{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 277,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0036101083032490976,
      "grad_norm": 1.1218351125717163,
      "learning_rate": 1e-05,
      "loss": 2.188,
      "step": 1
    },
    {
      "epoch": 0.007220216606498195,
      "grad_norm": 1.0706913471221924,
      "learning_rate": 9.96389891696751e-06,
      "loss": 2.1819,
      "step": 2
    },
    {
      "epoch": 0.010830324909747292,
      "grad_norm": 1.0887945890426636,
      "learning_rate": 9.92779783393502e-06,
      "loss": 2.1912,
      "step": 3
    },
    {
      "epoch": 0.01444043321299639,
      "grad_norm": 1.0803953409194946,
      "learning_rate": 9.891696750902527e-06,
      "loss": 2.2241,
      "step": 4
    },
    {
      "epoch": 0.018050541516245487,
      "grad_norm": 1.0402294397354126,
      "learning_rate": 9.855595667870036e-06,
      "loss": 2.2483,
      "step": 5
    },
    {
      "epoch": 0.021660649819494584,
      "grad_norm": 1.0075435638427734,
      "learning_rate": 9.819494584837546e-06,
      "loss": 2.2007,
      "step": 6
    },
    {
      "epoch": 0.02527075812274368,
      "grad_norm": 0.9559407830238342,
      "learning_rate": 9.783393501805055e-06,
      "loss": 2.1599,
      "step": 7
    },
    {
      "epoch": 0.02888086642599278,
      "grad_norm": 0.940712571144104,
      "learning_rate": 9.747292418772564e-06,
      "loss": 2.1482,
      "step": 8
    },
    {
      "epoch": 0.032490974729241874,
      "grad_norm": 0.8784303069114685,
      "learning_rate": 9.711191335740074e-06,
      "loss": 2.1772,
      "step": 9
    },
    {
      "epoch": 0.036101083032490974,
      "grad_norm": 0.8126964569091797,
      "learning_rate": 9.675090252707581e-06,
      "loss": 2.1021,
      "step": 10
    },
    {
      "epoch": 0.039711191335740074,
      "grad_norm": 0.7782945036888123,
      "learning_rate": 9.63898916967509e-06,
      "loss": 2.0967,
      "step": 11
    },
    {
      "epoch": 0.04332129963898917,
      "grad_norm": 0.7165591716766357,
      "learning_rate": 9.6028880866426e-06,
      "loss": 2.0123,
      "step": 12
    },
    {
      "epoch": 0.04693140794223827,
      "grad_norm": 0.6946887373924255,
      "learning_rate": 9.56678700361011e-06,
      "loss": 2.0483,
      "step": 13
    },
    {
      "epoch": 0.05054151624548736,
      "grad_norm": 0.6478677988052368,
      "learning_rate": 9.530685920577619e-06,
      "loss": 2.0618,
      "step": 14
    },
    {
      "epoch": 0.05415162454873646,
      "grad_norm": 0.6255316138267517,
      "learning_rate": 9.494584837545126e-06,
      "loss": 2.0202,
      "step": 15
    },
    {
      "epoch": 0.05776173285198556,
      "grad_norm": 0.5520123839378357,
      "learning_rate": 9.458483754512636e-06,
      "loss": 1.9815,
      "step": 16
    },
    {
      "epoch": 0.061371841155234655,
      "grad_norm": 0.554833710193634,
      "learning_rate": 9.422382671480145e-06,
      "loss": 2.0234,
      "step": 17
    },
    {
      "epoch": 0.06498194945848375,
      "grad_norm": 0.5348688364028931,
      "learning_rate": 9.386281588447654e-06,
      "loss": 2.0197,
      "step": 18
    },
    {
      "epoch": 0.06859205776173286,
      "grad_norm": 0.4817121922969818,
      "learning_rate": 9.350180505415164e-06,
      "loss": 1.9425,
      "step": 19
    },
    {
      "epoch": 0.07220216606498195,
      "grad_norm": 0.44491052627563477,
      "learning_rate": 9.314079422382673e-06,
      "loss": 1.9109,
      "step": 20
    },
    {
      "epoch": 0.07581227436823104,
      "grad_norm": 0.43117740750312805,
      "learning_rate": 9.27797833935018e-06,
      "loss": 1.8809,
      "step": 21
    },
    {
      "epoch": 0.07942238267148015,
      "grad_norm": 0.4064043164253235,
      "learning_rate": 9.24187725631769e-06,
      "loss": 1.7955,
      "step": 22
    },
    {
      "epoch": 0.08303249097472924,
      "grad_norm": 0.4002125859260559,
      "learning_rate": 9.2057761732852e-06,
      "loss": 1.8682,
      "step": 23
    },
    {
      "epoch": 0.08664259927797834,
      "grad_norm": 0.4068574905395508,
      "learning_rate": 9.169675090252709e-06,
      "loss": 1.9182,
      "step": 24
    },
    {
      "epoch": 0.09025270758122744,
      "grad_norm": 0.39439308643341064,
      "learning_rate": 9.133574007220218e-06,
      "loss": 1.8909,
      "step": 25
    },
    {
      "epoch": 0.09386281588447654,
      "grad_norm": 0.4042982757091522,
      "learning_rate": 9.097472924187727e-06,
      "loss": 1.9381,
      "step": 26
    },
    {
      "epoch": 0.09747292418772563,
      "grad_norm": 0.3561984598636627,
      "learning_rate": 9.061371841155235e-06,
      "loss": 1.839,
      "step": 27
    },
    {
      "epoch": 0.10108303249097472,
      "grad_norm": 0.39211294054985046,
      "learning_rate": 9.025270758122744e-06,
      "loss": 1.8968,
      "step": 28
    },
    {
      "epoch": 0.10469314079422383,
      "grad_norm": 0.36844152212142944,
      "learning_rate": 8.989169675090254e-06,
      "loss": 1.8805,
      "step": 29
    },
    {
      "epoch": 0.10830324909747292,
      "grad_norm": 0.3706235885620117,
      "learning_rate": 8.953068592057763e-06,
      "loss": 1.8828,
      "step": 30
    },
    {
      "epoch": 0.11191335740072202,
      "grad_norm": 0.3436739444732666,
      "learning_rate": 8.916967509025272e-06,
      "loss": 1.8159,
      "step": 31
    },
    {
      "epoch": 0.11552346570397112,
      "grad_norm": 0.33314648270606995,
      "learning_rate": 8.88086642599278e-06,
      "loss": 1.8116,
      "step": 32
    },
    {
      "epoch": 0.11913357400722022,
      "grad_norm": 0.3236274719238281,
      "learning_rate": 8.84476534296029e-06,
      "loss": 1.7399,
      "step": 33
    },
    {
      "epoch": 0.12274368231046931,
      "grad_norm": 0.30125582218170166,
      "learning_rate": 8.808664259927798e-06,
      "loss": 1.7315,
      "step": 34
    },
    {
      "epoch": 0.1263537906137184,
      "grad_norm": 0.31622451543807983,
