919410705566,
      "learning_rate": 6.752671600956295e-05,
      "loss": 0.5761,
      "step": 1930
    },
    {
      "epoch": 12.53896103896104,
      "grad_norm": 2.122612237930298,
      "learning_rate": 6.749728996942463e-05,
      "loss": 0.5902,
      "step": 1931
    },
    {
      "epoch": 12.545454545454545,
      "grad_norm": 2.474417209625244,
      "learning_rate": 6.746785702163336e-05,
      "loss": 0.6628,
      "step": 1932
    },
    {
      "epoch": 12.551948051948052,
      "grad_norm": 2.0963616371154785,
      "learning_rate": 6.743841717780874e-05,
      "loss": 0.6209,
      "step": 1933
    },
    {
      "epoch": 12.558441558441558,
      "grad_norm": 1.8984389305114746,
      "learning_rate": 6.740897044957322e-05,
      "loss": 0.5674,
      "step": 1934
    },
    {
      "epoch": 12.564935064935066,
      "grad_norm": 2.200878620147705,
      "learning_rate": 6.737951684855185e-05,
      "loss": 0.6193,
      "step": 1935
    },
    {
      "epoch": 12.571428571428571,
      "grad_norm": 1.947191834449768,
      "learning_rate": 6.735005638637248e-05,
      "loss": 0.5233,
      "step": 1936
    },
    {
      "epoch": 12.577922077922079,
      "grad_norm": 1.9447444677352905,
      "learning_rate": 6.73205890746656e-05,
      "loss": 0.5566,
      "step": 1937
    },
    {
      "epoch": 12.584415584415584,
      "grad_norm": 2.252058267593384,
      "learning_rate": 6.729111492506449e-05,
      "loss": 0.648,
      "step": 1938
    },
    {
      "epoch": 12.590909090909092,
      "grad_norm": 1.99386465549469,
      "learning_rate": 6.726163394920503e-05,
      "loss": 0.5509,
      "step": 1939
    },
    {
      "epoch": 12.597402597402597,
      "grad_norm": 2.0936877727508545,
      "learning_rate": 6.723214615872585e-05,
      "loss": 0.572,
      "step": 1940
    },
    {
      "epoch": 12.603896103896105,
      "grad_norm": 2.110973596572876,
      "learning_rate": 6.720265156526828e-05,
      "loss": 0.5817,
      "step": 1941
    },
    {
      "epoch": 12.61038961038961,
      "grad_norm": 2.3618104457855225,
      "learning_rate": 6.71731501804763e-05,
      "loss": 0.6308,
      "step": 1942
    },
    {
      "epoch": 12.616883116883116,
      "grad_norm": 2.1770923137664795,
      "learning_rate": 6.714364201599662e-05,
      "loss": 0.5744,
      "step": 1943
    },
    {
      "epoch": 12.623376623376624,
      "grad_norm": 2.2065415382385254,
      "learning_rate": 6.711412708347856e-05,
      "loss": 0.6431,
      "step": 1944
    },
    {
      "epoch": 12.62987012987013,
      "grad_norm": 2.123046398162842,
      "learning_rate": 6.708460539457418e-05,
      "loss": 0.577,
      "step": 1945
    },
    {
      "epoch": 12.636363636363637,
      "grad_norm": 2.3211543560028076,
      "learning_rate": 6.705507696093814e-05,
      "loss": 0.6264,
      "step": 1946
    },
    {
      "epoch": 12.642857142857142,
      "grad_norm": 1.948301076889038,
      "learning_rate": 6.702554179422783e-05,
      "loss": 0.5625,
      "step": 1947
    },
    {
      "epoch": 12.64935064935065,
      "grad_norm": 1.979990839958191,
      "learning_rate": 6.699599990610323e-05,
      "loss": 0.5623,
      "step": 1948
    },
    {
      "epoch": 12.655844155844155,
      "grad_norm": 1.8006978034973145,
      "learning_rate": 6.696645130822705e-05,
      "loss": 0.5595,
      "step": 1949
    },
    {
      "epoch": 12.662337662337663,
      "grad_norm": 2.1235246658325195,
      "learning_rate": 6.693689601226458e-05,
      "loss": 0.5588,
      "step": 1950
    },
    {
      "epoch": 12.668831168831169,
      "grad_norm": 2.1899781227111816,
      "learning_rate": 6.690733402988379e-05,
      "loss": 0.615,
      "step": 1951
    },
    {
      "epoch": 12.675324675324676,
      "grad_norm": 1.9493566751480103,
      "learning_rate": 6.68777653727553e-05,
      "loss": 0.5755,
      "step": 1952
    },
    {
      "epoch": 12.681818181818182,
      "grad_norm": 2.196223497390747,
      "learning_rate": 6.684819005255232e-05,
      "loss": 0.6394,
      "step": 1953
    },
    {
      "epoch": 12.688311688311689,
      "grad_norm": 2.0495898723602295,
