{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9993607889690439,
  "eval_steps": 500,
  "global_step": 684,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.001461053785042462,
      "grad_norm": 0.30625924468040466,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.1669,
      "step": 1
    },
    {
      "epoch": 0.002922107570084924,
      "grad_norm": 0.29808640480041504,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.0395,
      "step": 2
    },
    {
      "epoch": 0.004383161355127386,
      "grad_norm": 0.24509362876415253,
      "learning_rate": 6.000000000000001e-07,
      "loss": 1.8973,
      "step": 3
    },
    {
      "epoch": 0.005844215140169848,
      "grad_norm": 0.27399635314941406,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.0403,
      "step": 4
    },
    {
      "epoch": 0.00730526892521231,
      "grad_norm": 0.30136558413505554,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 2.0947,
      "step": 5
    },
    {
      "epoch": 0.008766322710254772,
      "grad_norm": 0.30337047576904297,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.0886,
      "step": 6
    },
    {
      "epoch": 0.010227376495297234,
      "grad_norm": 0.28456181287765503,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 1.988,
      "step": 7
    },
    {
      "epoch": 0.011688430280339696,
      "grad_norm": 0.2553723454475403,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 2.0434,
      "step": 8
    },
    {
      "epoch": 0.013149484065382157,
      "grad_norm": 0.2616969645023346,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.9692,
      "step": 9
    },
    {
      "epoch": 0.01461053785042462,
      "grad_norm": 0.2795903980731964,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.9994,
      "step": 10
    },
    {
      "epoch": 0.01607159163546708,
      "grad_norm": 0.27633437514305115,
      "learning_rate": 2.2e-06,
      "loss": 1.9672,
      "step": 11
    },
    {
      "epoch": 0.017532645420509543,
      "grad_norm": 0.3074289858341217,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 2.0757,
      "step": 12
    },
    {
      "epoch": 0.018993699205552005,
      "grad_norm": 0.27444779872894287,
      "learning_rate": 2.6e-06,
      "loss": 1.9697,
      "step": 13
    },
    {
      "epoch": 0.020454752990594467,
      "grad_norm": 0.28576454520225525,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 2.0516,
      "step": 14
    },
    {
      "epoch": 0.02191580677563693,
      "grad_norm": 0.25402647256851196,
      "learning_rate": 3e-06,
      "loss": 1.9916,
      "step": 15
    },
    {
      "epoch": 0.02337686056067939,
      "grad_norm": 0.27070897817611694,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 2.0079,
      "step": 16
    },
    {
      "epoch": 0.024837914345721853,
      "grad_norm": 0.26685598492622375,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.997,
      "step": 17
    },
    {
      "epoch": 0.026298968130764315,
      "grad_norm": 0.2818976044654846,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 2.0454,
      "step": 18
    },
    {
      "epoch": 0.027760021915806777,
      "grad_norm": 0.2832131087779999,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 2.0227,
      "step": 19
    },
    {
      "epoch": 0.02922107570084924,
      "grad_norm": 0.28652554750442505,
      "learning_rate": 4.000000000000001e-06,
      "loss": 2.0293,
      "step": 20
    },
    {
      "epoch": 0.0306821294858917,
      "grad_norm": 0.28600171208381653,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 2.0214,
      "step": 21
    },
    {
      "epoch": 0.03214318327093416,
      "grad_norm": 0.2540723383426666,
      "learning_rate": 4.4e-06,
      "loss": 1.998,
      "step": 22
    },
    {
      "epoch": 0.03360423705597662,
      "grad_norm": 0.2570485472679138,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.9908,
      "step": 23
    },
    {
      "epoch": 0.03506529084101909,
      "grad_norm": 0.2579878866672516,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.9355,
      "step": 24
    },
    {
      "epoch": 0.036526344626061545,
      "grad_norm": 0.29550811648368835,
      "learning_rate": 5e-06,
      "loss": 2.0992,
      "step": 25
    },
    {
      "epoch": 0.03798739841110401,
      "grad_norm": 0.2850530445575714,
      "learning_rate": 5.2e-06,
      "loss": 2.042,
      "step": 26
    },
    {
      "epoch": 0.03944845219614647,
      "grad_norm": 0.2779623568058014,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.9592,
      "step": 27
    },
    {
      "epoch": 0.040909505981188934,
      "grad_norm": 0.2843893766403198,
      "learning_rate": 5.600000000000001e-06,
      "loss": 2.0435,
      "step": 28
    },
    {
      "epoch": 0.04237055976623139,
      "grad_norm": 0.26713600754737854,
      "learning_rate": 5.8e-06,
      "loss": 2.0014,
      "step": 29
    },
    {
      "epoch": 0.04383161355127386,
      "grad_norm": 0.2971515357494354,
      "learning_rate": 6e-06,
      "loss": 2.0246,
      "step": 30
    },
    {
      "epoch": 0.04529266733631632,
      "grad_norm": 0.2626894414424896,
      "learning_rate": 6.200000000000001e-06,
      "loss": 2.0402,
      "step": 31
    },
    {
      "epoch": 0.04675372112135878,
      "grad_norm": 0.27606457471847534,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 2.0549,
      "step": 32
    },
    {
      "epoch": 0.04821477490640124,
      "grad_norm": 0.27802756428718567,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.9889,
      "step": 33
    },
    {
      "epoch": 0.049675828691443706,
      "grad_norm": 0.2772068977355957,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.9913,
      "step": 34
    },
    {
      "epoch": 0.051136882476486165,
      "grad_norm": 0.2816425561904907,
      "learning_rate": 7e-06,
      "loss": 1.9677,
      "step": 35
    },
    {
      "epoch": 0.05259793626152863,
      "grad_norm": 0.2521456182003021,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.9705,
      "step": 36
    },
    {
      "epoch": 0.05405899004657109,
      "grad_norm": 0.24795399606227875,
      "learning_rate": 7.4e-06,
      "loss": 2.0483,
      "step": 37
    },
    {
      "epoch": 0.055520043831613554,
      "grad_norm": 0.2924613654613495,
      "learning_rate": 7.600000000000001e-06,
      "loss": 2.0585,
      "step": 38
    },
    {
      "epoch": 0.05698109761665601,
      "grad_norm": 0.2479739636182785,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.9588,
      "step": 39
    },
    {
      "epoch": 0.05844215140169848,
      "grad_norm": 0.27038994431495667,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.9146,
      "step": 40
    },
    {
      "epoch": 0.059903205186740936,
      "grad_norm": 0.263032466173172,
      "learning_rate": 8.2e-06,
      "loss": 1.925,
      "step": 41
    },
    {
      "epoch": 0.0613642589717834,
      "grad_norm": 0.25826576352119446,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.9664,
      "step": 42
    },
    {
      "epoch": 0.06282531275682586,
      "grad_norm": 0.3021976351737976,
      "learning_rate": 8.6e-06,
      "loss": 2.0211,
      "step": 43
    },
    {
      "epoch": 0.06428636654186833,
      "grad_norm": 0.2642092704772949,
      "learning_rate": 8.8e-06,
      "loss": 1.8985,
      "step": 44
    },
    {
      "epoch": 0.06574742032691079,
      "grad_norm": 0.2693665623664856,
      "learning_rate": 9e-06,
      "loss": 2.0221,
      "step": 45
    },
    {
      "epoch": 0.06720847411195324,
      "grad_norm": 0.24328897893428802,
      "learning_rate": 9.200000000000002e-06,
      "loss": 1.9289,
      "step": 46
    },
    {
      "epoch": 0.06866952789699571,
      "grad_norm": 0.24804741144180298,
      "learning_rate": 9.4e-06,
      "loss": 1.9128,
      "step": 47
    },
    {
      "epoch": 0.07013058168203817,
      "grad_norm": 0.2629912197589874,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.8998,
      "step": 48
    },
    {
      "epoch": 0.07159163546708064,
      "grad_norm": 0.22671160101890564,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.8634,
      "step": 49
    },
    {
      "epoch": 0.07305268925212309,
      "grad_norm": 0.22378858923912048,
      "learning_rate": 1e-05,
      "loss": 1.9038,
      "step": 50
    },
    {
      "epoch": 0.07451374303716556,
      "grad_norm": 0.25769534707069397,
      "learning_rate": 1.02e-05,
      "loss": 1.8324,
      "step": 51
    },
    {
      "epoch": 0.07597479682220802,
      "grad_norm": 0.22694693505764008,
      "learning_rate": 1.04e-05,
      "loss": 1.8374,
      "step": 52
    },
    {
      "epoch": 0.07743585060725047,
      "grad_norm": 0.23865583539009094,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 1.9129,
      "step": 53
    },
    {
      "epoch": 0.07889690439229294,
      "grad_norm": 0.23314256966114044,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 1.8512,
      "step": 54
    },
    {
      "epoch": 0.0803579581773354,
      "grad_norm": 0.2111833393573761,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.8608,
      "step": 55
    },
    {
      "epoch": 0.08181901196237787,
      "grad_norm": 0.24742205440998077,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 1.9129,
      "step": 56
    },
    {
      "epoch": 0.08328006574742032,
      "grad_norm": 0.2268109768629074,
      "learning_rate": 1.14e-05,
      "loss": 1.7996,
      "step": 57
    },
    {
      "epoch": 0.08474111953246279,
      "grad_norm": 0.24065515398979187,
      "learning_rate": 1.16e-05,
      "loss": 1.9203,
      "step": 58
    },
    {
      "epoch": 0.08620217331750525,
      "grad_norm": 0.23937389254570007,
      "learning_rate": 1.18e-05,
      "loss": 1.8845,
      "step": 59
    },
    {
      "epoch": 0.08766322710254772,
      "grad_norm": 0.22344107925891876,
      "learning_rate": 1.2e-05,
      "loss": 1.94,
      "step": 60
    },
    {
      "epoch": 0.08912428088759017,
      "grad_norm": 0.24668040871620178,
      "learning_rate": 1.22e-05,
      "loss": 1.9473,
      "step": 61
    },
    {
      "epoch": 0.09058533467263263,
      "grad_norm": 0.21312135457992554,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 1.8225,
      "step": 62
    },
    {
      "epoch": 0.0920463884576751,
      "grad_norm": 0.2296803593635559,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 1.7675,
      "step": 63
    },
    {
      "epoch": 0.09350744224271756,
      "grad_norm": 0.20073962211608887,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 1.7844,
      "step": 64
    },
    {
      "epoch": 0.09496849602776002,
      "grad_norm": 0.23866504430770874,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 1.7766,
      "step": 65
    },
    {
      "epoch": 0.09642954981280248,
      "grad_norm": 0.24550433456897736,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 1.9283,
      "step": 66
    },
    {
      "epoch": 0.09789060359784495,
      "grad_norm": 0.24073684215545654,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 1.841,
      "step": 67
    },
    {
      "epoch": 0.09935165738288741,
      "grad_norm": 0.21318458020687103,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 1.7465,
      "step": 68
    },
    {
      "epoch": 0.10081271116792986,
      "grad_norm": 0.2440112829208374,
      "learning_rate": 1.38e-05,
      "loss": 1.8503,
      "step": 69
    },
    {
      "epoch": 0.10227376495297233,
      "grad_norm": 0.21730957925319672,
      "learning_rate": 1.4e-05,
      "loss": 1.7624,
      "step": 70
    },
    {
      "epoch": 0.1037348187380148,
      "grad_norm": 0.22476676106452942,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 1.8127,
      "step": 71
    },
    {
      "epoch": 0.10519587252305726,
      "grad_norm": 0.2405499368906021,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 1.7755,
      "step": 72
    },
    {
      "epoch": 0.10665692630809971,
      "grad_norm": 0.24996110796928406,
      "learning_rate": 1.46e-05,
      "loss": 1.841,
      "step": 73
    },
    {
      "epoch": 0.10811798009314218,
      "grad_norm": 0.26147621870040894,
      "learning_rate": 1.48e-05,
      "loss": 1.7769,
      "step": 74
    },
    {
      "epoch": 0.10957903387818464,
      "grad_norm": 0.23422400653362274,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.7561,
      "step": 75
    },
    {
      "epoch": 0.11104008766322711,
      "grad_norm": 0.24707835912704468,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 1.7038,
      "step": 76
    },
    {
      "epoch": 0.11250114144826956,
      "grad_norm": 0.25473567843437195,
      "learning_rate": 1.54e-05,
      "loss": 1.7496,
      "step": 77
    },
    {
      "epoch": 0.11396219523331202,
      "grad_norm": 0.3016836941242218,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 1.7323,
      "step": 78
    },
    {
      "epoch": 0.11542324901835449,
      "grad_norm": 0.26107528805732727,
      "learning_rate": 1.58e-05,
      "loss": 1.7511,
      "step": 79
    },
    {
      "epoch": 0.11688430280339696,
      "grad_norm": 0.3029772639274597,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.691,
      "step": 80
    },
    {
      "epoch": 0.11834535658843941,
      "grad_norm": 0.30675315856933594,
      "learning_rate": 1.62e-05,
      "loss": 1.6696,
      "step": 81
    },
    {
      "epoch": 0.11980641037348187,
      "grad_norm": 0.3136438727378845,
      "learning_rate": 1.64e-05,
      "loss": 1.6624,
      "step": 82
    },
    {
      "epoch": 0.12126746415852434,
      "grad_norm": 0.32071617245674133,
      "learning_rate": 1.66e-05,
      "loss": 1.6458,
      "step": 83
    },
    {
      "epoch": 0.1227285179435668,
      "grad_norm": 0.3253510594367981,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.6169,
      "step": 84
    },
    {
      "epoch": 0.12418957172860925,
      "grad_norm": 0.3446858525276184,
      "learning_rate": 1.7e-05,
      "loss": 1.6721,
      "step": 85
    },
    {
      "epoch": 0.12565062551365172,
      "grad_norm": 0.33179470896720886,
      "learning_rate": 1.72e-05,
      "loss": 1.578,
      "step": 86
    },
    {
      "epoch": 0.12711167929869419,
      "grad_norm": 0.3791605830192566,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 1.6055,
      "step": 87
    },
    {
      "epoch": 0.12857273308373665,
      "grad_norm": 0.5245212912559509,
      "learning_rate": 1.76e-05,
      "loss": 1.626,
      "step": 88
    },
    {
      "epoch": 0.13003378686877912,
      "grad_norm": 0.43215855956077576,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 1.6177,
      "step": 89
    },
    {
      "epoch": 0.13149484065382158,
      "grad_norm": 0.4050828516483307,