      "learning_rate": 8.772563176895308e-06,
      "loss": 1.7964,
      "step": 35
    },
    {
      "epoch": 0.1299638989169675,
      "grad_norm": 0.30419063568115234,
      "learning_rate": 8.736462093862817e-06,
      "loss": 1.7526,
      "step": 36
    },
    {
      "epoch": 0.13357400722021662,
      "grad_norm": 0.30785295367240906,
      "learning_rate": 8.700361010830326e-06,
      "loss": 1.7974,
      "step": 37
    },
    {
      "epoch": 0.1371841155234657,
      "grad_norm": 0.27913904190063477,
      "learning_rate": 8.664259927797834e-06,
      "loss": 1.7141,
      "step": 38
    },
    {
      "epoch": 0.1407942238267148,
      "grad_norm": 0.2888409495353699,
      "learning_rate": 8.628158844765343e-06,
      "loss": 1.7374,
      "step": 39
    },
    {
      "epoch": 0.1444043321299639,
      "grad_norm": 0.29159021377563477,
      "learning_rate": 8.592057761732853e-06,
      "loss": 1.7762,
      "step": 40
    },
    {
      "epoch": 0.148014440433213,
      "grad_norm": 0.28307339549064636,
      "learning_rate": 8.55595667870036e-06,
      "loss": 1.7281,
      "step": 41
    },
    {
      "epoch": 0.15162454873646208,
      "grad_norm": 0.28662118315696716,
      "learning_rate": 8.519855595667871e-06,
      "loss": 1.7282,
      "step": 42
    },
    {
      "epoch": 0.1552346570397112,
      "grad_norm": 0.29065990447998047,
      "learning_rate": 8.483754512635379e-06,
      "loss": 1.7636,
      "step": 43
    },
    {
      "epoch": 0.1588447653429603,
      "grad_norm": 0.2732274532318115,
      "learning_rate": 8.447653429602888e-06,
      "loss": 1.6769,
      "step": 44
    },
    {
      "epoch": 0.1624548736462094,
      "grad_norm": 0.28228530287742615,
      "learning_rate": 8.411552346570398e-06,
      "loss": 1.7207,
      "step": 45
    },
    {
      "epoch": 0.16606498194945848,
      "grad_norm": 0.2729104161262512,
      "learning_rate": 8.375451263537907e-06,
      "loss": 1.6888,
      "step": 46
    },
    {
      "epoch": 0.16967509025270758,
      "grad_norm": 0.26737287640571594,
      "learning_rate": 8.339350180505416e-06,
      "loss": 1.6536,
      "step": 47
    },
    {
      "epoch": 0.17328519855595667,
      "grad_norm": 0.2611989378929138,
      "learning_rate": 8.303249097472926e-06,
      "loss": 1.6595,
      "step": 48
    },
    {
      "epoch": 0.17689530685920576,
      "grad_norm": 0.2761898338794708,
      "learning_rate": 8.267148014440433e-06,
      "loss": 1.6674,
      "step": 49
    },
    {
      "epoch": 0.18050541516245489,
      "grad_norm": 0.26143068075180054,
      "learning_rate": 8.231046931407943e-06,
      "loss": 1.6382,
      "step": 50
    },
    {
      "epoch": 0.18411552346570398,
      "grad_norm": 0.2734948992729187,
      "learning_rate": 8.194945848375452e-06,
      "loss": 1.7228,
      "step": 51
    },
    {
      "epoch": 0.18772563176895307,
      "grad_norm": 0.2682507634162903,
      "learning_rate": 8.158844765342961e-06,
      "loss": 1.6465,
      "step": 52
    },
    {
      "epoch": 0.19133574007220217,
      "grad_norm": 0.2903922498226166,
      "learning_rate": 8.12274368231047e-06,
      "loss": 1.6962,
      "step": 53
    },
    {
      "epoch": 0.19494584837545126,
      "grad_norm": 0.2923874855041504,
      "learning_rate": 8.086642599277978e-06,
      "loss": 1.7269,
      "step": 54
    },
    {
      "epoch": 0.19855595667870035,
      "grad_norm": 0.26083049178123474,
      "learning_rate": 8.050541516245488e-06,
      "loss": 1.6442,
      "step": 55
    },
    {
      "epoch": 0.20216606498194944,
      "grad_norm": 0.24434912204742432,
      "learning_rate": 8.014440433212997e-06,
      "loss": 1.586,
      "step": 56
    },
    {
      "epoch": 0.20577617328519857,
      "grad_norm": 0.24050913751125336,
      "learning_rate": 7.978339350180506e-06,
      "loss": 1.6119,
      "step": 57
    },
    {
      "epoch": 0.20938628158844766,
      "grad_norm": 0.24940010905265808,
      "learning_rate": 7.942238267148014e-06,
      "loss": 1.5994,
      "step": 58
    },
    {
      "epoch": 0.21299638989169675,
      "grad_norm": 0.24224700033664703,
      "learning_rate": 7.906137184115525e-06,
      "loss": 1.5855,
      "step": 59
    },
    {
      "epoch": 0.21660649819494585,
      "grad_norm": 0.2521527409553528,
      "learning_rate": 7.870036101083033e-06,
      "loss": 1.5845,
      "step": 60
    },
    {
      "epoch": 0.22021660649819494,
      "grad_norm": 0.24293367564678192,
      "learning_rate": 7.833935018050542e-06,
      "loss": 1.5839,
      "step": 61
    },
    {
      "epoch": 0.22382671480144403,
      "grad_norm": 0.23730704188346863,
      "learning_rate": 7.797833935018051e-06,
      "loss": 1.5611,
      "step": 62
    },
    {
      "epoch": 0.22743682310469315,
      "grad_norm": 0.23844872415065765,
      "learning_rate": 7.76173285198556e-06,
      "loss": 1.5947,
      "step": 63
    },
    {
      "epoch": 0.23104693140794225,
      "grad_norm": 0.243188738822937,
      "learning_rate": 7.72563176895307e-06,
      "loss": 1.58,
      "step": 64
    },
    {
      "epoch": 0.23465703971119134,
      "grad_norm": 0.22330617904663086,
      "learning_rate": 7.68953068592058e-06,
      "loss": 1.5329,
      "step": 65
    },
    {
      "epoch": 0.23826714801444043,
      "grad_norm": 0.2230277955532074,
      "learning_rate": 7.653429602888087e-06,
      "loss": 1.5251,
      "step": 66
    },
    {
      "epoch": 0.24187725631768953,
      "grad_norm": 0.23090355098247528,
      "learning_rate": 7.617328519855596e-06,
      "loss": 1.5686,
      "step": 67
    },
    {
      "epoch": 0.24548736462093862,
      "grad_norm": 0.23035725951194763,
      "learning_rate": 7.5812274368231055e-06,
      "loss": 1.5596,
      "step": 68
    },
    {
      "epoch": 0.2490974729241877,
      "grad_norm": 0.2521900236606598,
      "learning_rate": 7.545126353790614e-06,
      "loss": 1.57,
      "step": 69
    },
    {
      "epoch": 0.2527075812274368,
      "grad_norm": 0.23036813735961914,