      "learning_rate": 6.681860808095074e-05,
      "loss": 0.5432,
      "step": 1954
    },
    {
      "epoch": 12.694805194805195,
      "grad_norm": 2.1735448837280273,
      "learning_rate": 6.678901946962903e-05,
      "loss": 0.6549,
      "step": 1955
    },
    {
      "epoch": 12.7012987012987,
      "grad_norm": 2.1591930389404297,
      "learning_rate": 6.675942423026833e-05,
      "loss": 0.5885,
      "step": 1956
    },
    {
      "epoch": 12.707792207792208,
      "grad_norm": 2.046126127243042,
      "learning_rate": 6.672982237455237e-05,
      "loss": 0.5849,
      "step": 1957
    },
    {
      "epoch": 12.714285714285714,
      "grad_norm": 2.0392959117889404,
      "learning_rate": 6.670021391416749e-05,
      "loss": 0.5539,
      "step": 1958
    },
    {
      "epoch": 12.720779220779221,
      "grad_norm": 2.007087230682373,
      "learning_rate": 6.667059886080262e-05,
      "loss": 0.5605,
      "step": 1959
    },
    {
      "epoch": 12.727272727272727,
      "grad_norm": 2.019771099090576,
      "learning_rate": 6.664097722614934e-05,
      "loss": 0.612,
      "step": 1960
    },
    {
      "epoch": 12.733766233766234,
      "grad_norm": 2.1823174953460693,
      "learning_rate": 6.66113490219018e-05,
      "loss": 0.6417,
      "step": 1961
    },
    {
      "epoch": 12.74025974025974,
      "grad_norm": 2.022130250930786,
      "learning_rate": 6.658171425975672e-05,
      "loss": 0.5359,
      "step": 1962
    },
    {
      "epoch": 12.746753246753247,
      "grad_norm": 2.0918095111846924,
      "learning_rate": 6.655207295141346e-05,
      "loss": 0.5774,
      "step": 1963
    },
    {
      "epoch": 12.753246753246753,
      "grad_norm": 2.2147998809814453,
      "learning_rate": 6.652242510857394e-05,
      "loss": 0.6046,
      "step": 1964
    },
    {
      "epoch": 12.75974025974026,
      "grad_norm": 2.1559667587280273,
      "learning_rate": 6.649277074294264e-05,
      "loss": 0.5821,
      "step": 1965
    },
    {
      "epoch": 12.766233766233766,
      "grad_norm": 1.8574299812316895,
      "learning_rate": 6.646310986622668e-05,
      "loss": 0.5442,
      "step": 1966
    },
    {
      "epoch": 12.772727272727273,
      "grad_norm": 2.072185754776001,
      "learning_rate": 6.643344249013562e-05,
      "loss": 0.5747,
      "step": 1967
    },
    {
      "epoch": 12.779220779220779,
      "grad_norm": 1.8414840698242188,
      "learning_rate": 6.640376862638176e-05,
      "loss": 0.4852,
      "step": 1968
    },
    {
      "epoch": 12.785714285714286,
      "grad_norm": 2.0400550365448,
      "learning_rate": 6.637408828667982e-05,
      "loss": 0.5534,
      "step": 1969
    },
    {
      "epoch": 12.792207792207792,
      "grad_norm": 2.1789374351501465,
      "learning_rate": 6.634440148274713e-05,
      "loss": 0.5932,
      "step": 1970
    },
    {
      "epoch": 12.7987012987013,
      "grad_norm": 2.1213948726654053,
      "learning_rate": 6.631470822630359e-05,
      "loss": 0.6047,
      "step": 1971
    },
    {
      "epoch": 12.805194805194805,
      "grad_norm": 1.9664524793624878,
      "learning_rate": 6.628500852907161e-05,
      "loss": 0.5529,
      "step": 1972
    },
    {
      "epoch": 12.811688311688311,
      "grad_norm": 2.3267581462860107,
      "learning_rate": 6.625530240277618e-05,
      "loss": 0.6179,
      "step": 1973
    },
    {
      "epoch": 12.818181818181818,
      "grad_norm": 2.1917669773101807,
      "learning_rate": 6.622558985914477e-05,
      "loss": 0.6275,
      "step": 1974
    },
    {
      "epoch": 12.824675324675324,
      "grad_norm": 2.1599512100219727,
      "learning_rate": 6.619587090990746e-05,
      "loss": 0.6153,
      "step": 1975
    },
    {
      "epoch": 12.831168831168831,
      "grad_norm": 2.2975082397460938,
      "learning_rate": 6.616614556679683e-05,
      "loss": 0.6387,
      "step": 1976
    },
    {
      "epoch": 12.837662337662337,
      "grad_norm": 2.11966609954834,
      "learning_rate": 6.613641384154794e-05,
      "loss": 0.5674,
      "step": 1977
    },
    {
      "epoch": 12.844155844155845,