      "learning_rate": 1.8e-05,
      "loss": 1.4903,
      "step": 90
    },
    {
      "epoch": 0.13295589443886402,
      "grad_norm": 0.399501234292984,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 1.6079,
      "step": 91
    },
    {
      "epoch": 0.13441694822390649,
      "grad_norm": 0.439622700214386,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 1.5405,
      "step": 92
    },
    {
      "epoch": 0.13587800200894895,
      "grad_norm": 0.4368193447589874,
      "learning_rate": 1.86e-05,
      "loss": 1.415,
      "step": 93
    },
    {
      "epoch": 0.13733905579399142,
      "grad_norm": 0.3644118010997772,
      "learning_rate": 1.88e-05,
      "loss": 1.525,
      "step": 94
    },
    {
      "epoch": 0.13880010957903388,
      "grad_norm": 0.3868708312511444,
      "learning_rate": 1.9e-05,
      "loss": 1.4903,
      "step": 95
    },
    {
      "epoch": 0.14026116336407635,
      "grad_norm": 0.43034952878952026,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 1.4605,
      "step": 96
    },
    {
      "epoch": 0.1417222171491188,
      "grad_norm": 0.4087560772895813,
      "learning_rate": 1.94e-05,
      "loss": 1.3544,
      "step": 97
    },
    {
      "epoch": 0.14318327093416128,
      "grad_norm": 0.29801666736602783,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 1.4098,
      "step": 98
    },
    {
      "epoch": 0.14464432471920372,
      "grad_norm": 0.275905966758728,
      "learning_rate": 1.98e-05,
      "loss": 1.4732,
      "step": 99
    },
    {
      "epoch": 0.14610537850424618,
      "grad_norm": 0.32271912693977356,
      "learning_rate": 2e-05,
      "loss": 1.4156,
      "step": 100
    },
    {
      "epoch": 0.14756643228928865,
      "grad_norm": 0.3191397190093994,
      "learning_rate": 1.9965753424657538e-05,
      "loss": 1.3254,
      "step": 101
    },
    {
      "epoch": 0.1490274860743311,
      "grad_norm": 0.26260653138160706,
      "learning_rate": 1.993150684931507e-05,
      "loss": 1.3877,
      "step": 102
    },
    {
      "epoch": 0.15048853985937358,
      "grad_norm": 0.2782766819000244,
      "learning_rate": 1.9897260273972604e-05,
      "loss": 1.3683,
      "step": 103
    },
    {
      "epoch": 0.15194959364441604,
      "grad_norm": 0.2510565221309662,
      "learning_rate": 1.9863013698630137e-05,
      "loss": 1.3996,
      "step": 104
    },
    {
      "epoch": 0.1534106474294585,
      "grad_norm": 0.2523151934146881,
      "learning_rate": 1.9828767123287674e-05,
      "loss": 1.3192,
      "step": 105
    },
    {
      "epoch": 0.15487170121450095,
      "grad_norm": 0.20559488236904144,
      "learning_rate": 1.9794520547945207e-05,
      "loss": 1.2096,
      "step": 106
    },
    {
      "epoch": 0.1563327549995434,
      "grad_norm": 0.17568816244602203,
      "learning_rate": 1.9760273972602743e-05,
      "loss": 1.3795,
      "step": 107
    },
    {
      "epoch": 0.15779380878458588,
      "grad_norm": 0.1778278350830078,
      "learning_rate": 1.9726027397260276e-05,
      "loss": 1.3146,
      "step": 108
    },
    {
      "epoch": 0.15925486256962834,
      "grad_norm": 0.18488670885562897,
      "learning_rate": 1.969178082191781e-05,
      "loss": 1.4105,
      "step": 109
    },
    {
      "epoch": 0.1607159163546708,
      "grad_norm": 0.1593291312456131,
      "learning_rate": 1.9657534246575346e-05,
      "loss": 1.3054,
      "step": 110
    },
    {
      "epoch": 0.16217697013971327,
      "grad_norm": 0.14311783015727997,
      "learning_rate": 1.962328767123288e-05,
      "loss": 1.3985,
      "step": 111
    },
    {
      "epoch": 0.16363802392475574,
      "grad_norm": 0.14948627352714539,
      "learning_rate": 1.9589041095890412e-05,
      "loss": 1.3395,
      "step": 112
    },
    {
      "epoch": 0.1650990777097982,
      "grad_norm": 0.14075608551502228,
      "learning_rate": 1.9554794520547945e-05,
      "loss": 1.3868,
      "step": 113
    },
    {
      "epoch": 0.16656013149484064,
      "grad_norm": 0.14439420402050018,
      "learning_rate": 1.952054794520548e-05,
      "loss": 1.2985,
      "step": 114
    },
    {
      "epoch": 0.1680211852798831,
      "grad_norm": 0.13425147533416748,
      "learning_rate": 1.9486301369863014e-05,
      "loss": 1.3855,
      "step": 115
    },
    {
      "epoch": 0.16948223906492557,
      "grad_norm": 0.1291724145412445,
      "learning_rate": 1.945205479452055e-05,
      "loss": 1.2942,
      "step": 116
    },
    {
      "epoch": 0.17094329284996804,
      "grad_norm": 0.1350926160812378,
      "learning_rate": 1.9417808219178084e-05,
      "loss": 1.3649,
      "step": 117
    },
    {
      "epoch": 0.1724043466350105,
      "grad_norm": 0.1309756338596344,
      "learning_rate": 1.9383561643835617e-05,
      "loss": 1.3241,
      "step": 118
    },
    {
      "epoch": 0.17386540042005297,
      "grad_norm": 0.12676255404949188,
      "learning_rate": 1.9349315068493153e-05,
      "loss": 1.3773,
      "step": 119
    },
    {
      "epoch": 0.17532645420509543,
      "grad_norm": 0.1361505538225174,
      "learning_rate": 1.9315068493150686e-05,
      "loss": 1.3259,
      "step": 120
    },
    {
      "epoch": 0.1767875079901379,
      "grad_norm": 0.1225372925400734,
      "learning_rate": 1.9280821917808223e-05,
      "loss": 1.2854,
      "step": 121
    },
    {
      "epoch": 0.17824856177518034,
      "grad_norm": 0.12647689878940582,
      "learning_rate": 1.9246575342465756e-05,
      "loss": 1.2741,
      "step": 122
    },
    {
      "epoch": 0.1797096155602228,
      "grad_norm": 0.12492359429597855,
      "learning_rate": 1.921232876712329e-05,
      "loss": 1.2728,
      "step": 123
    },
    {
      "epoch": 0.18117066934526527,
      "grad_norm": 0.1245495080947876,
      "learning_rate": 1.9178082191780822e-05,
      "loss": 1.2829,
      "step": 124
    },
    {
      "epoch": 0.18263172313030773,
      "grad_norm": 0.12087871879339218,
      "learning_rate": 1.9143835616438358e-05,
      "loss": 1.3673,
      "step": 125
    },
    {
      "epoch": 0.1840927769153502,
      "grad_norm": 0.13881978392601013,
      "learning_rate": 1.910958904109589e-05,
      "loss": 1.2749,
      "step": 126
    },
    {
      "epoch": 0.18555383070039266,
      "grad_norm": 0.12226665765047073,
      "learning_rate": 1.9075342465753424e-05,
      "loss": 1.3216,
      "step": 127
    },
    {
      "epoch": 0.18701488448543513,
      "grad_norm": 0.12877057492733002,
      "learning_rate": 1.904109589041096e-05,
      "loss": 1.3632,
      "step": 128
    },
    {
      "epoch": 0.1884759382704776,
      "grad_norm": 0.13042791187763214,
      "learning_rate": 1.9006849315068494e-05,
      "loss": 1.3087,
      "step": 129
    },
    {
      "epoch": 0.18993699205552003,
      "grad_norm": 0.1289220005273819,
      "learning_rate": 1.897260273972603e-05,
      "loss": 1.3569,
      "step": 130
    },
    {
      "epoch": 0.1913980458405625,
      "grad_norm": 0.12027924507856369,
      "learning_rate": 1.8938356164383563e-05,
      "loss": 1.3487,
      "step": 131
    },
    {
      "epoch": 0.19285909962560496,
      "grad_norm": 0.13009122014045715,
      "learning_rate": 1.8904109589041096e-05,
      "loss": 1.2639,
      "step": 132
    },
    {
      "epoch": 0.19432015341064743,
      "grad_norm": 0.13321325182914734,
      "learning_rate": 1.8869863013698633e-05,
      "loss": 1.2982,
      "step": 133
    },
    {
      "epoch": 0.1957812071956899,
      "grad_norm": 0.1279006004333496,
      "learning_rate": 1.8835616438356166e-05,
      "loss": 1.3291,
      "step": 134
    },
    {
      "epoch": 0.19724226098073236,
      "grad_norm": 0.13129518926143646,
      "learning_rate": 1.8801369863013702e-05,
      "loss": 1.2805,
      "step": 135
    },
    {
      "epoch": 0.19870331476577482,
      "grad_norm": 0.1193300113081932,
      "learning_rate": 1.8767123287671235e-05,
      "loss": 1.3173,
      "step": 136
    },
    {
      "epoch": 0.2001643685508173,
      "grad_norm": 0.12838037312030792,
      "learning_rate": 1.8732876712328768e-05,
      "loss": 1.3416,
      "step": 137
    },
    {
      "epoch": 0.20162542233585973,
      "grad_norm": 0.1269877851009369,
      "learning_rate": 1.86986301369863e-05,
      "loss": 1.295,
      "step": 138
    },
    {
      "epoch": 0.2030864761209022,
      "grad_norm": 0.13122180104255676,
      "learning_rate": 1.8664383561643838e-05,
      "loss": 1.293,
      "step": 139
    },
    {
      "epoch": 0.20454752990594466,
      "grad_norm": 0.1385333091020584,
      "learning_rate": 1.863013698630137e-05,
      "loss": 1.2237,
      "step": 140
    },
    {
      "epoch": 0.20600858369098712,
      "grad_norm": 0.1442401111125946,
      "learning_rate": 1.8595890410958907e-05,
      "loss": 1.2833,
      "step": 141
    },
    {
      "epoch": 0.2074696374760296,
      "grad_norm": 0.1236981600522995,
      "learning_rate": 1.856164383561644e-05,
      "loss": 1.3252,
      "step": 142
    },
    {
      "epoch": 0.20893069126107205,
      "grad_norm": 0.13414394855499268,
      "learning_rate": 1.8527397260273973e-05,
      "loss": 1.4099,
      "step": 143
    },
    {
      "epoch": 0.21039174504611452,
      "grad_norm": 0.1376720666885376,
      "learning_rate": 1.849315068493151e-05,
      "loss": 1.3735,
      "step": 144
    },
    {
      "epoch": 0.21185279883115699,
      "grad_norm": 0.13452522456645966,
      "learning_rate": 1.8458904109589043e-05,
      "loss": 1.2442,
      "step": 145
    },
    {
      "epoch": 0.21331385261619942,
      "grad_norm": 0.12783923745155334,
      "learning_rate": 1.8424657534246576e-05,
      "loss": 1.3028,
      "step": 146
    },
    {
      "epoch": 0.2147749064012419,
      "grad_norm": 0.1320268213748932,
      "learning_rate": 1.839041095890411e-05,
      "loss": 1.2906,
      "step": 147
    },
    {
      "epoch": 0.21623596018628435,
      "grad_norm": 0.14116773009300232,
      "learning_rate": 1.8356164383561645e-05,
      "loss": 1.3724,
      "step": 148
    },
    {
      "epoch": 0.21769701397132682,
      "grad_norm": 0.13176654279232025,
      "learning_rate": 1.8321917808219182e-05,
      "loss": 1.2719,
      "step": 149
    },
    {
      "epoch": 0.21915806775636928,
      "grad_norm": 0.13379769027233124,
      "learning_rate": 1.8287671232876715e-05,
      "loss": 1.3286,
      "step": 150
    },
    {
      "epoch": 0.22061912154141175,
      "grad_norm": 0.13172045350074768,
      "learning_rate": 1.8253424657534248e-05,
      "loss": 1.3104,
      "step": 151
    },
    {
      "epoch": 0.22208017532645422,
      "grad_norm": 0.14471982419490814,
      "learning_rate": 1.821917808219178e-05,
      "loss": 1.3111,
      "step": 152
    },
    {
      "epoch": 0.22354122911149665,
      "grad_norm": 0.14093175530433655,
      "learning_rate": 1.8184931506849317e-05,
      "loss": 1.2696,
      "step": 153
    },
    {
      "epoch": 0.22500228289653912,
      "grad_norm": 0.14340919256210327,
      "learning_rate": 1.815068493150685e-05,
      "loss": 1.3105,
      "step": 154
    },
    {
      "epoch": 0.22646333668158158,
      "grad_norm": 0.15136584639549255,
      "learning_rate": 1.8116438356164387e-05,
      "loss": 1.1514,
      "step": 155
    },
    {
      "epoch": 0.22792439046662405,
      "grad_norm": 0.16463331878185272,
      "learning_rate": 1.808219178082192e-05,
      "loss": 1.2292,
      "step": 156
    },
    {
      "epoch": 0.22938544425166651,
      "grad_norm": 0.14684943854808807,
      "learning_rate": 1.8047945205479453e-05,
      "loss": 1.1744,
      "step": 157
    },
    {
      "epoch": 0.23084649803670898,
      "grad_norm": 0.17171254754066467,
      "learning_rate": 1.801369863013699e-05,
      "loss": 1.2522,
      "step": 158
    },
    {
      "epoch": 0.23230755182175145,
      "grad_norm": 0.1535484492778778,
      "learning_rate": 1.7979452054794522e-05,
      "loss": 1.3086,
      "step": 159
    },
    {
      "epoch": 0.2337686056067939,
      "grad_norm": 0.15115784108638763,
      "learning_rate": 1.7945205479452055e-05,
      "loss": 1.3126,
      "step": 160
    },
    {
      "epoch": 0.23522965939183635,
      "grad_norm": 0.15851254761219025,
      "learning_rate": 1.791095890410959e-05,
      "loss": 1.3022,
      "step": 161
    },
    {
      "epoch": 0.23669071317687881,
      "grad_norm": 0.16716252267360687,
      "learning_rate": 1.7876712328767125e-05,
      "loss": 1.2339,
      "step": 162
    },
    {
      "epoch": 0.23815176696192128,
      "grad_norm": 0.1609048694372177,
      "learning_rate": 1.7842465753424658e-05,
      "loss": 1.3374,
      "step": 163
    },
    {
      "epoch": 0.23961282074696375,
      "grad_norm": 0.17205393314361572,
      "learning_rate": 1.7808219178082194e-05,
      "loss": 1.2043,
      "step": 164
    },
    {
      "epoch": 0.2410738745320062,
      "grad_norm": 0.17497386038303375,
      "learning_rate": 1.7773972602739727e-05,
      "loss": 1.2267,
      "step": 165
    },
    {
      "epoch": 0.24253492831704868,
      "grad_norm": 0.17982399463653564,
      "learning_rate": 1.773972602739726e-05,
      "loss": 1.2363,
      "step": 166
    },
    {
      "epoch": 0.24399598210209114,
      "grad_norm": 0.17415454983711243,
      "learning_rate": 1.7705479452054797e-05,
      "loss": 1.2339,
      "step": 167
    },
    {
      "epoch": 0.2454570358871336,
      "grad_norm": 0.17761662602424622,
      "learning_rate": 1.767123287671233e-05,
      "loss": 1.2549,
      "step": 168
    },
    {
      "epoch": 0.24691808967217604,
      "grad_norm": 0.17029732465744019,
      "learning_rate": 1.7636986301369866e-05,
      "loss": 1.2175,
      "step": 169
    },
    {
      "epoch": 0.2483791434572185,
      "grad_norm": 0.1831396073102951,
      "learning_rate": 1.76027397260274e-05,
      "loss": 1.2975,
      "step": 170
    },
    {
      "epoch": 0.24984019724226098,
      "grad_norm": 0.1648183912038803,
      "learning_rate": 1.7568493150684932e-05,
      "loss": 1.317,
      "step": 171
    },
    {
      "epoch": 0.25130125102730344,
      "grad_norm": 0.19559577107429504,
      "learning_rate": 1.7534246575342465e-05,
      "loss": 1.2482,
      "step": 172
    },
    {
      "epoch": 0.2527623048123459,
      "grad_norm": 0.17486748099327087,
      "learning_rate": 1.7500000000000002e-05,
      "loss": 1.2638,
      "step": 173
    },
    {
      "epoch": 0.25422335859738837,
      "grad_norm": 0.18816767632961273,
      "learning_rate": 1.7465753424657538e-05,
      "loss": 1.2739,
      "step": 174
    },
    {
      "epoch": 0.25568441238243084,
      "grad_norm": 0.1921810507774353,
      "learning_rate": 1.743150684931507e-05,
      "loss": 1.2028,