      "learning_rate": 7.509025270758123e-06,
      "loss": 1.5087,
      "step": 70
    },
    {
      "epoch": 0.2563176895306859,
      "grad_norm": 0.24895897507667542,
      "learning_rate": 7.472924187725632e-06,
      "loss": 1.5433,
      "step": 71
    },
    {
      "epoch": 0.259927797833935,
      "grad_norm": 0.2232387810945511,
      "learning_rate": 7.436823104693142e-06,
      "loss": 1.5283,
      "step": 72
    },
    {
      "epoch": 0.26353790613718414,
      "grad_norm": 0.23834098875522614,
      "learning_rate": 7.40072202166065e-06,
      "loss": 1.5719,
      "step": 73
    },
    {
      "epoch": 0.26714801444043323,
      "grad_norm": 0.24203678965568542,
      "learning_rate": 7.36462093862816e-06,
      "loss": 1.5768,
      "step": 74
    },
    {
      "epoch": 0.27075812274368233,
      "grad_norm": 0.2327604442834854,
      "learning_rate": 7.328519855595668e-06,
      "loss": 1.5387,
      "step": 75
    },
    {
      "epoch": 0.2743682310469314,
      "grad_norm": 0.2127024084329605,
      "learning_rate": 7.2924187725631776e-06,
      "loss": 1.4921,
      "step": 76
    },
    {
      "epoch": 0.2779783393501805,
      "grad_norm": 0.2509189248085022,
      "learning_rate": 7.256317689530686e-06,
      "loss": 1.565,
      "step": 77
    },
    {
      "epoch": 0.2815884476534296,
      "grad_norm": 0.23548553884029388,
      "learning_rate": 7.220216606498196e-06,
      "loss": 1.5078,
      "step": 78
    },
    {
      "epoch": 0.2851985559566787,
      "grad_norm": 0.23434241116046906,
      "learning_rate": 7.184115523465705e-06,
      "loss": 1.5232,
      "step": 79
    },
    {
      "epoch": 0.2888086642599278,
      "grad_norm": 0.22626014053821564,
      "learning_rate": 7.148014440433214e-06,
      "loss": 1.5143,
      "step": 80
    },
    {
      "epoch": 0.2924187725631769,
      "grad_norm": 0.232896089553833,
      "learning_rate": 7.1119133574007225e-06,
      "loss": 1.5495,
      "step": 81
    },
    {
      "epoch": 0.296028880866426,
      "grad_norm": 0.22791652381420135,
      "learning_rate": 7.075812274368231e-06,
      "loss": 1.5021,
      "step": 82
    },
    {
      "epoch": 0.2996389891696751,
      "grad_norm": 0.2307957261800766,
      "learning_rate": 7.039711191335741e-06,
      "loss": 1.5721,
      "step": 83
    },
    {
      "epoch": 0.30324909747292417,
      "grad_norm": 0.26458626985549927,
      "learning_rate": 7.00361010830325e-06,
      "loss": 1.5046,
      "step": 84
    },
    {
      "epoch": 0.30685920577617326,
      "grad_norm": 0.20370201766490936,
      "learning_rate": 6.967509025270759e-06,
      "loss": 1.4624,
      "step": 85
    },
    {
      "epoch": 0.3104693140794224,
      "grad_norm": 0.2249036282300949,
      "learning_rate": 6.9314079422382674e-06,
      "loss": 1.5127,
      "step": 86
    },
    {
      "epoch": 0.3140794223826715,
      "grad_norm": 0.22176551818847656,
      "learning_rate": 6.895306859205777e-06,
      "loss": 1.4924,
      "step": 87
    },
    {
      "epoch": 0.3176895306859206,
      "grad_norm": 0.19935470819473267,
      "learning_rate": 6.859205776173285e-06,
      "loss": 1.4207,
      "step": 88
    },
    {
      "epoch": 0.3212996389891697,
      "grad_norm": 0.21199044585227966,
      "learning_rate": 6.8231046931407954e-06,
      "loss": 1.4677,
      "step": 89
    },
    {
      "epoch": 0.3249097472924188,
      "grad_norm": 0.2081277221441269,
      "learning_rate": 6.787003610108304e-06,
      "loss": 1.443,
      "step": 90
    },
    {
      "epoch": 0.3285198555956679,
      "grad_norm": 0.20040768384933472,
      "learning_rate": 6.750902527075813e-06,
      "loss": 1.474,
      "step": 91
    },
    {
      "epoch": 0.33212996389891697,
      "grad_norm": 0.2211485207080841,
      "learning_rate": 6.714801444043322e-06,
      "loss": 1.4466,
      "step": 92
    },
    {
      "epoch": 0.33574007220216606,
      "grad_norm": 0.21461904048919678,
      "learning_rate": 6.678700361010831e-06,
      "loss": 1.4769,
      "step": 93
    },
    {
      "epoch": 0.33935018050541516,
      "grad_norm": 0.20499303936958313,
      "learning_rate": 6.6425992779783395e-06,
      "loss": 1.4367,
      "step": 94
    },
    {
      "epoch": 0.34296028880866425,
      "grad_norm": 0.2087281048297882,
      "learning_rate": 6.606498194945848e-06,
      "loss": 1.4919,
      "step": 95
    },
    {
      "epoch": 0.34657039711191334,
      "grad_norm": 0.22192475199699402,
      "learning_rate": 6.570397111913358e-06,
      "loss": 1.4722,
      "step": 96
    },
    {
      "epoch": 0.35018050541516244,
      "grad_norm": 0.21253931522369385,
      "learning_rate": 6.534296028880867e-06,
      "loss": 1.4654,
      "step": 97
    },
    {
      "epoch": 0.35379061371841153,
      "grad_norm": 0.19912089407444,
      "learning_rate": 6.498194945848376e-06,
      "loss": 1.4187,
      "step": 98
    },
    {
      "epoch": 0.3574007220216607,
      "grad_norm": 0.1975143700838089,
      "learning_rate": 6.4620938628158845e-06,
      "loss": 1.4604,
      "step": 99
    },
    {
      "epoch": 0.36101083032490977,
      "grad_norm": 0.1989564597606659,
      "learning_rate": 6.425992779783395e-06,
      "loss": 1.4471,
      "step": 100
    },
    {
      "epoch": 0.36462093862815886,
      "grad_norm": 0.19660496711730957,
      "learning_rate": 6.389891696750903e-06,
      "loss": 1.4487,
      "step": 101
    },
    {
      "epoch": 0.36823104693140796,
      "grad_norm": 0.20031259953975677,
      "learning_rate": 6.3537906137184125e-06,
      "loss": 1.4529,
      "step": 102
    },
    {
      "epoch": 0.37184115523465705,
      "grad_norm": 0.22296781837940216,
      "learning_rate": 6.317689530685921e-06,
      "loss": 1.5193,
      "step": 103
    },
    {
      "epoch": 0.37545126353790614,
      "grad_norm": 0.23671561479568481,
      "learning_rate": 6.28158844765343e-06,
      "loss": 1.5017,
      "step": 104
    },
    {
      "epoch": 0.37906137184115524,