      "grad_norm": 2.677170991897583,
      "learning_rate": 6.61066757458984e-05,
      "loss": 0.701,
      "step": 1978
    },
    {
      "epoch": 12.85064935064935,
      "grad_norm": 2.129204273223877,
      "learning_rate": 6.607693129158837e-05,
      "loss": 0.5654,
      "step": 1979
    },
    {
      "epoch": 12.857142857142858,
      "grad_norm": 1.9715017080307007,
      "learning_rate": 6.604718049036048e-05,
      "loss": 0.5539,
      "step": 1980
    },
    {
      "epoch": 12.863636363636363,
      "grad_norm": 2.3241770267486572,
      "learning_rate": 6.601742335395987e-05,
      "loss": 0.6669,
      "step": 1981
    },
    {
      "epoch": 12.87012987012987,
      "grad_norm": 2.105726718902588,
      "learning_rate": 6.598765989413419e-05,
      "loss": 0.6325,
      "step": 1982
    },
    {
      "epoch": 12.876623376623376,
      "grad_norm": 1.9479717016220093,
      "learning_rate": 6.595789012263355e-05,
      "loss": 0.5448,
      "step": 1983
    },
    {
      "epoch": 12.883116883116884,
      "grad_norm": 2.1983015537261963,
      "learning_rate": 6.592811405121064e-05,
      "loss": 0.6193,
      "step": 1984
    },
    {
      "epoch": 12.88961038961039,
      "grad_norm": 2.3505144119262695,
      "learning_rate": 6.589833169162054e-05,
      "loss": 0.677,
      "step": 1985
    },
    {
      "epoch": 12.896103896103895,
      "grad_norm": 2.245614767074585,
      "learning_rate": 6.586854305562088e-05,
      "loss": 0.6554,
      "step": 1986
    },
    {
      "epoch": 12.902597402597403,
      "grad_norm": 1.899040937423706,
      "learning_rate": 6.583874815497175e-05,
      "loss": 0.5566,
      "step": 1987
    },
    {
      "epoch": 12.909090909090908,
      "grad_norm": 2.100031614303589,
      "learning_rate": 6.580894700143565e-05,
      "loss": 0.5871,
      "step": 1988
    },
    {
      "epoch": 12.915584415584416,
      "grad_norm": 2.2208034992218018,
      "learning_rate": 6.577913960677766e-05,
      "loss": 0.6821,
      "step": 1989
    },
    {
      "epoch": 12.922077922077921,
      "grad_norm": 2.1982948780059814,
      "learning_rate": 6.574932598276525e-05,
      "loss": 0.6474,
      "step": 1990
    },
    {
      "epoch": 12.928571428571429,
      "grad_norm": 1.8316792249679565,
      "learning_rate": 6.571950614116835e-05,
      "loss": 0.5621,
      "step": 1991
    },
    {
      "epoch": 12.935064935064934,
      "grad_norm": 2.208531379699707,
      "learning_rate": 6.568968009375937e-05,
      "loss": 0.6198,
      "step": 1992
    },
    {
      "epoch": 12.941558441558442,
      "grad_norm": 2.179624557495117,
      "learning_rate": 6.565984785231318e-05,
      "loss": 0.6272,
      "step": 1993
    },
    {
      "epoch": 12.948051948051948,
      "grad_norm": 2.1235885620117188,
      "learning_rate": 6.563000942860706e-05,
      "loss": 0.6084,
      "step": 1994
    },
    {
      "epoch": 12.954545454545455,
      "grad_norm": 2.1948156356811523,
      "learning_rate": 6.560016483442075e-05,
      "loss": 0.6801,
      "step": 1995
    },
    {
      "epoch": 12.96103896103896,
      "grad_norm": 1.9589859247207642,
      "learning_rate": 6.557031408153642e-05,
      "loss": 0.5824,
      "step": 1996
    },
    {
      "epoch": 12.967532467532468,
      "grad_norm": 2.1491918563842773,
      "learning_rate": 6.554045718173867e-05,
      "loss": 0.6364,
      "step": 1997
    },
    {
      "epoch": 12.974025974025974,
      "grad_norm": 2.1677355766296387,
      "learning_rate": 6.551059414681455e-05,
      "loss": 0.6551,
      "step": 1998
    },
    {
      "epoch": 12.980519480519481,
      "grad_norm": 1.8698400259017944,
      "learning_rate": 6.54807249885535e-05,
      "loss": 0.5743,
      "step": 1999
    },
    {
      "epoch": 12.987012987012987,
      "grad_norm": 1.9512232542037964,
      "learning_rate": 6.545084971874738e-05,
      "loss": 0.5545,
      "step": 2000
    }
  ],
  "logging_steps": 1,
  "max_steps": 5000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 33,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1551733255296000.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}