      "step": 175
    },
    {
      "epoch": 0.2571454661674733,
      "grad_norm": 0.20229558646678925,
      "learning_rate": 1.7397260273972604e-05,
      "loss": 1.2518,
      "step": 176
    },
    {
      "epoch": 0.25860651995251577,
      "grad_norm": 0.21454092860221863,
      "learning_rate": 1.7363013698630137e-05,
      "loss": 1.2257,
      "step": 177
    },
    {
      "epoch": 0.26006757373755823,
      "grad_norm": 0.20349366962909698,
      "learning_rate": 1.7328767123287674e-05,
      "loss": 1.229,
      "step": 178
    },
    {
      "epoch": 0.2615286275226007,
      "grad_norm": 0.22348977625370026,
      "learning_rate": 1.7294520547945207e-05,
      "loss": 1.3047,
      "step": 179
    },
    {
      "epoch": 0.26298968130764316,
      "grad_norm": 0.18413174152374268,
      "learning_rate": 1.726027397260274e-05,
      "loss": 1.2372,
      "step": 180
    },
    {
      "epoch": 0.2644507350926856,
      "grad_norm": 0.22073839604854584,
      "learning_rate": 1.7226027397260273e-05,
      "loss": 1.2719,
      "step": 181
    },
    {
      "epoch": 0.26591178887772804,
      "grad_norm": 0.25906509160995483,
      "learning_rate": 1.719178082191781e-05,
      "loss": 1.2132,
      "step": 182
    },
    {
      "epoch": 0.2673728426627705,
      "grad_norm": 0.22006724774837494,
      "learning_rate": 1.7157534246575346e-05,
      "loss": 1.2595,
      "step": 183
    },
    {
      "epoch": 0.26883389644781297,
      "grad_norm": 0.2628309428691864,
      "learning_rate": 1.712328767123288e-05,
      "loss": 1.1812,
      "step": 184
    },
    {
      "epoch": 0.27029495023285544,
      "grad_norm": 0.21583111584186554,
      "learning_rate": 1.7089041095890412e-05,
      "loss": 1.2631,
      "step": 185
    },
    {
      "epoch": 0.2717560040178979,
      "grad_norm": 0.19774451851844788,
      "learning_rate": 1.7054794520547945e-05,
      "loss": 1.2511,
      "step": 186
    },
    {
      "epoch": 0.27321705780294037,
      "grad_norm": 0.199305459856987,
      "learning_rate": 1.702054794520548e-05,
      "loss": 1.2836,
      "step": 187
    },
    {
      "epoch": 0.27467811158798283,
      "grad_norm": 0.17964106798171997,
      "learning_rate": 1.6986301369863014e-05,
      "loss": 1.2784,
      "step": 188
    },
    {
      "epoch": 0.2761391653730253,
      "grad_norm": 0.15635524690151215,
      "learning_rate": 1.695205479452055e-05,
      "loss": 1.2959,
      "step": 189
    },
    {
      "epoch": 0.27760021915806776,
      "grad_norm": 0.14683344960212708,
      "learning_rate": 1.6917808219178084e-05,
      "loss": 1.2453,
      "step": 190
    },
    {
      "epoch": 0.27906127294311023,
      "grad_norm": 0.1619580239057541,
      "learning_rate": 1.6883561643835617e-05,
      "loss": 1.2602,
      "step": 191
    },
    {
      "epoch": 0.2805223267281527,
      "grad_norm": 0.19470493495464325,
      "learning_rate": 1.6849315068493153e-05,
      "loss": 1.1052,
      "step": 192
    },
    {
      "epoch": 0.28198338051319516,
      "grad_norm": 0.17526264488697052,
      "learning_rate": 1.6815068493150686e-05,
      "loss": 1.2553,
      "step": 193
    },
    {
      "epoch": 0.2834444342982376,
      "grad_norm": 0.1810723841190338,
      "learning_rate": 1.678082191780822e-05,
      "loss": 1.2121,
      "step": 194
    },
    {
      "epoch": 0.2849054880832801,
      "grad_norm": 0.16387374699115753,
      "learning_rate": 1.6746575342465753e-05,
      "loss": 1.1702,
      "step": 195
    },
    {
      "epoch": 0.28636654186832255,
      "grad_norm": 0.1537161022424698,
      "learning_rate": 1.671232876712329e-05,
      "loss": 1.1865,
      "step": 196
    },
    {
      "epoch": 0.28782759565336496,
      "grad_norm": 0.13615332543849945,
      "learning_rate": 1.6678082191780822e-05,
      "loss": 1.2578,
      "step": 197
    },
    {
      "epoch": 0.28928864943840743,
      "grad_norm": 0.13642196357250214,
      "learning_rate": 1.664383561643836e-05,
      "loss": 1.1813,
      "step": 198
    },
    {
      "epoch": 0.2907497032234499,
      "grad_norm": 0.1444728523492813,
      "learning_rate": 1.660958904109589e-05,
      "loss": 1.2815,
      "step": 199
    },
    {
      "epoch": 0.29221075700849236,
      "grad_norm": 0.13030050694942474,
      "learning_rate": 1.6575342465753425e-05,
      "loss": 1.2848,
      "step": 200
    },
    {
      "epoch": 0.2936718107935348,
      "grad_norm": 0.13471786677837372,
      "learning_rate": 1.654109589041096e-05,
      "loss": 1.1634,
      "step": 201
    },
    {
      "epoch": 0.2951328645785773,
      "grad_norm": 0.11596754193305969,
      "learning_rate": 1.6506849315068494e-05,
      "loss": 1.2522,
      "step": 202
    },
    {
      "epoch": 0.29659391836361976,
      "grad_norm": 0.11978977173566818,
      "learning_rate": 1.647260273972603e-05,
      "loss": 1.2585,
      "step": 203
    },
    {
      "epoch": 0.2980549721486622,
      "grad_norm": 0.11857204139232635,
      "learning_rate": 1.6438356164383563e-05,
      "loss": 1.1372,
      "step": 204
    },
    {
      "epoch": 0.2995160259337047,
      "grad_norm": 0.12098690867424011,
      "learning_rate": 1.6404109589041096e-05,
      "loss": 1.1391,
      "step": 205
    },
    {
      "epoch": 0.30097707971874715,
      "grad_norm": 0.12197306752204895,
      "learning_rate": 1.6369863013698633e-05,
      "loss": 1.2073,
      "step": 206
    },
    {
      "epoch": 0.3024381335037896,
      "grad_norm": 0.11093982309103012,
      "learning_rate": 1.6335616438356166e-05,
      "loss": 1.3644,
      "step": 207
    },
    {
      "epoch": 0.3038991872888321,
      "grad_norm": 0.11352576315402985,
      "learning_rate": 1.6301369863013702e-05,
      "loss": 1.2034,
      "step": 208
    },
    {
      "epoch": 0.30536024107387455,
      "grad_norm": 0.1127958819270134,
      "learning_rate": 1.6267123287671232e-05,
      "loss": 1.241,
      "step": 209
    },
    {
      "epoch": 0.306821294858917,
      "grad_norm": 0.10320553183555603,
      "learning_rate": 1.623287671232877e-05,
      "loss": 1.2169,
      "step": 210
    },
    {
      "epoch": 0.3082823486439595,
      "grad_norm": 0.11075271666049957,
      "learning_rate": 1.61986301369863e-05,
      "loss": 1.2867,
      "step": 211
    },
    {
      "epoch": 0.3097434024290019,
      "grad_norm": 0.10098811239004135,
      "learning_rate": 1.6164383561643838e-05,
      "loss": 1.2625,
      "step": 212
    },
    {
      "epoch": 0.31120445621404436,
      "grad_norm": 0.10328993201255798,
      "learning_rate": 1.613013698630137e-05,
      "loss": 1.2121,
      "step": 213
    },
    {
      "epoch": 0.3126655099990868,
      "grad_norm": 0.11792083084583282,
      "learning_rate": 1.6095890410958904e-05,
      "loss": 1.2316,
      "step": 214
    },
    {
      "epoch": 0.3141265637841293,
      "grad_norm": 0.1143653616309166,
      "learning_rate": 1.606164383561644e-05,
      "loss": 1.2084,
      "step": 215
    },
    {
      "epoch": 0.31558761756917175,
      "grad_norm": 0.10241192579269409,
      "learning_rate": 1.6027397260273974e-05,
      "loss": 1.2924,
      "step": 216
    },
    {
      "epoch": 0.3170486713542142,
      "grad_norm": 0.09764024615287781,
      "learning_rate": 1.599315068493151e-05,
      "loss": 1.2387,
      "step": 217
    },
    {
      "epoch": 0.3185097251392567,
      "grad_norm": 0.0964062437415123,
      "learning_rate": 1.5958904109589043e-05,
      "loss": 1.2143,
      "step": 218
    },
    {
      "epoch": 0.31997077892429915,
      "grad_norm": 0.10618474334478378,
      "learning_rate": 1.5924657534246576e-05,
      "loss": 1.1912,
      "step": 219
    },
    {
      "epoch": 0.3214318327093416,
      "grad_norm": 0.10085848718881607,
      "learning_rate": 1.589041095890411e-05,
      "loss": 1.1691,
      "step": 220
    },
    {
      "epoch": 0.3228928864943841,
      "grad_norm": 0.10805616527795792,
      "learning_rate": 1.5856164383561646e-05,
      "loss": 1.2028,
      "step": 221
    },
    {
      "epoch": 0.32435394027942654,
      "grad_norm": 0.10075750201940536,
      "learning_rate": 1.5821917808219182e-05,
      "loss": 1.2286,
      "step": 222
    },
    {
      "epoch": 0.325814994064469,
      "grad_norm": 0.09440125524997711,
      "learning_rate": 1.5787671232876715e-05,
      "loss": 1.24,
      "step": 223
    },
    {
      "epoch": 0.3272760478495115,
      "grad_norm": 0.09914068877696991,
      "learning_rate": 1.5753424657534248e-05,
      "loss": 1.2225,
      "step": 224
    },
    {
      "epoch": 0.32873710163455394,
      "grad_norm": 0.10779386013746262,
      "learning_rate": 1.571917808219178e-05,
      "loss": 1.1821,
      "step": 225
    },
    {
      "epoch": 0.3301981554195964,
      "grad_norm": 0.09676062315702438,
      "learning_rate": 1.5684931506849318e-05,
      "loss": 1.2975,
      "step": 226
    },
    {
      "epoch": 0.33165920920463887,
      "grad_norm": 0.09622418135404587,
      "learning_rate": 1.565068493150685e-05,
      "loss": 1.2935,
      "step": 227
    },
    {
      "epoch": 0.3331202629896813,
      "grad_norm": 0.10332711786031723,
      "learning_rate": 1.5616438356164384e-05,
      "loss": 1.2178,
      "step": 228
    },
    {
      "epoch": 0.33458131677472375,
      "grad_norm": 0.10794605314731598,
      "learning_rate": 1.5582191780821917e-05,
      "loss": 1.1103,
      "step": 229
    },
    {
      "epoch": 0.3360423705597662,
      "grad_norm": 0.10352062433958054,
      "learning_rate": 1.5547945205479453e-05,
      "loss": 1.2187,
      "step": 230
    },
    {
      "epoch": 0.3375034243448087,
      "grad_norm": 0.10264533013105392,
      "learning_rate": 1.551369863013699e-05,
      "loss": 1.3449,
      "step": 231
    },
    {
      "epoch": 0.33896447812985114,
      "grad_norm": 0.10094834864139557,
      "learning_rate": 1.5479452054794523e-05,
      "loss": 1.2042,
      "step": 232
    },
    {
      "epoch": 0.3404255319148936,
      "grad_norm": 0.09931448101997375,
      "learning_rate": 1.5445205479452056e-05,
      "loss": 1.1876,
      "step": 233
    },
    {
      "epoch": 0.3418865856999361,
      "grad_norm": 0.09288407117128372,
      "learning_rate": 1.541095890410959e-05,
      "loss": 1.2163,
      "step": 234
    },
    {
      "epoch": 0.34334763948497854,
      "grad_norm": 0.09983450174331665,
      "learning_rate": 1.5376712328767125e-05,
      "loss": 1.2118,
      "step": 235
    },
    {
      "epoch": 0.344808693270021,
      "grad_norm": 0.10181832313537598,
      "learning_rate": 1.5342465753424658e-05,
      "loss": 1.2189,
      "step": 236
    },
    {
      "epoch": 0.34626974705506347,
      "grad_norm": 0.10253550857305527,
      "learning_rate": 1.5308219178082195e-05,
      "loss": 1.2415,
      "step": 237
    },
    {
      "epoch": 0.34773080084010594,
      "grad_norm": 0.09938843548297882,
      "learning_rate": 1.5273972602739728e-05,
      "loss": 1.2391,
      "step": 238
    },
    {
      "epoch": 0.3491918546251484,
      "grad_norm": 0.09904040396213531,
      "learning_rate": 1.523972602739726e-05,
      "loss": 1.147,
      "step": 239
    },
    {
      "epoch": 0.35065290841019087,
      "grad_norm": 0.1011345386505127,
      "learning_rate": 1.5205479452054797e-05,
      "loss": 1.2801,
      "step": 240
    },
    {
      "epoch": 0.35211396219523333,
      "grad_norm": 0.10546337813138962,
      "learning_rate": 1.517123287671233e-05,
      "loss": 1.2179,
      "step": 241
    },
    {
      "epoch": 0.3535750159802758,
      "grad_norm": 0.09379958361387253,
      "learning_rate": 1.5136986301369865e-05,
      "loss": 1.2078,
      "step": 242
    },
    {
      "epoch": 0.35503606976531826,
      "grad_norm": 0.09210502356290817,
      "learning_rate": 1.5102739726027398e-05,
      "loss": 1.2126,
      "step": 243
    },
    {
      "epoch": 0.3564971235503607,
      "grad_norm": 0.0911347046494484,
      "learning_rate": 1.5068493150684933e-05,
      "loss": 1.2353,
      "step": 244
    },
    {
      "epoch": 0.35795817733540314,
      "grad_norm": 0.10343588888645172,
      "learning_rate": 1.5034246575342466e-05,
      "loss": 1.2063,
      "step": 245
    },
    {
      "epoch": 0.3594192311204456,
      "grad_norm": 0.10774116218090057,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.1895,
      "step": 246
    },
    {
      "epoch": 0.36088028490548807,
      "grad_norm": 0.09309092164039612,
      "learning_rate": 1.4965753424657537e-05,
      "loss": 1.2661,
      "step": 247
    },
    {
      "epoch": 0.36234133869053053,
      "grad_norm": 0.09687670320272446,
      "learning_rate": 1.493150684931507e-05,
      "loss": 1.2154,
      "step": 248
    },
    {
      "epoch": 0.363802392475573,
      "grad_norm": 0.09125279635190964,
      "learning_rate": 1.4897260273972605e-05,
      "loss": 1.2367,
      "step": 249
    },
    {
      "epoch": 0.36526344626061547,
      "grad_norm": 0.11119771748781204,
      "learning_rate": 1.4863013698630138e-05,
      "loss": 1.1032,
      "step": 250
    },
    {
      "epoch": 0.36672450004565793,
      "grad_norm": 0.11136704683303833,
      "learning_rate": 1.4828767123287672e-05,
      "loss": 1.1679,
      "step": 251
    },
    {
      "epoch": 0.3681855538307004,
      "grad_norm": 0.11276744306087494,
      "learning_rate": 1.4794520547945205e-05,
      "loss": 1.1268,
      "step": 252
    },
    {
      "epoch": 0.36964660761574286,
      "grad_norm": 0.10903234779834747,
      "learning_rate": 1.4760273972602742e-05,
      "loss": 1.1681,
      "step": 253
    },
    {
      "epoch": 0.3711076614007853,
      "grad_norm": 0.09586647897958755,
      "learning_rate": 1.4726027397260275e-05,
      "loss": 1.2486,
      "step": 254
    },
    {
      "epoch": 0.3725687151858278,
      "grad_norm": 0.09748208522796631,
      "learning_rate": 1.469178082191781e-05,
      "loss": 1.3186,
      "step": 255
    },
    {
      "epoch": 0.37402976897087026,
      "grad_norm": 0.10351759195327759,
      "learning_rate": 1.4657534246575344e-05,
      "loss": 1.1344,
      "step": 256
    },
    {
      "epoch": 0.3754908227559127,
      "grad_norm": 0.11112543940544128,
      "learning_rate": 1.4623287671232877e-05,
      "loss": 1.2433,
      "step": 257
    },
    {
      "epoch": 0.3769518765409552,
      "grad_norm": 0.09291627258062363,
      "learning_rate": 1.4589041095890412e-05,
      "loss": 1.1717,
      "step": 258
    },
    {
      "epoch": 0.3784129303259976,
      "grad_norm": 0.11696401238441467,
      "learning_rate": 1.4554794520547945e-05,
      "loss": 1.1889,
      "step": 259
    },
    {
      "epoch": 0.37987398411104006,
      "grad_norm": 0.09728217869997025,