      "grad_norm": 0.20504920184612274,
      "learning_rate": 6.245487364620939e-06,
      "loss": 1.453,
      "step": 105
    },
    {
      "epoch": 0.38267148014440433,
      "grad_norm": 0.2209184765815735,
      "learning_rate": 6.209386281588449e-06,
      "loss": 1.4509,
      "step": 106
    },
    {
      "epoch": 0.3862815884476534,
      "grad_norm": 0.20382221043109894,
      "learning_rate": 6.173285198555957e-06,
      "loss": 1.4356,
      "step": 107
    },
    {
      "epoch": 0.3898916967509025,
      "grad_norm": 0.2022784799337387,
      "learning_rate": 6.137184115523466e-06,
      "loss": 1.4164,
      "step": 108
    },
    {
      "epoch": 0.3935018050541516,
      "grad_norm": 0.1905573010444641,
      "learning_rate": 6.101083032490975e-06,
      "loss": 1.4002,
      "step": 109
    },
    {
      "epoch": 0.3971119133574007,
      "grad_norm": 0.20327116549015045,
      "learning_rate": 6.064981949458484e-06,
      "loss": 1.3884,
      "step": 110
    },
    {
      "epoch": 0.4007220216606498,
      "grad_norm": 0.19279006123542786,
      "learning_rate": 6.028880866425994e-06,
      "loss": 1.4392,
      "step": 111
    },
    {
      "epoch": 0.4043321299638989,
      "grad_norm": 0.19752241671085358,
      "learning_rate": 5.992779783393502e-06,
      "loss": 1.4033,
      "step": 112
    },
    {
      "epoch": 0.40794223826714804,
      "grad_norm": 0.20032241940498352,
      "learning_rate": 5.956678700361012e-06,
      "loss": 1.4291,
      "step": 113
    },
    {
      "epoch": 0.41155234657039713,
      "grad_norm": 0.21105395257472992,
      "learning_rate": 5.92057761732852e-06,
      "loss": 1.4483,
      "step": 114
    },
    {
      "epoch": 0.4151624548736462,
      "grad_norm": 0.19472186267375946,
      "learning_rate": 5.8844765342960295e-06,
      "loss": 1.4071,
      "step": 115
    },
    {
      "epoch": 0.4187725631768953,
      "grad_norm": 0.1926383376121521,
      "learning_rate": 5.848375451263538e-06,
      "loss": 1.4419,
      "step": 116
    },
    {
      "epoch": 0.4223826714801444,
      "grad_norm": 0.18957462906837463,
      "learning_rate": 5.812274368231048e-06,
      "loss": 1.4055,
      "step": 117
    },
    {
      "epoch": 0.4259927797833935,
      "grad_norm": 0.19267480075359344,
      "learning_rate": 5.776173285198557e-06,
      "loss": 1.4112,
      "step": 118
    },
    {
      "epoch": 0.4296028880866426,
      "grad_norm": 0.18771202862262726,
      "learning_rate": 5.740072202166066e-06,
      "loss": 1.4294,
      "step": 119
    },
    {
      "epoch": 0.4332129963898917,
      "grad_norm": 0.19841831922531128,
      "learning_rate": 5.7039711191335744e-06,
      "loss": 1.4272,
      "step": 120
    },
    {
      "epoch": 0.4368231046931408,
      "grad_norm": 0.20646512508392334,
      "learning_rate": 5.667870036101083e-06,
      "loss": 1.3947,
      "step": 121
    },
    {
      "epoch": 0.4404332129963899,
      "grad_norm": 0.20326825976371765,
      "learning_rate": 5.631768953068592e-06,
      "loss": 1.4393,
      "step": 122
    },
    {
      "epoch": 0.44404332129963897,
      "grad_norm": 0.19117839634418488,
      "learning_rate": 5.595667870036101e-06,
      "loss": 1.3686,
      "step": 123
    },
    {
      "epoch": 0.44765342960288806,
      "grad_norm": 0.19725248217582703,
      "learning_rate": 5.559566787003611e-06,
      "loss": 1.4325,
      "step": 124
    },
    {
      "epoch": 0.45126353790613716,
      "grad_norm": 0.20171114802360535,
      "learning_rate": 5.523465703971119e-06,
      "loss": 1.3884,
      "step": 125
    },
    {
      "epoch": 0.4548736462093863,
      "grad_norm": 0.21069733798503876,
      "learning_rate": 5.487364620938629e-06,
      "loss": 1.4115,
      "step": 126
    },
    {
      "epoch": 0.4584837545126354,
      "grad_norm": 0.19288866221904755,
      "learning_rate": 5.451263537906137e-06,
      "loss": 1.4237,
      "step": 127
    },
    {
      "epoch": 0.4620938628158845,
      "grad_norm": 0.21202199161052704,
      "learning_rate": 5.415162454873647e-06,
      "loss": 1.4356,
      "step": 128
    },
    {
      "epoch": 0.4657039711191336,
      "grad_norm": 0.19240565598011017,
      "learning_rate": 5.379061371841156e-06,
      "loss": 1.3668,
      "step": 129
    },
    {
      "epoch": 0.4693140794223827,
      "grad_norm": 0.20009998977184296,
      "learning_rate": 5.342960288808665e-06,
      "loss": 1.4231,
      "step": 130
    },
    {
      "epoch": 0.4729241877256318,
      "grad_norm": 0.19136758148670197,
      "learning_rate": 5.306859205776174e-06,
      "loss": 1.4169,
      "step": 131
    },
    {
      "epoch": 0.47653429602888087,
      "grad_norm": 0.21354779601097107,
      "learning_rate": 5.270758122743683e-06,
      "loss": 1.43,
      "step": 132
    },
    {
      "epoch": 0.48014440433212996,
      "grad_norm": 0.20004898309707642,
      "learning_rate": 5.2346570397111915e-06,
      "loss": 1.4465,
      "step": 133
    },
    {
      "epoch": 0.48375451263537905,
      "grad_norm": 0.21821723878383636,
      "learning_rate": 5.1985559566787e-06,
      "loss": 1.4209,
      "step": 134
    },
    {
      "epoch": 0.48736462093862815,
      "grad_norm": 0.19806598126888275,
      "learning_rate": 5.16245487364621e-06,
      "loss": 1.3948,
      "step": 135
    },
    {
      "epoch": 0.49097472924187724,
      "grad_norm": 0.202366441488266,
      "learning_rate": 5.126353790613719e-06,
      "loss": 1.3961,
      "step": 136
    },
    {
      "epoch": 0.49458483754512633,
      "grad_norm": 0.20319777727127075,
      "learning_rate": 5.090252707581228e-06,
      "loss": 1.4122,
      "step": 137
    },
    {
      "epoch": 0.4981949458483754,
      "grad_norm": 0.1868642121553421,
      "learning_rate": 5.054151624548736e-06,
      "loss": 1.3635,
      "step": 138
    },
    {
      "epoch": 0.5018050541516246,
      "grad_norm": 0.2048720419406891,
      "learning_rate": 5.018050541516246e-06,
      "loss": 1.4029,
      "step": 139
    },
    {
      "epoch": 0.5054151624548736,
      "grad_norm": 0.19640134274959564,
      "learning_rate": 4.981949458483755e-06,
      "loss": 1.3501,
      "step": 140
    },
    {
      "epoch": 0.5090252707581228,
      "grad_norm": 0.19755016267299652,
      "learning_rate": 4.9458483754512636e-06,
      "loss": 1.3714,
      "step": 141
    },
    {
      "epoch": 0.5126353790613718,
      "grad_norm": 0.2008562535047531,
      "learning_rate": 4.909747292418773e-06,
      "loss": 1.371,
      "step": 142
    },
    {
      "epoch": 0.516245487364621,
      "grad_norm": 0.20218950510025024,
      "learning_rate": 4.873646209386282e-06,
      "loss": 1.4145,
      "step": 143
    },
    {
      "epoch": 0.51985559566787,
      "grad_norm": 0.19422906637191772,
      "learning_rate": 4.837545126353791e-06,
      "loss": 1.3604,
      "step": 144
    },
    {
      "epoch": 0.5234657039711191,
      "grad_norm": 0.20418153703212738,
      "learning_rate": 4.8014440433213e-06,
      "loss": 1.3712,
      "step": 145
    },
    {
      "epoch": 0.5270758122743683,
      "grad_norm": 0.20942777395248413,
      "learning_rate": 4.765342960288809e-06,
      "loss": 1.402,
      "step": 146
    },
    {
      "epoch": 0.5306859205776173,
      "grad_norm": 0.18804116547107697,
      "learning_rate": 4.729241877256318e-06,
      "loss": 1.3649,
      "step": 147
    },
    {
      "epoch": 0.5342960288808665,
      "grad_norm": 0.22690172493457794,
      "learning_rate": 4.693140794223827e-06,
      "loss": 1.3621,
      "step": 148
    },
    {
      "epoch": 0.5379061371841155,
      "grad_norm": 0.18934603035449982,
      "learning_rate": 4.6570397111913365e-06,
      "loss": 1.3749,
      "step": 149
    },
    {
      "epoch": 0.5415162454873647,
      "grad_norm": 0.18682512640953064,
      "learning_rate": 4.620938628158845e-06,
      "loss": 1.3191,
      "step": 150
    },
    {
      "epoch": 0.5451263537906137,
      "grad_norm": 0.19843855500221252,
      "learning_rate": 4.584837545126354e-06,
      "loss": 1.3605,
      "step": 151
    },
    {
      "epoch": 0.5487364620938628,
      "grad_norm": 0.18685825169086456,
      "learning_rate": 4.548736462093864e-06,
      "loss": 1.3702,
      "step": 152
    },
    {
      "epoch": 0.5523465703971119,
      "grad_norm": 0.19559639692306519,
      "learning_rate": 4.512635379061372e-06,
      "loss": 1.3844,
      "step": 153
    },
    {
      "epoch": 0.555956678700361,
      "grad_norm": 0.2046169936656952,
      "learning_rate": 4.4765342960288814e-06,
      "loss": 1.3849,
      "step": 154
    },
    {
      "epoch": 0.5595667870036101,
      "grad_norm": 0.20239655673503876,
      "learning_rate": 4.44043321299639e-06,
      "loss": 1.3784,
      "step": 155
    },
    {
      "epoch": 0.5631768953068592,
      "grad_norm": 0.19824427366256714,
      "learning_rate": 4.404332129963899e-06,
      "loss": 1.3398,
      "step": 156
    },
    {
      "epoch": 0.5667870036101083,
      "grad_norm": 0.1943778544664383,
      "learning_rate": 4.3682310469314086e-06,
      "loss": 1.3683,
      "step": 157
    },
    {
      "epoch": 0.5703971119133574,
      "grad_norm": 0.2199310064315796,
      "learning_rate": 4.332129963898917e-06,
      "loss": 1.4149,
      "step": 158
    },
    {
      "epoch": 0.5740072202166066,
      "grad_norm": 0.21495173871517181,
      "learning_rate": 4.296028880866426e-06,
      "loss": 1.4314,
      "step": 159
    },
    {
      "epoch": 0.5776173285198556,
      "grad_norm": 0.2000962346792221,
      "learning_rate": 4.259927797833936e-06,
      "loss": 1.3201,
      "step": 160
    },
    {
      "epoch": 0.5812274368231047,
      "grad_norm": 0.22053833305835724,
      "learning_rate": 4.223826714801444e-06,
      "loss": 1.4319,
      "step": 161
    },
    {
      "epoch": 0.5848375451263538,
      "grad_norm": 0.20683500170707703,
      "learning_rate": 4.1877256317689535e-06,
      "loss": 1.3582,
      "step": 162
    },
    {
      "epoch": 0.5884476534296029,
      "grad_norm": 0.1908424198627472,
      "learning_rate": 4.151624548736463e-06,
      "loss": 1.349,
      "step": 163
    },
    {
      "epoch": 0.592057761732852,
      "grad_norm": 0.20667223632335663,
      "learning_rate": 4.115523465703971e-06,
      "loss": 1.3862,
      "step": 164
    },
    {
      "epoch": 0.5956678700361011,
      "grad_norm": 0.20890134572982788,
      "learning_rate": 4.079422382671481e-06,
      "loss": 1.3584,
      "step": 165
    },
    {
      "epoch": 0.5992779783393501,
      "grad_norm": 0.20082373917102814,
      "learning_rate": 4.043321299638989e-06,
      "loss": 1.2859,
      "step": 166
    },
    {
      "epoch": 0.6028880866425993,
      "grad_norm": 0.19916357100009918,
      "learning_rate": 4.0072202166064985e-06,
      "loss": 1.3676,
      "step": 167
    },
    {
      "epoch": 0.6064981949458483,
      "grad_norm": 0.20149283111095428,
      "learning_rate": 3.971119133574007e-06,
      "loss": 1.3477,
      "step": 168
    },
    {
      "epoch": 0.6101083032490975,
      "grad_norm": 0.20036858320236206,
      "learning_rate": 3.935018050541516e-06,
      "loss": 1.3326,
      "step": 169
    },
    {
      "epoch": 0.6137184115523465,
      "grad_norm": 0.20373424887657166,
      "learning_rate": 3.898916967509026e-06,
      "loss": 1.3656,
      "step": 170
    },
    {
      "epoch": 0.6173285198555957,
      "grad_norm": 0.1955791711807251,
      "learning_rate": 3.862815884476535e-06,
      "loss": 1.3238,
      "step": 171
    },
    {
      "epoch": 0.6209386281588448,
      "grad_norm": 0.20329371094703674,
      "learning_rate": 3.826714801444043e-06,
      "loss": 1.3233,
      "step": 172
    },
    {
      "epoch": 0.6245487364620939,
      "grad_norm": 0.19480253756046295,
      "learning_rate": 3.7906137184115527e-06,
      "loss": 1.3483,
      "step": 173
    },
    {
      "epoch": 0.628158844765343,
      "grad_norm": 0.20373401045799255,
      "learning_rate": 3.7545126353790616e-06,
      "loss": 1.3276,