      "learning_rate": 1.4520547945205482e-05,
      "loss": 1.1756,
      "step": 260
    },
    {
      "epoch": 0.38133503789608253,
      "grad_norm": 0.10936015099287033,
      "learning_rate": 1.4486301369863015e-05,
      "loss": 1.1526,
      "step": 261
    },
    {
      "epoch": 0.382796091681125,
      "grad_norm": 0.09887027740478516,
      "learning_rate": 1.445205479452055e-05,
      "loss": 1.1556,
      "step": 262
    },
    {
      "epoch": 0.38425714546616746,
      "grad_norm": 0.09080694615840912,
      "learning_rate": 1.4417808219178084e-05,
      "loss": 1.1592,
      "step": 263
    },
    {
      "epoch": 0.3857181992512099,
      "grad_norm": 0.09273724257946014,
      "learning_rate": 1.4383561643835617e-05,
      "loss": 1.1994,
      "step": 264
    },
    {
      "epoch": 0.3871792530362524,
      "grad_norm": 0.10300930589437485,
      "learning_rate": 1.4349315068493152e-05,
      "loss": 1.2292,
      "step": 265
    },
    {
      "epoch": 0.38864030682129486,
      "grad_norm": 0.10504985600709915,
      "learning_rate": 1.4315068493150685e-05,
      "loss": 1.1623,
      "step": 266
    },
    {
      "epoch": 0.3901013606063373,
      "grad_norm": 0.10123489052057266,
      "learning_rate": 1.4280821917808221e-05,
      "loss": 1.1435,
      "step": 267
    },
    {
      "epoch": 0.3915624143913798,
      "grad_norm": 0.09777438640594482,
      "learning_rate": 1.4246575342465754e-05,
      "loss": 1.2014,
      "step": 268
    },
    {
      "epoch": 0.39302346817642225,
      "grad_norm": 0.10096925497055054,
      "learning_rate": 1.421232876712329e-05,
      "loss": 1.243,
      "step": 269
    },
    {
      "epoch": 0.3944845219614647,
      "grad_norm": 0.11357256770133972,
      "learning_rate": 1.4178082191780822e-05,
      "loss": 1.1434,
      "step": 270
    },
    {
      "epoch": 0.3959455757465072,
      "grad_norm": 0.09454260766506195,
      "learning_rate": 1.4143835616438357e-05,
      "loss": 1.257,
      "step": 271
    },
    {
      "epoch": 0.39740662953154965,
      "grad_norm": 0.0995330736041069,
      "learning_rate": 1.4109589041095892e-05,
      "loss": 1.2173,
      "step": 272
    },
    {
      "epoch": 0.3988676833165921,
      "grad_norm": 0.09766160696744919,
      "learning_rate": 1.4075342465753425e-05,
      "loss": 1.1523,
      "step": 273
    },
    {
      "epoch": 0.4003287371016346,
      "grad_norm": 0.09799221158027649,
      "learning_rate": 1.4041095890410961e-05,
      "loss": 1.2785,
      "step": 274
    },
    {
      "epoch": 0.401789790886677,
      "grad_norm": 0.11043940484523773,
      "learning_rate": 1.4006849315068494e-05,
      "loss": 1.1715,
      "step": 275
    },
    {
      "epoch": 0.40325084467171946,
      "grad_norm": 0.09611232578754425,
      "learning_rate": 1.3972602739726029e-05,
      "loss": 1.1689,
      "step": 276
    },
    {
      "epoch": 0.4047118984567619,
      "grad_norm": 0.0990489274263382,
      "learning_rate": 1.3938356164383562e-05,
      "loss": 1.2633,
      "step": 277
    },
    {
      "epoch": 0.4061729522418044,
      "grad_norm": 0.09898124635219574,
      "learning_rate": 1.3904109589041097e-05,
      "loss": 1.2649,
      "step": 278
    },
    {
      "epoch": 0.40763400602684685,
      "grad_norm": 0.10052936524152756,
      "learning_rate": 1.3869863013698633e-05,
      "loss": 1.1938,
      "step": 279
    },
    {
      "epoch": 0.4090950598118893,
      "grad_norm": 0.10725940018892288,
      "learning_rate": 1.3835616438356164e-05,
      "loss": 1.2371,
      "step": 280
    },
    {
      "epoch": 0.4105561135969318,
      "grad_norm": 0.10019299387931824,
      "learning_rate": 1.3801369863013701e-05,
      "loss": 1.2738,
      "step": 281
    },
    {
      "epoch": 0.41201716738197425,
      "grad_norm": 0.10612376034259796,
      "learning_rate": 1.3767123287671234e-05,
      "loss": 1.1666,
      "step": 282
    },
    {
      "epoch": 0.4134782211670167,
      "grad_norm": 0.1012573391199112,
      "learning_rate": 1.3732876712328769e-05,
      "loss": 1.2489,
      "step": 283
    },
    {
      "epoch": 0.4149392749520592,
      "grad_norm": 0.10012490302324295,
      "learning_rate": 1.3698630136986302e-05,
      "loss": 1.2934,
      "step": 284
    },
    {
      "epoch": 0.41640032873710164,
      "grad_norm": 0.09684241563081741,
      "learning_rate": 1.3664383561643836e-05,
      "loss": 1.2547,
      "step": 285
    },
    {
      "epoch": 0.4178613825221441,
      "grad_norm": 0.10791518539190292,
      "learning_rate": 1.363013698630137e-05,
      "loss": 1.2118,
      "step": 286
    },
    {
      "epoch": 0.4193224363071866,
      "grad_norm": 0.10714226961135864,
      "learning_rate": 1.3595890410958906e-05,
      "loss": 1.218,
      "step": 287
    },
    {
      "epoch": 0.42078349009222904,
      "grad_norm": 0.10130172967910767,
      "learning_rate": 1.356164383561644e-05,
      "loss": 1.2244,
      "step": 288
    },
    {
      "epoch": 0.4222445438772715,
      "grad_norm": 0.0985652357339859,
      "learning_rate": 1.3527397260273974e-05,
      "loss": 1.2228,
      "step": 289
    },
    {
      "epoch": 0.42370559766231397,
      "grad_norm": 0.09767841547727585,
      "learning_rate": 1.3493150684931508e-05,
      "loss": 1.1981,
      "step": 290
    },
    {
      "epoch": 0.4251666514473564,
      "grad_norm": 0.09126268327236176,
      "learning_rate": 1.3458904109589042e-05,
      "loss": 1.2296,
      "step": 291
    },
    {
      "epoch": 0.42662770523239885,
      "grad_norm": 0.09404008835554123,
      "learning_rate": 1.3424657534246576e-05,
      "loss": 1.3013,
      "step": 292
    },
    {
      "epoch": 0.4280887590174413,
      "grad_norm": 0.09837393462657928,
      "learning_rate": 1.339041095890411e-05,
      "loss": 1.2102,
      "step": 293
    },
    {
      "epoch": 0.4295498128024838,
      "grad_norm": 0.0978068932890892,
      "learning_rate": 1.3356164383561646e-05,
      "loss": 1.1995,
      "step": 294
    },
    {
      "epoch": 0.43101086658752624,
      "grad_norm": 0.09333440661430359,
      "learning_rate": 1.332191780821918e-05,
      "loss": 1.2284,
      "step": 295
    },
    {
      "epoch": 0.4324719203725687,
      "grad_norm": 0.10575850307941437,
      "learning_rate": 1.3287671232876714e-05,
      "loss": 1.1442,
      "step": 296
    },
    {
      "epoch": 0.4339329741576112,
      "grad_norm": 0.10079637169837952,
      "learning_rate": 1.3253424657534248e-05,
      "loss": 1.2114,
      "step": 297
    },
    {
      "epoch": 0.43539402794265364,
      "grad_norm": 0.10534002631902695,
      "learning_rate": 1.3219178082191781e-05,
      "loss": 1.1894,
      "step": 298
    },
    {
      "epoch": 0.4368550817276961,
      "grad_norm": 0.09364209324121475,
      "learning_rate": 1.3184931506849316e-05,
      "loss": 1.1565,
      "step": 299
    },
    {
      "epoch": 0.43831613551273857,
      "grad_norm": 0.09970992058515549,
      "learning_rate": 1.3150684931506849e-05,
      "loss": 1.234,
      "step": 300
    },
    {
      "epoch": 0.43977718929778103,
      "grad_norm": 0.10115770250558853,
      "learning_rate": 1.3116438356164385e-05,
      "loss": 1.2113,
      "step": 301
    },
    {
      "epoch": 0.4412382430828235,
      "grad_norm": 0.10583057254552841,
      "learning_rate": 1.3082191780821919e-05,
      "loss": 1.1469,
      "step": 302
    },
    {
      "epoch": 0.44269929686786597,
      "grad_norm": 0.10508885979652405,
      "learning_rate": 1.3047945205479453e-05,
      "loss": 1.1648,
      "step": 303
    },
    {
      "epoch": 0.44416035065290843,
      "grad_norm": 0.09814278036355972,
      "learning_rate": 1.3013698630136988e-05,
      "loss": 1.1365,
      "step": 304
    },
    {
      "epoch": 0.4456214044379509,
      "grad_norm": 0.0985088050365448,
      "learning_rate": 1.2979452054794521e-05,
      "loss": 1.2589,
      "step": 305
    },
    {
      "epoch": 0.4470824582229933,
      "grad_norm": 0.10917991399765015,
      "learning_rate": 1.2945205479452056e-05,
      "loss": 1.2256,
      "step": 306
    },
    {
      "epoch": 0.44854351200803577,
      "grad_norm": 0.10406248271465302,
      "learning_rate": 1.2910958904109589e-05,
      "loss": 1.2597,
      "step": 307
    },
    {
      "epoch": 0.45000456579307824,
      "grad_norm": 0.09909600764513016,
      "learning_rate": 1.2876712328767125e-05,
      "loss": 1.2125,
      "step": 308
    },
    {
      "epoch": 0.4514656195781207,
      "grad_norm": 0.10518831014633179,
      "learning_rate": 1.2842465753424658e-05,
      "loss": 1.1912,
      "step": 309
    },
    {
      "epoch": 0.45292667336316317,
      "grad_norm": 0.09970947355031967,
      "learning_rate": 1.2808219178082193e-05,
      "loss": 1.207,
      "step": 310
    },
    {
      "epoch": 0.45438772714820563,
      "grad_norm": 0.10252334922552109,
      "learning_rate": 1.2773972602739728e-05,
      "loss": 1.2133,
      "step": 311
    },
    {
      "epoch": 0.4558487809332481,
      "grad_norm": 0.10719390958547592,
      "learning_rate": 1.273972602739726e-05,
      "loss": 1.146,
      "step": 312
    },
    {
      "epoch": 0.45730983471829056,
      "grad_norm": 0.11196637153625488,
      "learning_rate": 1.2705479452054796e-05,
      "loss": 1.2922,
      "step": 313
    },
    {
      "epoch": 0.45877088850333303,
      "grad_norm": 0.11517077684402466,
      "learning_rate": 1.2671232876712329e-05,
      "loss": 1.2577,
      "step": 314
    },
    {
      "epoch": 0.4602319422883755,
      "grad_norm": 0.09453711658716202,
      "learning_rate": 1.2636986301369865e-05,
      "loss": 1.3212,
      "step": 315
    },
    {
      "epoch": 0.46169299607341796,
      "grad_norm": 0.1068459153175354,
      "learning_rate": 1.2602739726027398e-05,
      "loss": 1.1981,
      "step": 316
    },
    {
      "epoch": 0.4631540498584604,
      "grad_norm": 0.0985800251364708,
      "learning_rate": 1.2568493150684933e-05,
      "loss": 1.189,
      "step": 317
    },
    {
      "epoch": 0.4646151036435029,
      "grad_norm": 0.10522795468568802,
      "learning_rate": 1.2534246575342466e-05,
      "loss": 1.2341,
      "step": 318
    },
    {
      "epoch": 0.46607615742854536,
      "grad_norm": 0.10663071274757385,
      "learning_rate": 1.25e-05,
      "loss": 1.1654,
      "step": 319
    },
    {
      "epoch": 0.4675372112135878,
      "grad_norm": 0.10541412234306335,
      "learning_rate": 1.2465753424657537e-05,
      "loss": 1.3409,
      "step": 320
    },
    {
      "epoch": 0.4689982649986303,
      "grad_norm": 0.10836822539567947,
      "learning_rate": 1.243150684931507e-05,
      "loss": 1.2551,
      "step": 321
    },
    {
      "epoch": 0.4704593187836727,
      "grad_norm": 0.09797906875610352,
      "learning_rate": 1.2397260273972605e-05,
      "loss": 1.1954,
      "step": 322
    },
    {
      "epoch": 0.47192037256871516,
      "grad_norm": 0.10791884362697601,
      "learning_rate": 1.2363013698630138e-05,
      "loss": 1.2677,
      "step": 323
    },
    {
      "epoch": 0.47338142635375763,
      "grad_norm": 0.10603371262550354,
      "learning_rate": 1.2328767123287673e-05,
      "loss": 1.1747,
      "step": 324
    },
    {
      "epoch": 0.4748424801388001,
      "grad_norm": 0.09947334975004196,
      "learning_rate": 1.2294520547945206e-05,
      "loss": 1.24,
      "step": 325
    },
    {
      "epoch": 0.47630353392384256,
      "grad_norm": 0.0948692336678505,
      "learning_rate": 1.226027397260274e-05,
      "loss": 1.3252,
      "step": 326
    },
    {
      "epoch": 0.477764587708885,
      "grad_norm": 0.11293943971395493,
      "learning_rate": 1.2226027397260273e-05,
      "loss": 1.2193,
      "step": 327
    },
    {
      "epoch": 0.4792256414939275,
      "grad_norm": 0.10320023447275162,
      "learning_rate": 1.219178082191781e-05,
      "loss": 1.2418,
      "step": 328
    },
    {
      "epoch": 0.48068669527896996,
      "grad_norm": 0.1106739267706871,
      "learning_rate": 1.2157534246575345e-05,
      "loss": 1.1499,
      "step": 329
    },
    {
      "epoch": 0.4821477490640124,
      "grad_norm": 0.1147918552160263,
      "learning_rate": 1.2123287671232878e-05,
      "loss": 1.1246,
      "step": 330
    },
    {
      "epoch": 0.4836088028490549,
      "grad_norm": 0.10941941291093826,
      "learning_rate": 1.2089041095890412e-05,
      "loss": 1.185,
      "step": 331
    },
    {
      "epoch": 0.48506985663409735,
      "grad_norm": 0.10806426405906677,
      "learning_rate": 1.2054794520547945e-05,
      "loss": 1.2699,
      "step": 332
    },
    {
      "epoch": 0.4865309104191398,
      "grad_norm": 0.09904070943593979,
      "learning_rate": 1.202054794520548e-05,
      "loss": 1.1772,
      "step": 333
    },
    {
      "epoch": 0.4879919642041823,
      "grad_norm": 0.097396120429039,
      "learning_rate": 1.1986301369863013e-05,
      "loss": 1.1753,
      "step": 334
    },
    {
      "epoch": 0.48945301798922475,
      "grad_norm": 0.10030350089073181,
      "learning_rate": 1.195205479452055e-05,
      "loss": 1.22,
      "step": 335
    },
    {
      "epoch": 0.4909140717742672,
      "grad_norm": 0.1039762943983078,
      "learning_rate": 1.1917808219178084e-05,
      "loss": 1.2396,
      "step": 336
    },
    {
      "epoch": 0.4923751255593097,
      "grad_norm": 0.10109396278858185,
      "learning_rate": 1.1883561643835617e-05,
      "loss": 1.185,
      "step": 337
    },
    {
      "epoch": 0.4938361793443521,
      "grad_norm": 0.1067412868142128,
      "learning_rate": 1.1849315068493152e-05,
      "loss": 1.1804,
      "step": 338
    },
    {
      "epoch": 0.49529723312939455,
      "grad_norm": 0.1046527549624443,
      "learning_rate": 1.1815068493150685e-05,
      "loss": 1.1996,
      "step": 339
    },
    {
      "epoch": 0.496758286914437,
      "grad_norm": 0.10973203927278519,
      "learning_rate": 1.178082191780822e-05,
      "loss": 1.1812,
      "step": 340
    },
    {
      "epoch": 0.4982193406994795,
      "grad_norm": 0.11020953208208084,
      "learning_rate": 1.1746575342465753e-05,
      "loss": 1.2603,
      "step": 341
    },
    {
      "epoch": 0.49968039448452195,
      "grad_norm": 0.10667795687913895,
      "learning_rate": 1.171232876712329e-05,
      "loss": 1.2421,
      "step": 342
    },
    {
      "epoch": 0.5011414482695644,
      "grad_norm": 0.0992070883512497,
      "learning_rate": 1.1678082191780822e-05,
      "loss": 1.1995,
      "step": 343
    },
    {
      "epoch": 0.5026025020546069,
      "grad_norm": 0.11278413236141205,
      "learning_rate": 1.1643835616438357e-05,
      "loss": 1.2166,
      "step": 344
    },
    {
      "epoch": 0.5040635558396493,
      "grad_norm": 0.11119436472654343,
      "learning_rate": 1.1609589041095892e-05,
      "loss": 1.2551,
      "step": 345
    },
    {
      "epoch": 0.5055246096246918,
      "grad_norm": 0.11275441944599152,
      "learning_rate": 1.1575342465753425e-05,
      "loss": 1.2082,
      "step": 346
    },
    {
      "epoch": 0.5069856634097343,
      "grad_norm": 0.10059484094381332,
      "learning_rate": 1.154109589041096e-05,
      "loss": 1.172,
      "step": 347
    },
    {
      "epoch": 0.5084467171947767,
      "grad_norm": 0.11156380921602249,
      "learning_rate": 1.1506849315068493e-05,
      "loss": 1.1818,
      "step": 348
    },
    {
      "epoch": 0.5099077709798192,
      "grad_norm": 0.09762994199991226,
      "learning_rate": 1.147260273972603e-05,
      "loss": 1.1674,
      "step": 349
    },
    {
      "epoch": 0.5113688247648617,
      "grad_norm": 0.10659226775169373,
      "learning_rate": 1.1438356164383562e-05,
      "loss": 1.1513,
      "step": 350
    },
    {
      "epoch": 0.5128298785499041,
      "grad_norm": 0.11233004927635193,
      "learning_rate": 1.1404109589041097e-05,
      "loss": 1.1637,
      "step": 351
    },
    {
      "epoch": 0.5142909323349466,
      "grad_norm": 0.10431814193725586,
      "learning_rate": 1.1369863013698632e-05,
      "loss": 1.2148,
      "step": 352
    },
    {
      "epoch": 0.5157519861199891,
      "grad_norm": 0.09940113872289658,
      "learning_rate": 1.1335616438356165e-05,
      "loss": 1.1836,
      "step": 353
    },
    {
      "epoch": 0.5172130399050315,
      "grad_norm": 0.09862768650054932,
      "learning_rate": 1.1301369863013701e-05,
      "loss": 1.2356,
      "step": 354
    },
    {
      "epoch": 0.518674093690074,
      "grad_norm": 0.10437644273042679,
      "learning_rate": 1.1267123287671232e-05,
      "loss": 1.1193,
      "step": 355
    },
    {
      "epoch": 0.5201351474751165,
      "grad_norm": 0.09749601781368256,
      "learning_rate": 1.1232876712328769e-05,
      "loss": 1.2704,
      "step": 356
    },
    {
      "epoch": 0.5215962012601589,
      "grad_norm": 0.1042131707072258,
      "learning_rate": 1.1198630136986302e-05,
      "loss": 1.156,
      "step": 357
    },
    {
      "epoch": 0.5230572550452014,
      "grad_norm": 0.10424741357564926,
      "learning_rate": 1.1164383561643837e-05,
      "loss": 1.2334,
      "step": 358
    },
    {
      "epoch": 0.5245183088302439,
      "grad_norm": 0.09913278371095657,
      "learning_rate": 1.113013698630137e-05,
      "loss": 1.1721,
      "step": 359
    },
    {
      "epoch": 0.5259793626152863,
      "grad_norm": 0.11436072736978531,
      "learning_rate": 1.1095890410958904e-05,
      "loss": 1.1675,
      "step": 360
    },
    {
      "epoch": 0.5274404164003287,
      "grad_norm": 0.1049785315990448,
      "learning_rate": 1.1061643835616441e-05,
      "loss": 1.1788,
      "step": 361
    },
    {
      "epoch": 0.5289014701853711,
      "grad_norm": 0.12005290389060974,
      "learning_rate": 1.1027397260273974e-05,
      "loss": 1.1608,
      "step": 362
    },
    {
      "epoch": 0.5303625239704136,
      "grad_norm": 0.1049429252743721,
      "learning_rate": 1.0993150684931509e-05,
      "loss": 1.2217,
      "step": 363
    },
    {
      "epoch": 0.5318235777554561,
      "grad_norm": 0.10289900004863739,
      "learning_rate": 1.0958904109589042e-05,
      "loss": 1.1854,
      "step": 364
    },
    {
      "epoch": 0.5332846315404985,
      "grad_norm": 0.105230912566185,
      "learning_rate": 1.0924657534246576e-05,
      "loss": 1.2872,
      "step": 365
    },
    {
      "epoch": 0.534745685325541,
      "grad_norm": 0.10501307249069214,
      "learning_rate": 1.089041095890411e-05,
      "loss": 1.2098,
      "step": 366
    },
    {
      "epoch": 0.5362067391105835,
      "grad_norm": 0.11315510421991348,
      "learning_rate": 1.0856164383561644e-05,
      "loss": 1.2473,
      "step": 367
    },
    {
      "epoch": 0.5376677928956259,
      "grad_norm": 0.10925040394067764,
      "learning_rate": 1.082191780821918e-05,
      "loss": 1.2147,
      "step": 368
    },
    {
      "epoch": 0.5391288466806684,
      "grad_norm": 0.11574160307645798,
      "learning_rate": 1.0787671232876714e-05,
      "loss": 1.1954,
      "step": 369
    },
    {
      "epoch": 0.5405899004657109,
      "grad_norm": 0.09681655466556549,
      "learning_rate": 1.0753424657534248e-05,
      "loss": 1.2026,
      "step": 370
    },
    {
      "epoch": 0.5420509542507533,
      "grad_norm": 0.10630439221858978,
      "learning_rate": 1.0719178082191782e-05,
      "loss": 1.1083,
      "step": 371
    },
    {
      "epoch": 0.5435120080357958,
      "grad_norm": 0.1086338609457016,
      "learning_rate": 1.0684931506849316e-05,
      "loss": 1.2337,
      "step": 372
    },
    {
      "epoch": 0.5449730618208383,
      "grad_norm": 0.10121461749076843,
      "learning_rate": 1.065068493150685e-05,
      "loss": 1.2983,
      "step": 373
    },
    {
      "epoch": 0.5464341156058807,
      "grad_norm": 0.10418357700109482,
      "learning_rate": 1.0616438356164384e-05,
      "loss": 1.1552,
      "step": 374
    },
    {
      "epoch": 0.5478951693909232,
      "grad_norm": 0.09971540421247482,
      "learning_rate": 1.0582191780821917e-05,
      "loss": 1.131,
      "step": 375
    },
    {
      "epoch": 0.5493562231759657,
      "grad_norm": 0.09615826606750488,
      "learning_rate": 1.0547945205479453e-05,
      "loss": 1.2724,
      "step": 376
    },
    {
      "epoch": 0.5508172769610081,
      "grad_norm": 0.11235067993402481,
      "learning_rate": 1.0513698630136988e-05,
      "loss": 1.2207,
      "step": 377
    },
    {
      "epoch": 0.5522783307460506,
      "grad_norm": 0.12269837409257889,
      "learning_rate": 1.0479452054794521e-05,
      "loss": 1.1767,
      "step": 378
    },
    {
      "epoch": 0.5537393845310931,
      "grad_norm": 0.11360511928796768,
      "learning_rate": 1.0445205479452056e-05,
      "loss": 1.1903,
      "step": 379
    },
    {
      "epoch": 0.5552004383161355,
      "grad_norm": 0.11293426156044006,
      "learning_rate": 1.0410958904109589e-05,
      "loss": 1.231,
      "step": 380
    },
    {
      "epoch": 0.556661492101178,
      "grad_norm": 0.10496404767036438,
      "learning_rate": 1.0376712328767124e-05,
      "loss": 1.2697,
      "step": 381
    },
    {
      "epoch": 0.5581225458862205,
      "grad_norm": 0.09859599173069,
      "learning_rate": 1.0342465753424657e-05,
      "loss": 1.3125,
      "step": 382
    },
    {
      "epoch": 0.5595835996712629,
      "grad_norm": 0.10170820355415344,
      "learning_rate": 1.0308219178082193e-05,
      "loss": 1.1881,
      "step": 383
    },
    {
      "epoch": 0.5610446534563054,
      "grad_norm": 0.11982686072587967,
      "learning_rate": 1.0273972602739728e-05,
      "loss": 1.1198,
      "step": 384
    },
    {
      "epoch": 0.5625057072413479,
      "grad_norm": 0.10333485156297684,
      "learning_rate": 1.0239726027397261e-05,
      "loss": 1.2105,
      "step": 385
    },
    {
      "epoch": 0.5639667610263903,
      "grad_norm": 0.10605639219284058,
      "learning_rate": 1.0205479452054796e-05,
      "loss": 1.232,
      "step": 386
    },
    {
      "epoch": 0.5654278148114328,
      "grad_norm": 0.10501563549041748,
      "learning_rate": 1.0171232876712329e-05,
      "loss": 1.3193,
      "step": 387
    },
    {
      "epoch": 0.5668888685964752,
      "grad_norm": 0.1139717772603035,
      "learning_rate": 1.0136986301369864e-05,
      "loss": 1.1178,
      "step": 388
    },
    {
      "epoch": 0.5683499223815177,
      "grad_norm": 0.10598957538604736,
      "learning_rate": 1.0102739726027397e-05,
      "loss": 1.2438,
      "step": 389
    },
    {
      "epoch": 0.5698109761665602,
      "grad_norm": 0.10228073596954346,
      "learning_rate": 1.0068493150684933e-05,
      "loss": 1.2064,
      "step": 390
    },
    {
      "epoch": 0.5712720299516026,
      "grad_norm": 0.10198397189378738,
      "learning_rate": 1.0034246575342466e-05,
      "loss": 1.2535,
      "step": 391
    },
    {
      "epoch": 0.5727330837366451,
      "grad_norm": 0.12202975898981094,
      "learning_rate": 1e-05,
      "loss": 1.1619,
      "step": 392
    },
    {
      "epoch": 0.5741941375216875,
      "grad_norm": 0.11935204267501831,
      "learning_rate": 9.965753424657536e-06,
      "loss": 1.2056,
      "step": 393
    },
    {
      "epoch": 0.5756551913067299,
      "grad_norm": 0.10678353160619736,
      "learning_rate": 9.931506849315069e-06,
      "loss": 1.2223,
      "step": 394
    },
    {
      "epoch": 0.5771162450917724,
      "grad_norm": 0.10997404158115387,
      "learning_rate": 9.897260273972603e-06,
      "loss": 1.2401,
      "step": 395
    },
    {
      "epoch": 0.5785772988768149,
      "grad_norm": 0.11465183645486832,
      "learning_rate": 9.863013698630138e-06,
      "loss": 1.2073,
      "step": 396
    },
    {
      "epoch": 0.5800383526618573,
      "grad_norm": 0.13768929243087769,
      "learning_rate": 9.828767123287673e-06,
      "loss": 1.1872,
      "step": 397
    },
    {
      "epoch": 0.5814994064468998,
      "grad_norm": 0.12065139412879944,
      "learning_rate": 9.794520547945206e-06,
      "loss": 1.2145,
      "step": 398
    },
    {
      "epoch": 0.5829604602319423,
      "grad_norm": 0.10538379102945328,
      "learning_rate": 9.76027397260274e-06,
      "loss": 1.217,
      "step": 399
    },
    {
      "epoch": 0.5844215140169847,
      "grad_norm": 0.09868345409631729,
      "learning_rate": 9.726027397260275e-06,
      "loss": 1.255,
      "step": 400
    },
    {
      "epoch": 0.5858825678020272,
      "grad_norm": 0.10661034286022186,
      "learning_rate": 9.691780821917808e-06,
      "loss": 1.1742,
      "step": 401
    },
    {
      "epoch": 0.5873436215870697,
      "grad_norm": 0.11624684184789658,
      "learning_rate": 9.657534246575343e-06,
      "loss": 1.1662,
      "step": 402
    },
    {
      "epoch": 0.5888046753721121,
      "grad_norm": 0.11101629585027695,
      "learning_rate": 9.623287671232878e-06,
      "loss": 1.218,
      "step": 403
    },
    {
      "epoch": 0.5902657291571546,
      "grad_norm": 0.13213178515434265,
      "learning_rate": 9.589041095890411e-06,
      "loss": 1.133,
      "step": 404
    },
    {
      "epoch": 0.591726782942197,
      "grad_norm": 0.11517394334077835,
      "learning_rate": 9.554794520547946e-06,
      "loss": 1.1966,
      "step": 405
    },
    {
      "epoch": 0.5931878367272395,
      "grad_norm": 0.10408038645982742,
      "learning_rate": 9.52054794520548e-06,
      "loss": 1.1859,
      "step": 406
    },
    {
      "epoch": 0.594648890512282,
      "grad_norm": 0.1159515306353569,
      "learning_rate": 9.486301369863015e-06,
      "loss": 1.2286,
      "step": 407
    },
    {
      "epoch": 0.5961099442973244,
      "grad_norm": 0.11420222371816635,
      "learning_rate": 9.452054794520548e-06,
      "loss": 1.0952,
      "step": 408
    },
    {
      "epoch": 0.5975709980823669,
      "grad_norm": 0.11413077265024185,
      "learning_rate": 9.417808219178083e-06,
      "loss": 1.2159,
      "step": 409
    },
    {
      "epoch": 0.5990320518674094,
      "grad_norm": 0.12136485427618027,
      "learning_rate": 9.383561643835618e-06,
      "loss": 1.2166,
      "step": 410
    },
    {
      "epoch": 0.6004931056524518,
      "grad_norm": 0.12264648824930191,
      "learning_rate": 9.34931506849315e-06,
      "loss": 1.1057,
      "step": 411
    },
    {
      "epoch": 0.6019541594374943,
      "grad_norm": 0.10724509507417679,
      "learning_rate": 9.315068493150685e-06,
      "loss": 1.1633,
      "step": 412
    },
    {
      "epoch": 0.6034152132225368,
      "grad_norm": 0.11786479502916336,
      "learning_rate": 9.28082191780822e-06,
      "loss": 1.1942,
      "step": 413
    },
    {
      "epoch": 0.6048762670075792,
      "grad_norm": 0.10697019845247269,
      "learning_rate": 9.246575342465755e-06,
      "loss": 1.265,
      "step": 414
    },
    {
      "epoch": 0.6063373207926217,
      "grad_norm": 0.11874634027481079,
      "learning_rate": 9.212328767123288e-06,
      "loss": 1.2373,
      "step": 415
    },
    {
      "epoch": 0.6077983745776642,
      "grad_norm": 0.11465580761432648,
      "learning_rate": 9.178082191780823e-06,
      "loss": 1.1977,
      "step": 416
    },
    {
      "epoch": 0.6092594283627066,
      "grad_norm": 0.11860576272010803,
      "learning_rate": 9.143835616438357e-06,
      "loss": 1.1474,
      "step": 417
    },
    {
      "epoch": 0.6107204821477491,
      "grad_norm": 0.11340127140283585,
      "learning_rate": 9.10958904109589e-06,
      "loss": 1.219,
      "step": 418
    },
    {
      "epoch": 0.6121815359327916,
      "grad_norm": 0.1260974407196045,
      "learning_rate": 9.075342465753425e-06,
      "loss": 1.1837,
      "step": 419
    },
    {
      "epoch": 0.613642589717834,
      "grad_norm": 0.1299670934677124,
      "learning_rate": 9.04109589041096e-06,
      "loss": 1.0615,
      "step": 420
    },
    {
      "epoch": 0.6151036435028765,
      "grad_norm": 0.10845065861940384,
      "learning_rate": 9.006849315068495e-06,
      "loss": 1.2173,
      "step": 421
    },
    {
      "epoch": 0.616564697287919,
      "grad_norm": 0.10730204731225967,
      "learning_rate": 8.972602739726028e-06,
      "loss": 1.1314,
      "step": 422
    },
    {
      "epoch": 0.6180257510729614,
      "grad_norm": 0.10890056192874908,
      "learning_rate": 8.938356164383562e-06,
      "loss": 1.2128,
      "step": 423
    },
    {
      "epoch": 0.6194868048580038,
      "grad_norm": 0.11053816974163055,
      "learning_rate": 8.904109589041097e-06,
      "loss": 1.2554,
      "step": 424
    },
    {
      "epoch": 0.6209478586430462,
      "grad_norm": 0.1105181872844696,
      "learning_rate": 8.86986301369863e-06,
      "loss": 1.1943,
      "step": 425
    },
    {
      "epoch": 0.6224089124280887,
      "grad_norm": 0.12110709398984909,
      "learning_rate": 8.835616438356165e-06,
      "loss": 1.2072,
      "step": 426
    },
    {
      "epoch": 0.6238699662131312,
      "grad_norm": 0.12286946922540665,
      "learning_rate": 8.8013698630137e-06,
      "loss": 1.2465,
      "step": 427
    },
    {
      "epoch": 0.6253310199981736,
      "grad_norm": 0.132927805185318,
      "learning_rate": 8.767123287671233e-06,
      "loss": 1.1644,
      "step": 428
    },
    {
      "epoch": 0.6267920737832161,
      "grad_norm": 0.11811359971761703,
      "learning_rate": 8.732876712328769e-06,
      "loss": 1.1466,
      "step": 429
    },
    {
      "epoch": 0.6282531275682586,