      "step": 174
    },
    {
      "epoch": 0.631768953068592,
      "grad_norm": 0.21212375164031982,
      "learning_rate": 3.718411552346571e-06,
      "loss": 1.3661,
      "step": 175
    },
    {
      "epoch": 0.6353790613718412,
      "grad_norm": 0.196991965174675,
      "learning_rate": 3.68231046931408e-06,
      "loss": 1.3723,
      "step": 176
    },
    {
      "epoch": 0.6389891696750902,
      "grad_norm": 0.19335833191871643,
      "learning_rate": 3.6462093862815888e-06,
      "loss": 1.3218,
      "step": 177
    },
    {
      "epoch": 0.6425992779783394,
      "grad_norm": 0.20568343997001648,
      "learning_rate": 3.610108303249098e-06,
      "loss": 1.3535,
      "step": 178
    },
    {
      "epoch": 0.6462093862815884,
      "grad_norm": 0.20011267066001892,
      "learning_rate": 3.574007220216607e-06,
      "loss": 1.3505,
      "step": 179
    },
    {
      "epoch": 0.6498194945848376,
      "grad_norm": 0.21634064614772797,
      "learning_rate": 3.5379061371841155e-06,
      "loss": 1.3959,
      "step": 180
    },
    {
      "epoch": 0.6534296028880866,
      "grad_norm": 0.19454725086688995,
      "learning_rate": 3.501805054151625e-06,
      "loss": 1.3235,
      "step": 181
    },
    {
      "epoch": 0.6570397111913358,
      "grad_norm": 0.206266850233078,
      "learning_rate": 3.4657039711191337e-06,
      "loss": 1.3527,
      "step": 182
    },
    {
      "epoch": 0.6606498194945848,
      "grad_norm": 0.20744888484477997,
      "learning_rate": 3.4296028880866426e-06,
      "loss": 1.3609,
      "step": 183
    },
    {
      "epoch": 0.6642599277978339,
      "grad_norm": 0.19529123604297638,
      "learning_rate": 3.393501805054152e-06,
      "loss": 1.3409,
      "step": 184
    },
    {
      "epoch": 0.6678700361010831,
      "grad_norm": 0.20020008087158203,
      "learning_rate": 3.357400722021661e-06,
      "loss": 1.3426,
      "step": 185
    },
    {
      "epoch": 0.6714801444043321,
      "grad_norm": 0.23242731392383575,
      "learning_rate": 3.3212996389891698e-06,
      "loss": 1.3632,
      "step": 186
    },
    {
      "epoch": 0.6750902527075813,
      "grad_norm": 0.19558390974998474,
      "learning_rate": 3.285198555956679e-06,
      "loss": 1.356,
      "step": 187
    },
    {
      "epoch": 0.6787003610108303,
      "grad_norm": 0.21582651138305664,
      "learning_rate": 3.249097472924188e-06,
      "loss": 1.3416,
      "step": 188
    },
    {
      "epoch": 0.6823104693140795,
      "grad_norm": 0.19333289563655853,
      "learning_rate": 3.2129963898916973e-06,
      "loss": 1.2946,
      "step": 189
    },
    {
      "epoch": 0.6859205776173285,
      "grad_norm": 0.19065658748149872,
      "learning_rate": 3.1768953068592062e-06,
      "loss": 1.3152,
      "step": 190
    },
    {
      "epoch": 0.6895306859205776,
      "grad_norm": 0.19440875947475433,
      "learning_rate": 3.140794223826715e-06,
      "loss": 1.3243,
      "step": 191
    },
    {
      "epoch": 0.6931407942238267,
      "grad_norm": 0.18864405155181885,
      "learning_rate": 3.1046931407942245e-06,
      "loss": 1.3322,
      "step": 192
    },
    {
      "epoch": 0.6967509025270758,
      "grad_norm": 0.1976729929447174,
      "learning_rate": 3.068592057761733e-06,
      "loss": 1.3288,
      "step": 193
    },
    {
      "epoch": 0.7003610108303249,
      "grad_norm": 0.19966968894004822,
      "learning_rate": 3.032490974729242e-06,
      "loss": 1.3215,
      "step": 194
    },
    {
      "epoch": 0.703971119133574,
      "grad_norm": 0.19344742596149445,
      "learning_rate": 2.996389891696751e-06,
      "loss": 1.3391,
      "step": 195
    },
    {
      "epoch": 0.7075812274368231,
      "grad_norm": 0.1955103874206543,
      "learning_rate": 2.96028880866426e-06,
      "loss": 1.3181,
      "step": 196
    },
    {
      "epoch": 0.7111913357400722,
      "grad_norm": 0.21449753642082214,
      "learning_rate": 2.924187725631769e-06,
      "loss": 1.3647,
      "step": 197
    },
    {
      "epoch": 0.7148014440433214,
      "grad_norm": 0.1900106519460678,
      "learning_rate": 2.8880866425992783e-06,
      "loss": 1.3671,
      "step": 198
    },
    {
      "epoch": 0.7184115523465704,
      "grad_norm": 0.20474538207054138,
      "learning_rate": 2.8519855595667872e-06,
      "loss": 1.3417,
      "step": 199
    },
    {
      "epoch": 0.7220216606498195,
      "grad_norm": 0.2103157937526703,
      "learning_rate": 2.815884476534296e-06,
      "loss": 1.348,
      "step": 200
    },
    {
      "epoch": 0.7256317689530686,
      "grad_norm": 0.2024817168712616,
      "learning_rate": 2.7797833935018055e-06,
      "loss": 1.3295,
      "step": 201
    },
    {
      "epoch": 0.7292418772563177,
      "grad_norm": 0.19940879940986633,
      "learning_rate": 2.7436823104693144e-06,
      "loss": 1.3012,
      "step": 202
    },
    {
      "epoch": 0.7328519855595668,
      "grad_norm": 0.1980341076850891,
      "learning_rate": 2.7075812274368237e-06,
      "loss": 1.3099,
      "step": 203
    },
    {
      "epoch": 0.7364620938628159,
      "grad_norm": 0.201655313372612,
      "learning_rate": 2.6714801444043326e-06,
      "loss": 1.3601,
      "step": 204
    },
    {
      "epoch": 0.740072202166065,
      "grad_norm": 0.2011563926935196,
      "learning_rate": 2.6353790613718415e-06,
      "loss": 1.3372,
      "step": 205
    },
    {
      "epoch": 0.7436823104693141,
      "grad_norm": 0.19773615896701813,
      "learning_rate": 2.59927797833935e-06,
      "loss": 1.3081,
      "step": 206
    },
    {
      "epoch": 0.7472924187725631,
      "grad_norm": 0.20711122453212738,
      "learning_rate": 2.5631768953068593e-06,
      "loss": 1.3639,
      "step": 207
    },
    {
      "epoch": 0.7509025270758123,
      "grad_norm": 0.19647595286369324,
      "learning_rate": 2.527075812274368e-06,
      "loss": 1.305,
      "step": 208
    },
    {
      "epoch": 0.7545126353790613,
      "grad_norm": 0.19561149179935455,