      "grad_norm": 0.11676699668169022,
      "learning_rate": 8.698630136986302e-06,
      "loss": 1.1889,
      "step": 430
    },
    {
      "epoch": 0.629714181353301,
      "grad_norm": 0.10928516089916229,
      "learning_rate": 8.664383561643837e-06,
      "loss": 1.1603,
      "step": 431
    },
    {
      "epoch": 0.6311752351383435,
      "grad_norm": 0.13088025152683258,
      "learning_rate": 8.63013698630137e-06,
      "loss": 1.0952,
      "step": 432
    },
    {
      "epoch": 0.632636288923386,
      "grad_norm": 0.11683713644742966,
      "learning_rate": 8.595890410958905e-06,
      "loss": 1.2762,
      "step": 433
    },
    {
      "epoch": 0.6340973427084284,
      "grad_norm": 0.10292809456586838,
      "learning_rate": 8.56164383561644e-06,
      "loss": 1.1743,
      "step": 434
    },
    {
      "epoch": 0.6355583964934709,
      "grad_norm": 0.1085817888379097,
      "learning_rate": 8.527397260273972e-06,
      "loss": 1.2021,
      "step": 435
    },
    {
      "epoch": 0.6370194502785134,
      "grad_norm": 0.10217051953077316,
      "learning_rate": 8.493150684931507e-06,
      "loss": 1.1815,
      "step": 436
    },
    {
      "epoch": 0.6384805040635558,
      "grad_norm": 0.11223044246435165,
      "learning_rate": 8.458904109589042e-06,
      "loss": 1.1238,
      "step": 437
    },
    {
      "epoch": 0.6399415578485983,
      "grad_norm": 0.10959354788064957,
      "learning_rate": 8.424657534246577e-06,
      "loss": 1.116,
      "step": 438
    },
    {
      "epoch": 0.6414026116336408,
      "grad_norm": 0.12990103662014008,
      "learning_rate": 8.39041095890411e-06,
      "loss": 1.1134,
      "step": 439
    },
    {
      "epoch": 0.6428636654186832,
      "grad_norm": 0.11417476832866669,
      "learning_rate": 8.356164383561644e-06,
      "loss": 1.2019,
      "step": 440
    },
    {
      "epoch": 0.6443247192037257,
      "grad_norm": 0.10849736630916595,
      "learning_rate": 8.32191780821918e-06,
      "loss": 1.1575,
      "step": 441
    },
    {
      "epoch": 0.6457857729887682,
      "grad_norm": 0.12259836494922638,
      "learning_rate": 8.287671232876712e-06,
      "loss": 1.1847,
      "step": 442
    },
    {
      "epoch": 0.6472468267738106,
      "grad_norm": 0.11938966810703278,
      "learning_rate": 8.253424657534247e-06,
      "loss": 1.2109,
      "step": 443
    },
    {
      "epoch": 0.6487078805588531,
      "grad_norm": 0.11072079837322235,
      "learning_rate": 8.219178082191782e-06,
      "loss": 1.1742,
      "step": 444
    },
    {
      "epoch": 0.6501689343438956,
      "grad_norm": 0.10626699030399323,
      "learning_rate": 8.184931506849316e-06,
      "loss": 1.1866,
      "step": 445
    },
    {
      "epoch": 0.651629988128938,
      "grad_norm": 0.109890878200531,
      "learning_rate": 8.150684931506851e-06,
      "loss": 1.1477,
      "step": 446
    },
    {
      "epoch": 0.6530910419139805,
      "grad_norm": 0.11042490601539612,
      "learning_rate": 8.116438356164384e-06,
      "loss": 1.2544,
      "step": 447
    },
    {
      "epoch": 0.654552095699023,
      "grad_norm": 0.11169801652431488,
      "learning_rate": 8.082191780821919e-06,
      "loss": 1.1274,
      "step": 448
    },
    {
      "epoch": 0.6560131494840654,
      "grad_norm": 0.10873094201087952,
      "learning_rate": 8.047945205479452e-06,
      "loss": 1.1965,
      "step": 449
    },
    {
      "epoch": 0.6574742032691079,
      "grad_norm": 0.11143123358488083,
      "learning_rate": 8.013698630136987e-06,
      "loss": 1.1708,
      "step": 450
    },
    {
      "epoch": 0.6589352570541503,
      "grad_norm": 0.12092313915491104,
      "learning_rate": 7.979452054794521e-06,
      "loss": 1.2115,
      "step": 451
    },
    {
      "epoch": 0.6603963108391928,
      "grad_norm": 0.1247633770108223,
      "learning_rate": 7.945205479452055e-06,
      "loss": 1.1683,
      "step": 452
    },
    {
      "epoch": 0.6618573646242353,
      "grad_norm": 0.11757193505764008,
      "learning_rate": 7.910958904109591e-06,
      "loss": 1.187,
      "step": 453
    },
    {
      "epoch": 0.6633184184092777,
      "grad_norm": 0.10670476406812668,
      "learning_rate": 7.876712328767124e-06,
      "loss": 1.0998,
      "step": 454
    },
    {
      "epoch": 0.6647794721943201,
      "grad_norm": 0.11120694130659103,
      "learning_rate": 7.842465753424659e-06,
      "loss": 1.1952,
      "step": 455
    },
    {
      "epoch": 0.6662405259793626,
      "grad_norm": 0.10676517337560654,
      "learning_rate": 7.808219178082192e-06,
      "loss": 1.1609,
      "step": 456
    },
    {
      "epoch": 0.667701579764405,
      "grad_norm": 0.10845296084880829,
      "learning_rate": 7.773972602739727e-06,
      "loss": 1.1445,
      "step": 457
    },
    {
      "epoch": 0.6691626335494475,
      "grad_norm": 0.1130744218826294,
      "learning_rate": 7.739726027397261e-06,
      "loss": 1.2327,
      "step": 458
    },
    {
      "epoch": 0.67062368733449,
      "grad_norm": 0.12214113771915436,
      "learning_rate": 7.705479452054794e-06,
      "loss": 1.2415,
      "step": 459
    },
    {
      "epoch": 0.6720847411195324,
      "grad_norm": 0.10830514878034592,
      "learning_rate": 7.671232876712329e-06,
      "loss": 1.2456,
      "step": 460
    },
    {
      "epoch": 0.6735457949045749,
      "grad_norm": 0.11725237220525742,
      "learning_rate": 7.636986301369864e-06,
      "loss": 1.1838,
      "step": 461
    },
    {
      "epoch": 0.6750068486896174,
      "grad_norm": 0.12461910396814346,
      "learning_rate": 7.6027397260273985e-06,
      "loss": 1.1989,
      "step": 462
    },
    {
      "epoch": 0.6764679024746598,
      "grad_norm": 0.11189593374729156,
      "learning_rate": 7.568493150684932e-06,
      "loss": 1.1218,
      "step": 463
    },
    {
      "epoch": 0.6779289562597023,
      "grad_norm": 0.1076999306678772,
      "learning_rate": 7.534246575342466e-06,
      "loss": 1.125,
      "step": 464
    },
    {
      "epoch": 0.6793900100447448,
      "grad_norm": 0.13751359283924103,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.1334,
      "step": 465
    },
    {
      "epoch": 0.6808510638297872,
      "grad_norm": 0.11828191578388214,
      "learning_rate": 7.465753424657535e-06,
      "loss": 1.2438,
      "step": 466
    },
    {
      "epoch": 0.6823121176148297,
      "grad_norm": 0.11072523146867752,
      "learning_rate": 7.431506849315069e-06,
      "loss": 1.2299,
      "step": 467
    },
    {
      "epoch": 0.6837731713998721,
      "grad_norm": 0.1260717362165451,
      "learning_rate": 7.397260273972603e-06,
      "loss": 1.2204,
      "step": 468
    },
    {
      "epoch": 0.6852342251849146,
      "grad_norm": 0.11779427528381348,
      "learning_rate": 7.3630136986301374e-06,
      "loss": 1.1887,
      "step": 469
    },
    {
      "epoch": 0.6866952789699571,
      "grad_norm": 0.11070991307497025,
      "learning_rate": 7.328767123287672e-06,
      "loss": 1.2137,
      "step": 470
    },
    {
      "epoch": 0.6881563327549995,
      "grad_norm": 0.11925278604030609,
      "learning_rate": 7.294520547945206e-06,
      "loss": 1.1415,
      "step": 471
    },
    {
      "epoch": 0.689617386540042,
      "grad_norm": 0.11368401348590851,
      "learning_rate": 7.260273972602741e-06,
      "loss": 1.2588,
      "step": 472
    },
    {
      "epoch": 0.6910784403250845,
      "grad_norm": 0.11111228913068771,
      "learning_rate": 7.226027397260275e-06,
      "loss": 1.148,
      "step": 473
    },
    {
      "epoch": 0.6925394941101269,
      "grad_norm": 0.12571550905704498,
      "learning_rate": 7.191780821917809e-06,
      "loss": 1.167,
      "step": 474
    },
    {
      "epoch": 0.6940005478951694,
      "grad_norm": 0.11622565984725952,
      "learning_rate": 7.1575342465753425e-06,
      "loss": 1.2028,
      "step": 475
    },
    {
      "epoch": 0.6954616016802119,
      "grad_norm": 0.12074082344770432,
      "learning_rate": 7.123287671232877e-06,
      "loss": 1.3148,
      "step": 476
    },
    {
      "epoch": 0.6969226554652543,
      "grad_norm": 0.11308849602937698,
      "learning_rate": 7.089041095890411e-06,
      "loss": 1.2337,
      "step": 477
    },
    {
      "epoch": 0.6983837092502968,
      "grad_norm": 0.11705081909894943,
      "learning_rate": 7.054794520547946e-06,
      "loss": 1.1486,
      "step": 478
    },
    {
      "epoch": 0.6998447630353393,
      "grad_norm": 0.11467244476079941,
      "learning_rate": 7.020547945205481e-06,
      "loss": 1.3139,
      "step": 479
    },
    {
      "epoch": 0.7013058168203817,
      "grad_norm": 0.11780110001564026,
      "learning_rate": 6.9863013698630145e-06,
      "loss": 1.2417,
      "step": 480
    },
    {
      "epoch": 0.7027668706054242,
      "grad_norm": 0.11144915223121643,
      "learning_rate": 6.952054794520548e-06,
      "loss": 1.1892,
      "step": 481
    },
    {
      "epoch": 0.7042279243904667,
      "grad_norm": 0.13650654256343842,
      "learning_rate": 6.917808219178082e-06,
      "loss": 1.0756,
      "step": 482
    },
    {
      "epoch": 0.7056889781755091,
      "grad_norm": 0.11797624081373215,
      "learning_rate": 6.883561643835617e-06,
      "loss": 1.275,
      "step": 483
    },
    {
      "epoch": 0.7071500319605516,
      "grad_norm": 0.1288430392742157,
      "learning_rate": 6.849315068493151e-06,
      "loss": 1.1719,
      "step": 484
    },
    {
      "epoch": 0.7086110857455941,
      "grad_norm": 0.11288215219974518,
      "learning_rate": 6.815068493150685e-06,
      "loss": 1.1841,
      "step": 485
    },
    {
      "epoch": 0.7100721395306365,
      "grad_norm": 0.11503782123327255,
      "learning_rate": 6.78082191780822e-06,
      "loss": 1.1977,
      "step": 486
    },
    {
      "epoch": 0.7115331933156789,
      "grad_norm": 0.11059726774692535,
      "learning_rate": 6.746575342465754e-06,
      "loss": 1.1771,
      "step": 487
    },
    {
      "epoch": 0.7129942471007213,
      "grad_norm": 0.12105996906757355,
      "learning_rate": 6.712328767123288e-06,
      "loss": 1.218,
      "step": 488
    },
    {
      "epoch": 0.7144553008857638,
      "grad_norm": 0.1285051554441452,
      "learning_rate": 6.678082191780823e-06,
      "loss": 1.1423,
      "step": 489
    },
    {
      "epoch": 0.7159163546708063,
      "grad_norm": 0.11841081827878952,
      "learning_rate": 6.643835616438357e-06,
      "loss": 1.1433,
      "step": 490
    },
    {
      "epoch": 0.7173774084558487,
      "grad_norm": 0.11378856003284454,
      "learning_rate": 6.609589041095891e-06,
      "loss": 1.1941,
      "step": 491
    },
    {
      "epoch": 0.7188384622408912,
      "grad_norm": 0.10600030422210693,
      "learning_rate": 6.5753424657534245e-06,
      "loss": 1.188,
      "step": 492
    },
    {
      "epoch": 0.7202995160259337,
      "grad_norm": 0.10563358664512634,
      "learning_rate": 6.541095890410959e-06,
      "loss": 1.2165,
      "step": 493
    },
    {
      "epoch": 0.7217605698109761,
      "grad_norm": 0.11092449724674225,
      "learning_rate": 6.506849315068494e-06,
      "loss": 1.1956,
      "step": 494
    },
    {
      "epoch": 0.7232216235960186,
      "grad_norm": 0.11748078465461731,
      "learning_rate": 6.472602739726028e-06,
      "loss": 1.1741,
      "step": 495
    },
    {
      "epoch": 0.7246826773810611,
      "grad_norm": 0.12683749198913574,
      "learning_rate": 6.438356164383563e-06,
      "loss": 1.166,
      "step": 496
    },
    {
      "epoch": 0.7261437311661035,
      "grad_norm": 0.10706394910812378,
      "learning_rate": 6.4041095890410965e-06,
      "loss": 1.2278,
      "step": 497
    },
    {
      "epoch": 0.727604784951146,
      "grad_norm": 0.10801483690738678,
      "learning_rate": 6.36986301369863e-06,
      "loss": 1.1693,
      "step": 498
    },
    {
      "epoch": 0.7290658387361885,
      "grad_norm": 0.12337271869182587,
      "learning_rate": 6.335616438356164e-06,
      "loss": 1.229,
      "step": 499
    },
    {
      "epoch": 0.7305268925212309,
      "grad_norm": 0.13341547548770905,
      "learning_rate": 6.301369863013699e-06,
      "loss": 1.1274,
      "step": 500
    },
    {
      "epoch": 0.7319879463062734,
      "grad_norm": 0.11283931136131287,
      "learning_rate": 6.267123287671233e-06,
      "loss": 1.1867,
      "step": 501
    },
    {
      "epoch": 0.7334490000913159,
      "grad_norm": 0.1254453957080841,
      "learning_rate": 6.2328767123287685e-06,
      "loss": 1.2634,
      "step": 502
    },
    {
      "epoch": 0.7349100538763583,
      "grad_norm": 0.125976100564003,
      "learning_rate": 6.198630136986302e-06,
      "loss": 1.0729,
      "step": 503
    },
    {
      "epoch": 0.7363711076614008,
      "grad_norm": 0.10732964426279068,
      "learning_rate": 6.164383561643836e-06,
      "loss": 1.2072,
      "step": 504
    },
    {
      "epoch": 0.7378321614464433,
      "grad_norm": 0.12129033356904984,
      "learning_rate": 6.13013698630137e-06,
      "loss": 1.1197,
      "step": 505
    },
    {
      "epoch": 0.7392932152314857,
      "grad_norm": 0.12821297347545624,
      "learning_rate": 6.095890410958905e-06,
      "loss": 1.1515,
      "step": 506
    },
    {
      "epoch": 0.7407542690165282,
      "grad_norm": 0.11655119061470032,
      "learning_rate": 6.061643835616439e-06,
      "loss": 1.172,
      "step": 507
    },
    {
      "epoch": 0.7422153228015707,
      "grad_norm": 0.11455903202295303,
      "learning_rate": 6.027397260273973e-06,
      "loss": 1.1762,
      "step": 508
    },
    {
      "epoch": 0.7436763765866131,
      "grad_norm": 0.10701651871204376,
      "learning_rate": 5.993150684931507e-06,
      "loss": 1.1685,
      "step": 509
    },
    {
      "epoch": 0.7451374303716556,
      "grad_norm": 0.11655491590499878,
      "learning_rate": 5.958904109589042e-06,
      "loss": 1.1975,
      "step": 510
    },
    {
      "epoch": 0.746598484156698,
      "grad_norm": 0.11159254610538483,
      "learning_rate": 5.924657534246576e-06,
      "loss": 1.2226,
      "step": 511
    },
    {
      "epoch": 0.7480595379417405,
      "grad_norm": 0.11702670156955719,
      "learning_rate": 5.89041095890411e-06,
      "loss": 1.1552,
      "step": 512
    },
    {
      "epoch": 0.749520591726783,
      "grad_norm": 0.11294779181480408,
      "learning_rate": 5.856164383561645e-06,
      "loss": 1.189,
      "step": 513
    },
    {
      "epoch": 0.7509816455118254,
      "grad_norm": 0.10862728208303452,
      "learning_rate": 5.821917808219179e-06,
      "loss": 1.1997,
      "step": 514
    },
    {
      "epoch": 0.7524426992968679,
      "grad_norm": 0.12142271548509598,