      "learning_rate": 2.4909747292418775e-06,
      "loss": 1.2962,
      "step": 209
    },
    {
      "epoch": 0.7581227436823105,
      "grad_norm": 0.21176967024803162,
      "learning_rate": 2.4548736462093864e-06,
      "loss": 1.3721,
      "step": 210
    },
    {
      "epoch": 0.7617328519855595,
      "grad_norm": 0.2063320130109787,
      "learning_rate": 2.4187725631768953e-06,
      "loss": 1.2738,
      "step": 211
    },
    {
      "epoch": 0.7653429602888087,
      "grad_norm": 0.20307239890098572,
      "learning_rate": 2.3826714801444047e-06,
      "loss": 1.3045,
      "step": 212
    },
    {
      "epoch": 0.7689530685920578,
      "grad_norm": 0.21186378598213196,
      "learning_rate": 2.3465703971119136e-06,
      "loss": 1.3652,
      "step": 213
    },
    {
      "epoch": 0.7725631768953068,
      "grad_norm": 0.19658677279949188,
      "learning_rate": 2.3104693140794225e-06,
      "loss": 1.3755,
      "step": 214
    },
    {
      "epoch": 0.776173285198556,
      "grad_norm": 0.21189101040363312,
      "learning_rate": 2.274368231046932e-06,
      "loss": 1.3618,
      "step": 215
    },
    {
      "epoch": 0.779783393501805,
      "grad_norm": 0.2035369873046875,
      "learning_rate": 2.2382671480144407e-06,
      "loss": 1.3704,
      "step": 216
    },
    {
      "epoch": 0.7833935018050542,
      "grad_norm": 0.19990962743759155,
      "learning_rate": 2.2021660649819496e-06,
      "loss": 1.3603,
      "step": 217
    },
    {
      "epoch": 0.7870036101083032,
      "grad_norm": 0.19858066737651825,
      "learning_rate": 2.1660649819494585e-06,
      "loss": 1.3509,
      "step": 218
    },
    {
      "epoch": 0.7906137184115524,
      "grad_norm": 0.21003876626491547,
      "learning_rate": 2.129963898916968e-06,
      "loss": 1.3435,
      "step": 219
    },
    {
      "epoch": 0.7942238267148014,
      "grad_norm": 0.19455671310424805,
      "learning_rate": 2.0938628158844768e-06,
      "loss": 1.3381,
      "step": 220
    },
    {
      "epoch": 0.7978339350180506,
      "grad_norm": 0.20084233582019806,
      "learning_rate": 2.0577617328519857e-06,
      "loss": 1.3275,
      "step": 221
    },
    {
      "epoch": 0.8014440433212996,
      "grad_norm": 0.20646274089813232,
      "learning_rate": 2.0216606498194946e-06,
      "loss": 1.3127,
      "step": 222
    },
    {
      "epoch": 0.8050541516245487,
      "grad_norm": 0.1980086714029312,
      "learning_rate": 1.9855595667870035e-06,
      "loss": 1.355,
      "step": 223
    },
    {
      "epoch": 0.8086642599277978,
      "grad_norm": 0.19509509205818176,
      "learning_rate": 1.949458483754513e-06,
      "loss": 1.3236,
      "step": 224
    },
    {
      "epoch": 0.8122743682310469,
      "grad_norm": 0.19844523072242737,
      "learning_rate": 1.9133574007220217e-06,
      "loss": 1.3071,
      "step": 225
    },
    {
      "epoch": 0.8158844765342961,
      "grad_norm": 0.18677139282226562,
      "learning_rate": 1.8772563176895308e-06,
      "loss": 1.3124,
      "step": 226
    },
    {
      "epoch": 0.8194945848375451,
      "grad_norm": 0.2121885120868683,
      "learning_rate": 1.84115523465704e-06,
      "loss": 1.2569,
      "step": 227
    },
    {
      "epoch": 0.8231046931407943,
      "grad_norm": 0.21662482619285583,
      "learning_rate": 1.805054151624549e-06,
      "loss": 1.2857,
      "step": 228
    },
    {
      "epoch": 0.8267148014440433,
      "grad_norm": 0.20647631585597992,
      "learning_rate": 1.7689530685920577e-06,
      "loss": 1.3425,
      "step": 229
    },
    {
      "epoch": 0.8303249097472925,
      "grad_norm": 0.21089228987693787,
      "learning_rate": 1.7328519855595669e-06,
      "loss": 1.3092,
      "step": 230
    },
    {
      "epoch": 0.8339350180505415,
      "grad_norm": 0.19865262508392334,
      "learning_rate": 1.696750902527076e-06,
      "loss": 1.3075,
      "step": 231
    },
    {
      "epoch": 0.8375451263537906,
      "grad_norm": 0.20838730037212372,
      "learning_rate": 1.6606498194945849e-06,
      "loss": 1.3776,
      "step": 232
    },
    {
      "epoch": 0.8411552346570397,
      "grad_norm": 0.20943178236484528,
      "learning_rate": 1.624548736462094e-06,
      "loss": 1.3552,
      "step": 233
    },
    {
      "epoch": 0.8447653429602888,
      "grad_norm": 0.19375556707382202,
      "learning_rate": 1.5884476534296031e-06,
      "loss": 1.318,
      "step": 234
    },
    {
      "epoch": 0.8483754512635379,
      "grad_norm": 0.23595024645328522,
      "learning_rate": 1.5523465703971122e-06,
      "loss": 1.3337,
      "step": 235
    },
    {
      "epoch": 0.851985559566787,
      "grad_norm": 0.20870931446552277,
      "learning_rate": 1.516245487364621e-06,
      "loss": 1.3896,
      "step": 236
    },
    {
      "epoch": 0.855595667870036,
      "grad_norm": 0.1979263424873352,
      "learning_rate": 1.48014440433213e-06,
      "loss": 1.3133,
      "step": 237
    },
    {
      "epoch": 0.8592057761732852,
      "grad_norm": 0.2063916027545929,
      "learning_rate": 1.4440433212996392e-06,
      "loss": 1.3236,
      "step": 238
    },
    {
      "epoch": 0.8628158844765343,
      "grad_norm": 0.20475609600543976,
      "learning_rate": 1.407942238267148e-06,
      "loss": 1.3094,
      "step": 239
    },
    {
      "epoch": 0.8664259927797834,
      "grad_norm": 0.21313494443893433,
      "learning_rate": 1.3718411552346572e-06,
      "loss": 1.3487,
      "step": 240
    },
    {
      "epoch": 0.8700361010830325,
      "grad_norm": 0.20646831393241882,
      "learning_rate": 1.3357400722021663e-06,
      "loss": 1.3421,
      "step": 241
    },
    {
      "epoch": 0.8736462093862816,
      "grad_norm": 0.2132684737443924,
      "learning_rate": 1.299638989169675e-06,
      "loss": 1.3254,
      "step": 242
    },
    {
      "epoch": 0.8772563176895307,
      "grad_norm": 0.20300714671611786,
      "learning_rate": 1.263537906137184e-06,
      "loss": 1.2963,
      "step": 243
    },
    {
      "epoch": 0.8808664259927798,