      "learning_rate": 5.7876712328767125e-06,
      "loss": 1.1985,
      "step": 515
    },
    {
      "epoch": 0.7539037530819104,
      "grad_norm": 0.11388342082500458,
      "learning_rate": 5.753424657534246e-06,
      "loss": 1.1195,
      "step": 516
    },
    {
      "epoch": 0.7553648068669528,
      "grad_norm": 0.11917892098426819,
      "learning_rate": 5.719178082191781e-06,
      "loss": 1.1685,
      "step": 517
    },
    {
      "epoch": 0.7568258606519952,
      "grad_norm": 0.11466323584318161,
      "learning_rate": 5.684931506849316e-06,
      "loss": 1.1837,
      "step": 518
    },
    {
      "epoch": 0.7582869144370377,
      "grad_norm": 0.11477669328451157,
      "learning_rate": 5.6506849315068506e-06,
      "loss": 1.2336,
      "step": 519
    },
    {
      "epoch": 0.7597479682220801,
      "grad_norm": 0.12040074169635773,
      "learning_rate": 5.6164383561643845e-06,
      "loss": 1.2553,
      "step": 520
    },
    {
      "epoch": 0.7612090220071226,
      "grad_norm": 0.11461540311574936,
      "learning_rate": 5.582191780821918e-06,
      "loss": 1.1205,
      "step": 521
    },
    {
      "epoch": 0.7626700757921651,
      "grad_norm": 0.10867593437433243,
      "learning_rate": 5.547945205479452e-06,
      "loss": 1.1629,
      "step": 522
    },
    {
      "epoch": 0.7641311295772075,
      "grad_norm": 0.12366941571235657,
      "learning_rate": 5.513698630136987e-06,
      "loss": 1.2059,
      "step": 523
    },
    {
      "epoch": 0.76559218336225,
      "grad_norm": 0.11807534843683243,
      "learning_rate": 5.479452054794521e-06,
      "loss": 1.2863,
      "step": 524
    },
    {
      "epoch": 0.7670532371472925,
      "grad_norm": 0.11750409007072449,
      "learning_rate": 5.445205479452055e-06,
      "loss": 1.2293,
      "step": 525
    },
    {
      "epoch": 0.7685142909323349,
      "grad_norm": 0.11874507367610931,
      "learning_rate": 5.41095890410959e-06,
      "loss": 1.2275,
      "step": 526
    },
    {
      "epoch": 0.7699753447173774,
      "grad_norm": 0.12096529453992844,
      "learning_rate": 5.376712328767124e-06,
      "loss": 1.1525,
      "step": 527
    },
    {
      "epoch": 0.7714363985024199,
      "grad_norm": 0.10740137845277786,
      "learning_rate": 5.342465753424658e-06,
      "loss": 1.2267,
      "step": 528
    },
    {
      "epoch": 0.7728974522874623,
      "grad_norm": 0.12556499242782593,
      "learning_rate": 5.308219178082192e-06,
      "loss": 1.2042,
      "step": 529
    },
    {
      "epoch": 0.7743585060725048,
      "grad_norm": 0.1109650582075119,
      "learning_rate": 5.273972602739727e-06,
      "loss": 1.1813,
      "step": 530
    },
    {
      "epoch": 0.7758195598575472,
      "grad_norm": 0.10629246383905411,
      "learning_rate": 5.239726027397261e-06,
      "loss": 1.1959,
      "step": 531
    },
    {
      "epoch": 0.7772806136425897,
      "grad_norm": 0.1113128662109375,
      "learning_rate": 5.2054794520547945e-06,
      "loss": 1.1441,
      "step": 532
    },
    {
      "epoch": 0.7787416674276322,
      "grad_norm": 0.11163881421089172,
      "learning_rate": 5.171232876712328e-06,
      "loss": 1.1991,
      "step": 533
    },
    {
      "epoch": 0.7802027212126746,
      "grad_norm": 0.12462608516216278,
      "learning_rate": 5.136986301369864e-06,
      "loss": 1.1397,
      "step": 534
    },
    {
      "epoch": 0.7816637749977171,
      "grad_norm": 0.11631737649440765,
      "learning_rate": 5.102739726027398e-06,
      "loss": 1.1253,
      "step": 535
    },
    {
      "epoch": 0.7831248287827596,
      "grad_norm": 0.12405448406934738,
      "learning_rate": 5.068493150684932e-06,
      "loss": 1.1595,
      "step": 536
    },
    {
      "epoch": 0.784585882567802,
      "grad_norm": 0.1106800064444542,
      "learning_rate": 5.0342465753424665e-06,
      "loss": 1.2755,
      "step": 537
    },
    {
      "epoch": 0.7860469363528445,
      "grad_norm": 0.10687270015478134,
      "learning_rate": 5e-06,
      "loss": 1.1443,
      "step": 538
    },
    {
      "epoch": 0.787507990137887,
      "grad_norm": 0.10897688567638397,
      "learning_rate": 4.965753424657534e-06,
      "loss": 1.1917,
      "step": 539
    },
    {
      "epoch": 0.7889690439229294,
      "grad_norm": 0.12325593084096909,
      "learning_rate": 4.931506849315069e-06,
      "loss": 1.165,
      "step": 540
    },
    {
      "epoch": 0.7904300977079719,
      "grad_norm": 0.12180227786302567,
      "learning_rate": 4.897260273972603e-06,
      "loss": 1.15,
      "step": 541
    },
    {
      "epoch": 0.7918911514930144,
      "grad_norm": 0.1084585040807724,
      "learning_rate": 4.863013698630138e-06,
      "loss": 1.2431,
      "step": 542
    },
    {
      "epoch": 0.7933522052780568,
      "grad_norm": 0.11849282681941986,
      "learning_rate": 4.8287671232876716e-06,
      "loss": 1.2494,
      "step": 543
    },
    {
      "epoch": 0.7948132590630993,
      "grad_norm": 0.1109924465417862,
      "learning_rate": 4.7945205479452054e-06,
      "loss": 1.1517,
      "step": 544
    },
    {
      "epoch": 0.7962743128481418,
      "grad_norm": 0.1313486099243164,
      "learning_rate": 4.76027397260274e-06,
      "loss": 1.2044,
      "step": 545
    },
    {
      "epoch": 0.7977353666331842,
      "grad_norm": 0.12946535646915436,
      "learning_rate": 4.726027397260274e-06,
      "loss": 1.101,
      "step": 546
    },
    {
      "epoch": 0.7991964204182267,
      "grad_norm": 0.1083068773150444,
      "learning_rate": 4.691780821917809e-06,
      "loss": 1.1933,
      "step": 547
    },
    {
      "epoch": 0.8006574742032692,
      "grad_norm": 0.11590442061424255,
      "learning_rate": 4.657534246575343e-06,
      "loss": 1.198,
      "step": 548
    },
    {
      "epoch": 0.8021185279883116,
      "grad_norm": 0.11123711615800858,
      "learning_rate": 4.6232876712328774e-06,
      "loss": 1.1259,
      "step": 549
    },
    {
      "epoch": 0.803579581773354,
      "grad_norm": 0.1170891597867012,
      "learning_rate": 4.589041095890411e-06,
      "loss": 1.1977,
      "step": 550
    },
    {
      "epoch": 0.8050406355583964,
      "grad_norm": 0.12325557321310043,
      "learning_rate": 4.554794520547945e-06,
      "loss": 1.1549,
      "step": 551
    },
    {
      "epoch": 0.8065016893434389,
      "grad_norm": 0.10925264656543732,
      "learning_rate": 4.52054794520548e-06,
      "loss": 1.1376,
      "step": 552
    },
    {
      "epoch": 0.8079627431284814,
      "grad_norm": 0.10057859122753143,
      "learning_rate": 4.486301369863014e-06,
      "loss": 1.1237,
      "step": 553
    },
    {
      "epoch": 0.8094237969135238,
      "grad_norm": 0.1212565079331398,
      "learning_rate": 4.4520547945205486e-06,
      "loss": 1.1878,
      "step": 554
    },
    {
      "epoch": 0.8108848506985663,
      "grad_norm": 0.10832269489765167,
      "learning_rate": 4.4178082191780825e-06,
      "loss": 1.1239,
      "step": 555
    },
    {
      "epoch": 0.8123459044836088,
      "grad_norm": 0.11844736337661743,
      "learning_rate": 4.383561643835616e-06,
      "loss": 1.197,
      "step": 556
    },
    {
      "epoch": 0.8138069582686512,
      "grad_norm": 0.11942517757415771,
      "learning_rate": 4.349315068493151e-06,
      "loss": 1.1756,
      "step": 557
    },
    {
      "epoch": 0.8152680120536937,
      "grad_norm": 0.11740544438362122,
      "learning_rate": 4.315068493150685e-06,
      "loss": 1.1753,
      "step": 558
    },
    {
      "epoch": 0.8167290658387362,
      "grad_norm": 0.11186040937900543,
      "learning_rate": 4.28082191780822e-06,
      "loss": 1.2328,
      "step": 559
    },
    {
      "epoch": 0.8181901196237786,
      "grad_norm": 0.11647879332304001,
      "learning_rate": 4.246575342465754e-06,
      "loss": 1.1962,
      "step": 560
    },
    {
      "epoch": 0.8196511734088211,
      "grad_norm": 0.12523581087589264,
      "learning_rate": 4.212328767123288e-06,
      "loss": 1.2818,
      "step": 561
    },
    {
      "epoch": 0.8211122271938636,
      "grad_norm": 0.11687058955430984,
      "learning_rate": 4.178082191780822e-06,
      "loss": 1.173,
      "step": 562
    },
    {
      "epoch": 0.822573280978906,
      "grad_norm": 0.1250499188899994,
      "learning_rate": 4.143835616438356e-06,
      "loss": 1.1234,
      "step": 563
    },
    {
      "epoch": 0.8240343347639485,
      "grad_norm": 0.11505109816789627,
      "learning_rate": 4.109589041095891e-06,
      "loss": 1.2452,
      "step": 564
    },
    {
      "epoch": 0.825495388548991,
      "grad_norm": 0.1190369576215744,
      "learning_rate": 4.075342465753426e-06,
      "loss": 1.149,
      "step": 565
    },
    {
      "epoch": 0.8269564423340334,
      "grad_norm": 0.12453046441078186,
      "learning_rate": 4.0410958904109595e-06,
      "loss": 1.2214,
      "step": 566
    },
    {
      "epoch": 0.8284174961190759,
      "grad_norm": 0.10606851428747177,
      "learning_rate": 4.006849315068493e-06,
      "loss": 1.1749,
      "step": 567
    },
    {
      "epoch": 0.8298785499041184,
      "grad_norm": 0.11873757094144821,
      "learning_rate": 3.972602739726027e-06,
      "loss": 1.2741,
      "step": 568
    },
    {
      "epoch": 0.8313396036891608,
      "grad_norm": 0.12206880748271942,
      "learning_rate": 3.938356164383562e-06,
      "loss": 1.2122,
      "step": 569
    },
    {
      "epoch": 0.8328006574742033,
      "grad_norm": 0.10448160022497177,
      "learning_rate": 3.904109589041096e-06,
      "loss": 1.2251,
      "step": 570
    },
    {
      "epoch": 0.8342617112592458,
      "grad_norm": 0.10980773717164993,
      "learning_rate": 3.869863013698631e-06,
      "loss": 1.1791,
      "step": 571
    },
    {
      "epoch": 0.8357227650442882,
      "grad_norm": 0.11824549734592438,
      "learning_rate": 3.8356164383561645e-06,
      "loss": 1.2319,
      "step": 572
    },
    {
      "epoch": 0.8371838188293307,
      "grad_norm": 0.1143551915884018,
      "learning_rate": 3.8013698630136993e-06,
      "loss": 1.16,
      "step": 573
    },
    {
      "epoch": 0.8386448726143731,
      "grad_norm": 0.11645519733428955,
      "learning_rate": 3.767123287671233e-06,
      "loss": 1.1278,
      "step": 574
    },
    {
      "epoch": 0.8401059263994156,
      "grad_norm": 0.12313269078731537,
      "learning_rate": 3.7328767123287675e-06,
      "loss": 1.1707,
      "step": 575
    },
    {
      "epoch": 0.8415669801844581,
      "grad_norm": 0.1056065782904625,
      "learning_rate": 3.6986301369863014e-06,
      "loss": 1.0892,
      "step": 576
    },
    {
      "epoch": 0.8430280339695005,
      "grad_norm": 0.10892536491155624,
      "learning_rate": 3.664383561643836e-06,
      "loss": 1.2527,
      "step": 577
    },
    {
      "epoch": 0.844489087754543,
      "grad_norm": 0.10961074382066727,
      "learning_rate": 3.6301369863013704e-06,
      "loss": 1.2338,
      "step": 578
    },
    {
      "epoch": 0.8459501415395855,
      "grad_norm": 0.1143115982413292,
      "learning_rate": 3.5958904109589043e-06,
      "loss": 1.2256,
      "step": 579
    },
    {
      "epoch": 0.8474111953246279,
      "grad_norm": 0.1371782124042511,
      "learning_rate": 3.5616438356164386e-06,
      "loss": 1.1487,
      "step": 580
    },
    {
      "epoch": 0.8488722491096703,
      "grad_norm": 0.12069284915924072,
      "learning_rate": 3.527397260273973e-06,
      "loss": 1.1901,
      "step": 581
    },
    {
      "epoch": 0.8503333028947128,
      "grad_norm": 0.11841464787721634,
      "learning_rate": 3.4931506849315072e-06,
      "loss": 1.1714,
      "step": 582
    },
    {
      "epoch": 0.8517943566797552,
      "grad_norm": 0.12204127013683319,
      "learning_rate": 3.458904109589041e-06,
      "loss": 1.1501,
      "step": 583
    },
    {
      "epoch": 0.8532554104647977,
      "grad_norm": 0.13133108615875244,
      "learning_rate": 3.4246575342465754e-06,
      "loss": 1.2586,
      "step": 584
    },
    {
      "epoch": 0.8547164642498402,
      "grad_norm": 0.11903316527605057,
      "learning_rate": 3.39041095890411e-06,
      "loss": 1.2399,
      "step": 585
    },
    {
      "epoch": 0.8561775180348826,
      "grad_norm": 0.1073235422372818,
      "learning_rate": 3.356164383561644e-06,
      "loss": 1.2036,
      "step": 586
    },
    {
      "epoch": 0.8576385718199251,
      "grad_norm": 0.1356821060180664,
      "learning_rate": 3.3219178082191784e-06,
      "loss": 1.1168,
      "step": 587
    },
    {
      "epoch": 0.8590996256049676,
      "grad_norm": 0.1314031332731247,
      "learning_rate": 3.2876712328767123e-06,
      "loss": 1.1407,
      "step": 588
    },
    {
      "epoch": 0.86056067939001,
      "grad_norm": 0.1098315566778183,
      "learning_rate": 3.253424657534247e-06,
      "loss": 1.1737,
      "step": 589
    },
    {
      "epoch": 0.8620217331750525,
      "grad_norm": 0.12484846264123917,
      "learning_rate": 3.2191780821917813e-06,
      "loss": 1.29,
      "step": 590
    },
    {
      "epoch": 0.863482786960095,
      "grad_norm": 0.12023176997900009,
      "learning_rate": 3.184931506849315e-06,
      "loss": 1.2455,
      "step": 591
    },
    {
      "epoch": 0.8649438407451374,
      "grad_norm": 0.1124570220708847,
      "learning_rate": 3.1506849315068495e-06,
      "loss": 1.1792,
      "step": 592
    },
    {
      "epoch": 0.8664048945301799,
      "grad_norm": 0.11319974809885025,
      "learning_rate": 3.1164383561643843e-06,
      "loss": 1.1708,
      "step": 593
    },
    {
      "epoch": 0.8678659483152223,
      "grad_norm": 0.11844311654567719,
      "learning_rate": 3.082191780821918e-06,
      "loss": 1.1962,
      "step": 594
    },
    {
      "epoch": 0.8693270021002648,
      "grad_norm": 0.12392130494117737,
      "learning_rate": 3.0479452054794525e-06,
      "loss": 1.1748,
      "step": 595
    },
    {
      "epoch": 0.8707880558853073,
      "grad_norm": 0.1305224448442459,
      "learning_rate": 3.0136986301369864e-06,
      "loss": 1.2388,
      "step": 596
    },
    {
      "epoch": 0.8722491096703497,
      "grad_norm": 0.11098136007785797,
      "learning_rate": 2.979452054794521e-06,
      "loss": 1.2109,
      "step": 597
    },
    {
      "epoch": 0.8737101634553922,
      "grad_norm": 0.11488241702318192,
      "learning_rate": 2.945205479452055e-06,
      "loss": 1.2222,
      "step": 598
    },
    {
      "epoch": 0.8751712172404347,
      "grad_norm": 0.11296083778142929,
      "learning_rate": 2.9109589041095893e-06,
      "loss": 1.2425,
      "step": 599
    },
    {
      "epoch": 0.8766322710254771,
      "grad_norm": 0.11494144797325134,
      "learning_rate": 2.876712328767123e-06,