      "grad_norm": 0.20637428760528564,
      "learning_rate": 1.2274368231046932e-06,
      "loss": 1.3531,
      "step": 244
    },
    {
      "epoch": 0.8844765342960289,
      "grad_norm": 0.20430119335651398,
      "learning_rate": 1.1913357400722023e-06,
      "loss": 1.3097,
      "step": 245
    },
    {
      "epoch": 0.8880866425992779,
      "grad_norm": 0.2088884711265564,
      "learning_rate": 1.1552346570397112e-06,
      "loss": 1.3079,
      "step": 246
    },
    {
      "epoch": 0.8916967509025271,
      "grad_norm": 0.24322503805160522,
      "learning_rate": 1.1191335740072204e-06,
      "loss": 1.3841,
      "step": 247
    },
    {
      "epoch": 0.8953068592057761,
      "grad_norm": 0.2223634272813797,
      "learning_rate": 1.0830324909747293e-06,
      "loss": 1.2808,
      "step": 248
    },
    {
      "epoch": 0.8989169675090253,
      "grad_norm": 0.20814542472362518,
      "learning_rate": 1.0469314079422384e-06,
      "loss": 1.3186,
      "step": 249
    },
    {
      "epoch": 0.9025270758122743,
      "grad_norm": 0.203385591506958,
      "learning_rate": 1.0108303249097473e-06,
      "loss": 1.2978,
      "step": 250
    },
    {
      "epoch": 0.9061371841155235,
      "grad_norm": 0.19583970308303833,
      "learning_rate": 9.747292418772564e-07,
      "loss": 1.3459,
      "step": 251
    },
    {
      "epoch": 0.9097472924187726,
      "grad_norm": 0.1949346959590912,
      "learning_rate": 9.386281588447654e-07,
      "loss": 1.287,
      "step": 252
    },
    {
      "epoch": 0.9133574007220217,
      "grad_norm": 0.2315586656332016,
      "learning_rate": 9.025270758122745e-07,
      "loss": 1.3514,
      "step": 253
    },
    {
      "epoch": 0.9169675090252708,
      "grad_norm": 0.20318041741847992,
      "learning_rate": 8.664259927797834e-07,
      "loss": 1.2926,
      "step": 254
    },
    {
      "epoch": 0.9205776173285198,
      "grad_norm": 0.20969466865062714,
      "learning_rate": 8.303249097472924e-07,
      "loss": 1.3309,
      "step": 255
    },
    {
      "epoch": 0.924187725631769,
      "grad_norm": 0.20613840222358704,
      "learning_rate": 7.942238267148016e-07,
      "loss": 1.3204,
      "step": 256
    },
    {
      "epoch": 0.927797833935018,
      "grad_norm": 0.2004636526107788,
      "learning_rate": 7.581227436823105e-07,
      "loss": 1.3267,
      "step": 257
    },
    {
      "epoch": 0.9314079422382672,
      "grad_norm": 0.21113616228103638,
      "learning_rate": 7.220216606498196e-07,
      "loss": 1.2908,
      "step": 258
    },
    {
      "epoch": 0.9350180505415162,
      "grad_norm": 0.1977560967206955,
      "learning_rate": 6.859205776173286e-07,
      "loss": 1.3163,
      "step": 259
    },
    {
      "epoch": 0.9386281588447654,
      "grad_norm": 0.2119782716035843,
      "learning_rate": 6.498194945848375e-07,
      "loss": 1.3165,
      "step": 260
    },
    {
      "epoch": 0.9422382671480144,
      "grad_norm": 0.19387714564800262,
      "learning_rate": 6.137184115523466e-07,
      "loss": 1.2911,
      "step": 261
    },
    {
      "epoch": 0.9458483754512635,
      "grad_norm": 0.21642278134822845,
      "learning_rate": 5.776173285198556e-07,
      "loss": 1.303,
      "step": 262
    },
    {
      "epoch": 0.9494584837545126,
      "grad_norm": 0.21271206438541412,
      "learning_rate": 5.415162454873646e-07,
      "loss": 1.3273,
      "step": 263
    },
    {
      "epoch": 0.9530685920577617,
      "grad_norm": 0.21591399610042572,
      "learning_rate": 5.054151624548736e-07,
      "loss": 1.3196,
      "step": 264
    },
    {
      "epoch": 0.9566787003610109,
      "grad_norm": 0.2054390162229538,
      "learning_rate": 4.693140794223827e-07,
      "loss": 1.3363,
      "step": 265
    },
    {
      "epoch": 0.9602888086642599,
      "grad_norm": 0.2292117476463318,
      "learning_rate": 4.332129963898917e-07,
      "loss": 1.3768,
      "step": 266
    },
    {
      "epoch": 0.9638989169675091,
      "grad_norm": 0.2526095509529114,
      "learning_rate": 3.971119133574008e-07,
      "loss": 1.4071,
      "step": 267
    },
    {
      "epoch": 0.9675090252707581,
      "grad_norm": 0.21546489000320435,
      "learning_rate": 3.610108303249098e-07,
      "loss": 1.3661,
      "step": 268
    },
    {
      "epoch": 0.9711191335740073,
      "grad_norm": 0.20055970549583435,
      "learning_rate": 3.2490974729241875e-07,
      "loss": 1.3288,
      "step": 269
    },
    {
      "epoch": 0.9747292418772563,
      "grad_norm": 0.25362637639045715,
      "learning_rate": 2.888086642599278e-07,
      "loss": 1.4092,
      "step": 270
    },
    {
      "epoch": 0.9783393501805054,
      "grad_norm": 0.2010817676782608,
      "learning_rate": 2.527075812274368e-07,
      "loss": 1.3286,
      "step": 271
    },
    {
      "epoch": 0.9819494584837545,
      "grad_norm": 0.20375026762485504,
      "learning_rate": 2.1660649819494586e-07,
      "loss": 1.3446,
      "step": 272
    },
    {
      "epoch": 0.9855595667870036,
      "grad_norm": 0.2000964730978012,
      "learning_rate": 1.805054151624549e-07,
      "loss": 1.3223,
      "step": 273
    },
    {
      "epoch": 0.9891696750902527,
      "grad_norm": 0.19899840652942657,
      "learning_rate": 1.444043321299639e-07,
      "loss": 1.3393,
      "step": 274
    },
    {
      "epoch": 0.9927797833935018,
      "grad_norm": 0.20534487068653107,
      "learning_rate": 1.0830324909747293e-07,
      "loss": 1.3059,
      "step": 275
    },
    {
      "epoch": 0.9963898916967509,
      "grad_norm": 0.20792889595031738,
      "learning_rate": 7.220216606498195e-08,
      "loss": 1.3125,
      "step": 276
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.20410171151161194,
      "learning_rate": 3.6101083032490976e-08,
      "loss": 1.3321,
      "step": 277
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 277,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 9.827066833587405e+16,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}