      "loss": 1.1589,
      "step": 600
    },
    {
      "epoch": 0.8780933248105196,
      "grad_norm": 0.11837083101272583,
      "learning_rate": 2.842465753424658e-06,
      "loss": 1.1556,
      "step": 601
    },
    {
      "epoch": 0.8795543785955621,
      "grad_norm": 0.12680098414421082,
      "learning_rate": 2.8082191780821922e-06,
      "loss": 1.1742,
      "step": 602
    },
    {
      "epoch": 0.8810154323806045,
      "grad_norm": 0.12157981842756271,
      "learning_rate": 2.773972602739726e-06,
      "loss": 1.1715,
      "step": 603
    },
    {
      "epoch": 0.882476486165647,
      "grad_norm": 0.11859599500894547,
      "learning_rate": 2.7397260273972604e-06,
      "loss": 1.1988,
      "step": 604
    },
    {
      "epoch": 0.8839375399506895,
      "grad_norm": 0.11733684688806534,
      "learning_rate": 2.705479452054795e-06,
      "loss": 1.2507,
      "step": 605
    },
    {
      "epoch": 0.8853985937357319,
      "grad_norm": 0.11254255473613739,
      "learning_rate": 2.671232876712329e-06,
      "loss": 1.1843,
      "step": 606
    },
    {
      "epoch": 0.8868596475207744,
      "grad_norm": 0.11433933675289154,
      "learning_rate": 2.6369863013698634e-06,
      "loss": 1.1129,
      "step": 607
    },
    {
      "epoch": 0.8883207013058169,
      "grad_norm": 0.11361224204301834,
      "learning_rate": 2.6027397260273973e-06,
      "loss": 1.1883,
      "step": 608
    },
    {
      "epoch": 0.8897817550908593,
      "grad_norm": 0.11042333394289017,
      "learning_rate": 2.568493150684932e-06,
      "loss": 1.2322,
      "step": 609
    },
    {
      "epoch": 0.8912428088759018,
      "grad_norm": 0.1262194812297821,
      "learning_rate": 2.534246575342466e-06,
      "loss": 1.1767,
      "step": 610
    },
    {
      "epoch": 0.8927038626609443,
      "grad_norm": 0.10622208565473557,
      "learning_rate": 2.5e-06,
      "loss": 1.2355,
      "step": 611
    },
    {
      "epoch": 0.8941649164459866,
      "grad_norm": 0.11434955894947052,
      "learning_rate": 2.4657534246575345e-06,
      "loss": 1.1571,
      "step": 612
    },
    {
      "epoch": 0.8956259702310291,
      "grad_norm": 0.11948630958795547,
      "learning_rate": 2.431506849315069e-06,
      "loss": 1.1345,
      "step": 613
    },
    {
      "epoch": 0.8970870240160715,
      "grad_norm": 0.119502492249012,
      "learning_rate": 2.3972602739726027e-06,
      "loss": 1.1296,
      "step": 614
    },
    {
      "epoch": 0.898548077801114,
      "grad_norm": 0.1145474910736084,
      "learning_rate": 2.363013698630137e-06,
      "loss": 1.2316,
      "step": 615
    },
    {
      "epoch": 0.9000091315861565,
      "grad_norm": 0.11403004080057144,
      "learning_rate": 2.3287671232876713e-06,
      "loss": 1.1536,
      "step": 616
    },
    {
      "epoch": 0.9014701853711989,
      "grad_norm": 0.11280905455350876,
      "learning_rate": 2.2945205479452057e-06,
      "loss": 1.1922,
      "step": 617
    },
    {
      "epoch": 0.9029312391562414,
      "grad_norm": 0.10832927376031876,
      "learning_rate": 2.26027397260274e-06,
      "loss": 1.2386,
      "step": 618
    },
    {
      "epoch": 0.9043922929412839,
      "grad_norm": 0.11291555315256119,
      "learning_rate": 2.2260273972602743e-06,
      "loss": 1.1788,
      "step": 619
    },
    {
      "epoch": 0.9058533467263263,
      "grad_norm": 0.1250094771385193,
      "learning_rate": 2.191780821917808e-06,
      "loss": 1.2142,
      "step": 620
    },
    {
      "epoch": 0.9073144005113688,
      "grad_norm": 0.11631559580564499,
      "learning_rate": 2.1575342465753425e-06,
      "loss": 1.2121,
      "step": 621
    },
    {
      "epoch": 0.9087754542964113,
      "grad_norm": 0.11511734873056412,
      "learning_rate": 2.123287671232877e-06,
      "loss": 1.1137,
      "step": 622
    },
    {
      "epoch": 0.9102365080814537,
      "grad_norm": 0.12305217236280441,
      "learning_rate": 2.089041095890411e-06,
      "loss": 1.1745,
      "step": 623
    },
    {
      "epoch": 0.9116975618664962,
      "grad_norm": 0.14875584840774536,
      "learning_rate": 2.0547945205479454e-06,
      "loss": 1.1986,
      "step": 624
    },
    {
      "epoch": 0.9131586156515387,
      "grad_norm": 0.13255374133586884,
      "learning_rate": 2.0205479452054797e-06,
      "loss": 1.2716,
      "step": 625
    },
    {
      "epoch": 0.9146196694365811,
      "grad_norm": 0.13747917115688324,
      "learning_rate": 1.9863013698630136e-06,
      "loss": 1.1092,
      "step": 626
    },
    {
      "epoch": 0.9160807232216236,
      "grad_norm": 0.11340590566396713,
      "learning_rate": 1.952054794520548e-06,
      "loss": 1.2085,
      "step": 627
    },
    {
      "epoch": 0.9175417770066661,
      "grad_norm": 0.11387283354997635,
      "learning_rate": 1.9178082191780823e-06,
      "loss": 1.2549,
      "step": 628
    },
    {
      "epoch": 0.9190028307917085,
      "grad_norm": 0.11490115523338318,
      "learning_rate": 1.8835616438356166e-06,
      "loss": 1.1344,
      "step": 629
    },
    {
      "epoch": 0.920463884576751,
      "grad_norm": 0.10832976549863815,
      "learning_rate": 1.8493150684931507e-06,
      "loss": 1.2395,
      "step": 630
    },
    {
      "epoch": 0.9219249383617935,
      "grad_norm": 0.12825772166252136,
      "learning_rate": 1.8150684931506852e-06,
      "loss": 1.2296,
      "step": 631
    },
    {
      "epoch": 0.9233859921468359,
      "grad_norm": 0.1021864116191864,
      "learning_rate": 1.7808219178082193e-06,
      "loss": 1.2589,
      "step": 632
    },
    {
      "epoch": 0.9248470459318784,
      "grad_norm": 0.12429718673229218,
      "learning_rate": 1.7465753424657536e-06,
      "loss": 1.1167,
      "step": 633
    },
    {
      "epoch": 0.9263080997169209,
      "grad_norm": 0.10847421735525131,
      "learning_rate": 1.7123287671232877e-06,
      "loss": 1.1975,
      "step": 634
    },
    {
      "epoch": 0.9277691535019633,
      "grad_norm": 0.11923690140247345,
      "learning_rate": 1.678082191780822e-06,
      "loss": 1.1595,
      "step": 635
    },
    {
      "epoch": 0.9292302072870058,
      "grad_norm": 0.11933954805135727,
      "learning_rate": 1.6438356164383561e-06,
      "loss": 1.1694,
      "step": 636
    },
    {
      "epoch": 0.9306912610720482,
      "grad_norm": 0.11573485285043716,
      "learning_rate": 1.6095890410958907e-06,
      "loss": 1.1543,
      "step": 637
    },
    {
      "epoch": 0.9321523148570907,
      "grad_norm": 0.11935916543006897,
      "learning_rate": 1.5753424657534248e-06,
      "loss": 1.1666,
      "step": 638
    },
    {
      "epoch": 0.9336133686421332,
      "grad_norm": 0.1198139414191246,
      "learning_rate": 1.541095890410959e-06,
      "loss": 1.1385,
      "step": 639
    },
    {
      "epoch": 0.9350744224271756,
      "grad_norm": 0.12503781914710999,
      "learning_rate": 1.5068493150684932e-06,
      "loss": 1.2188,
      "step": 640
    },
    {
      "epoch": 0.9365354762122181,
      "grad_norm": 0.1288522630929947,
      "learning_rate": 1.4726027397260275e-06,
      "loss": 1.0925,
      "step": 641
    },
    {
      "epoch": 0.9379965299972606,
      "grad_norm": 0.11701378971338272,
      "learning_rate": 1.4383561643835616e-06,
      "loss": 1.1599,
      "step": 642
    },
    {
      "epoch": 0.939457583782303,
      "grad_norm": 0.10676533728837967,
      "learning_rate": 1.4041095890410961e-06,
      "loss": 1.2093,
      "step": 643
    },
    {
      "epoch": 0.9409186375673454,
      "grad_norm": 0.11982499808073044,
      "learning_rate": 1.3698630136986302e-06,
      "loss": 1.3175,
      "step": 644
    },
    {
      "epoch": 0.9423796913523879,
      "grad_norm": 0.12050015479326248,
      "learning_rate": 1.3356164383561645e-06,
      "loss": 1.1238,
      "step": 645
    },
    {
      "epoch": 0.9438407451374303,
      "grad_norm": 0.11197176575660706,
      "learning_rate": 1.3013698630136986e-06,
      "loss": 1.1744,
      "step": 646
    },
    {
      "epoch": 0.9453017989224728,
      "grad_norm": 0.11815720796585083,
      "learning_rate": 1.267123287671233e-06,
      "loss": 1.2155,
      "step": 647
    },
    {
      "epoch": 0.9467628527075153,
      "grad_norm": 0.1147996038198471,
      "learning_rate": 1.2328767123287673e-06,
      "loss": 1.2468,
      "step": 648
    },
    {
      "epoch": 0.9482239064925577,
      "grad_norm": 0.11407513171434402,
      "learning_rate": 1.1986301369863014e-06,
      "loss": 1.158,
      "step": 649
    },
    {
      "epoch": 0.9496849602776002,
      "grad_norm": 0.11162923276424408,
      "learning_rate": 1.1643835616438357e-06,
      "loss": 1.2155,
      "step": 650
    },
    {
      "epoch": 0.9511460140626427,
      "grad_norm": 0.1241702064871788,
      "learning_rate": 1.13013698630137e-06,
      "loss": 1.2487,
      "step": 651
    },
    {
      "epoch": 0.9526070678476851,
      "grad_norm": 0.11161646246910095,
      "learning_rate": 1.095890410958904e-06,
      "loss": 1.0931,
      "step": 652
    },
    {
      "epoch": 0.9540681216327276,
      "grad_norm": 0.11891372501850128,
      "learning_rate": 1.0616438356164384e-06,
      "loss": 1.2102,
      "step": 653
    },
    {
      "epoch": 0.95552917541777,
      "grad_norm": 0.11837035417556763,
      "learning_rate": 1.0273972602739727e-06,
      "loss": 1.1849,
      "step": 654
    },
    {
      "epoch": 0.9569902292028125,
      "grad_norm": 0.11559300124645233,
      "learning_rate": 9.931506849315068e-07,
      "loss": 1.2012,
      "step": 655
    },
    {
      "epoch": 0.958451282987855,
      "grad_norm": 0.1163790225982666,
      "learning_rate": 9.589041095890411e-07,
      "loss": 1.1365,
      "step": 656
    },
    {
      "epoch": 0.9599123367728974,
      "grad_norm": 0.10968944430351257,
      "learning_rate": 9.246575342465753e-07,
      "loss": 1.1666,
      "step": 657
    },
    {
      "epoch": 0.9613733905579399,
      "grad_norm": 0.11450762301683426,
      "learning_rate": 8.904109589041097e-07,
      "loss": 1.2182,
      "step": 658
    },
    {
      "epoch": 0.9628344443429824,
      "grad_norm": 0.11592509597539902,
      "learning_rate": 8.561643835616439e-07,
      "loss": 1.2405,
      "step": 659
    },
    {
      "epoch": 0.9642954981280248,
      "grad_norm": 0.1287543624639511,
      "learning_rate": 8.219178082191781e-07,
      "loss": 1.1236,
      "step": 660
    },
    {
      "epoch": 0.9657565519130673,
      "grad_norm": 0.10947469621896744,
      "learning_rate": 7.876712328767124e-07,
      "loss": 1.1835,
      "step": 661
    },
    {
      "epoch": 0.9672176056981098,
      "grad_norm": 0.11469222605228424,
      "learning_rate": 7.534246575342466e-07,
      "loss": 1.2212,
      "step": 662
    },
    {
      "epoch": 0.9686786594831522,
      "grad_norm": 0.11397232115268707,
      "learning_rate": 7.191780821917808e-07,
      "loss": 1.1476,
      "step": 663
    },
    {
      "epoch": 0.9701397132681947,
      "grad_norm": 0.11108572036027908,
      "learning_rate": 6.849315068493151e-07,
      "loss": 1.1669,
      "step": 664
    },
    {
      "epoch": 0.9716007670532372,
      "grad_norm": 0.10869178175926208,
      "learning_rate": 6.506849315068493e-07,
      "loss": 1.1264,
      "step": 665
    },
    {
      "epoch": 0.9730618208382796,
      "grad_norm": 0.10487518459558487,
      "learning_rate": 6.164383561643836e-07,
      "loss": 1.0964,
      "step": 666
    },
    {
      "epoch": 0.9745228746233221,
      "grad_norm": 0.1029355451464653,
      "learning_rate": 5.821917808219178e-07,
      "loss": 1.1172,
      "step": 667
    },
    {
      "epoch": 0.9759839284083646,
      "grad_norm": 0.12783120572566986,
      "learning_rate": 5.47945205479452e-07,
      "loss": 1.1128,
      "step": 668
    },
    {
      "epoch": 0.977444982193407,
      "grad_norm": 0.11498738080263138,
      "learning_rate": 5.136986301369864e-07,
      "loss": 1.1465,
      "step": 669
    },
    {
      "epoch": 0.9789060359784495,
      "grad_norm": 0.12313934415578842,
      "learning_rate": 4.794520547945206e-07,
      "loss": 1.1778,
      "step": 670
    },
    {
      "epoch": 0.980367089763492,
      "grad_norm": 0.12103428691625595,
      "learning_rate": 4.452054794520548e-07,
      "loss": 1.0667,
      "step": 671
    },
    {
      "epoch": 0.9818281435485344,
      "grad_norm": 0.11576645076274872,
      "learning_rate": 4.1095890410958903e-07,
      "loss": 1.217,
      "step": 672
    },
    {
      "epoch": 0.9832891973335769,
      "grad_norm": 0.10367967188358307,
      "learning_rate": 3.767123287671233e-07,
      "loss": 1.184,
      "step": 673
    },
    {
      "epoch": 0.9847502511186194,
      "grad_norm": 0.11073316633701324,
      "learning_rate": 3.4246575342465755e-07,
      "loss": 1.1595,
      "step": 674
    },
    {
      "epoch": 0.9862113049036617,
      "grad_norm": 0.12074057012796402,
      "learning_rate": 3.082191780821918e-07,
      "loss": 1.2545,
      "step": 675
    },
    {
      "epoch": 0.9876723586887042,
      "grad_norm": 0.11057423800230026,
      "learning_rate": 2.73972602739726e-07,
      "loss": 1.1668,
      "step": 676
    },
    {
      "epoch": 0.9891334124737466,
      "grad_norm": 0.125834122300148,
      "learning_rate": 2.397260273972603e-07,
      "loss": 1.1075,
      "step": 677
    },
    {
      "epoch": 0.9905944662587891,
      "grad_norm": 0.12944753468036652,
      "learning_rate": 2.0547945205479452e-07,
      "loss": 1.2271,
      "step": 678
    },
    {
      "epoch": 0.9920555200438316,
      "grad_norm": 0.11876823008060455,
      "learning_rate": 1.7123287671232878e-07,
      "loss": 1.1565,
      "step": 679
    },
    {
      "epoch": 0.993516573828874,
      "grad_norm": 0.12392497062683105,
      "learning_rate": 1.36986301369863e-07,
      "loss": 1.1049,
      "step": 680
    },
    {
      "epoch": 0.9949776276139165,
      "grad_norm": 0.11881987750530243,
      "learning_rate": 1.0273972602739726e-07,
      "loss": 1.2072,
      "step": 681
    },
    {
      "epoch": 0.996438681398959,
      "grad_norm": 0.12467513233423233,
      "learning_rate": 6.84931506849315e-08,
      "loss": 1.1384,
      "step": 682
    },
    {
      "epoch": 0.9978997351840014,
      "grad_norm": 0.11369941383600235,
      "learning_rate": 3.424657534246575e-08,
      "loss": 1.1439,
      "step": 683
    },
    {
      "epoch": 0.9993607889690439,
      "grad_norm": 0.1046438068151474,
      "learning_rate": 0.0,
      "loss": 1.1545,
      "step": 684
    }
  ],
  "logging_steps": 1,
  "max_steps": 684,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 7.73374714682794e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}