{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 4.423800044238001,
  "eval_steps": 500,
  "global_step": 20000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.022119000221190004,
      "grad_norm": 2561.85302734375,
      "learning_rate": 2.5e-06,
      "loss": 7244.9438,
      "step": 100
    },
    {
      "epoch": 0.04423800044238001,
      "grad_norm": 1641.6300048828125,
      "learning_rate": 5e-06,
      "loss": 278.0641,
      "step": 200
    },
    {
      "epoch": 0.06635700066357,
      "grad_norm": 413.9412841796875,
      "learning_rate": 7.5e-06,
      "loss": 305.1475,
      "step": 300
    },
    {
      "epoch": 0.08847600088476001,
      "grad_norm": 298.65997314453125,
      "learning_rate": 1e-05,
      "loss": 296.6687,
      "step": 400
    },
    {
      "epoch": 0.11059500110595001,
      "grad_norm": 612.1175537109375,
      "learning_rate": 1.25e-05,
      "loss": 247.2131,
      "step": 500
    },
    {
      "epoch": 0.13271400132714,
      "grad_norm": 1398.299560546875,
      "learning_rate": 1.5e-05,
      "loss": 273.615,
      "step": 600
    },
    {
      "epoch": 0.15483300154833002,
      "grad_norm": 538.8013305664062,
      "learning_rate": 1.75e-05,
      "loss": 277.284,
      "step": 700
    },
    {
      "epoch": 0.17695200176952003,
      "grad_norm": 834.0810546875,
      "learning_rate": 2e-05,
      "loss": 343.3682,
      "step": 800
    },
    {
      "epoch": 0.19907100199071,
      "grad_norm": 514.625732421875,
      "learning_rate": 2.25e-05,
      "loss": 172.4172,
      "step": 900
    },
    {
      "epoch": 0.22119000221190002,
      "grad_norm": 720.7930908203125,
      "learning_rate": 2.5e-05,
      "loss": 243.5434,
      "step": 1000
    },
    {
      "epoch": 0.24330900243309003,
      "grad_norm": 538.9693603515625,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 227.322,
      "step": 1100
    },
    {
      "epoch": 0.26542800265428,
      "grad_norm": 365.2975769042969,
      "learning_rate": 3e-05,
      "loss": 289.7278,
      "step": 1200
    },
    {
      "epoch": 0.28754700287547,
      "grad_norm": 1308.2694091796875,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 191.7522,
      "step": 1300
    },
    {
      "epoch": 0.30966600309666004,
      "grad_norm": 6047.15625,
      "learning_rate": 3.5e-05,
      "loss": 316.0845,
      "step": 1400
    },
    {
      "epoch": 0.33178500331785005,
      "grad_norm": 482.5833740234375,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 275.0296,
      "step": 1500
    },
    {
      "epoch": 0.35390400353904006,
      "grad_norm": 811.97607421875,
      "learning_rate": 4e-05,
      "loss": 232.9227,
      "step": 1600
    },
    {
      "epoch": 0.37602300376023,
      "grad_norm": 585.74365234375,
      "learning_rate": 4.25e-05,
      "loss": 157.1875,
      "step": 1700
    },
    {
      "epoch": 0.39814200398142,
      "grad_norm": 2657.204833984375,
      "learning_rate": 4.5e-05,
      "loss": 202.9244,
      "step": 1800
    },
    {
      "epoch": 0.42026100420261003,
      "grad_norm": 788.7911376953125,
      "learning_rate": 4.75e-05,
      "loss": 230.3789,
      "step": 1900
    },
    {
      "epoch": 0.44238000442380004,
      "grad_norm": 1466.7137451171875,
      "learning_rate": 5e-05,
      "loss": 184.6603,
      "step": 2000
    },
    {
      "epoch": 0.46449900464499005,
      "grad_norm": 717.4619750976562,
      "learning_rate": 4.972222222222223e-05,
      "loss": 160.0884,
      "step": 2100
    },
    {
      "epoch": 0.48661800486618007,
      "grad_norm": 1592.6282958984375,
      "learning_rate": 4.9444444444444446e-05,
      "loss": 159.9456,
      "step": 2200
    },
    {
      "epoch": 0.5087370050873701,
      "grad_norm": 258.3177795410156,
      "learning_rate": 4.9166666666666665e-05,
      "loss": 272.3899,
      "step": 2300
    },
    {
      "epoch": 0.53085600530856,
      "grad_norm": 2940.7119140625,
      "learning_rate": 4.888888888888889e-05,
      "loss": 108.9058,
      "step": 2400
    },
    {
      "epoch": 0.5529750055297501,
      "grad_norm": 1595.506103515625,
      "learning_rate": 4.8611111111111115e-05,
      "loss": 137.0249,
      "step": 2500
    },
    {
      "epoch": 0.57509400575094,
      "grad_norm": 476.0992126464844,
      "learning_rate": 4.8333333333333334e-05,
      "loss": 157.8203,
      "step": 2600
    },
    {
      "epoch": 0.59721300597213,
      "grad_norm": 462.79119873046875,
      "learning_rate": 4.805555555555556e-05,
      "loss": 202.3364,
      "step": 2700
    },
    {
      "epoch": 0.6193320061933201,
      "grad_norm": 7305.47900390625,
      "learning_rate": 4.7777777777777784e-05,
      "loss": 176.8393,
      "step": 2800
    },
    {
      "epoch": 0.64145100641451,
      "grad_norm": 1471.9434814453125,
      "learning_rate": 4.75e-05,
      "loss": 192.2721,
      "step": 2900
    },
    {
      "epoch": 0.6635700066357001,
      "grad_norm": 253.4272003173828,
      "learning_rate": 4.722222222222222e-05,
      "loss": 204.4471,
      "step": 3000
    },
    {
      "epoch": 0.68568900685689,
      "grad_norm": 4707.13916015625,
      "learning_rate": 4.6944444444444446e-05,
      "loss": 154.4462,
      "step": 3100
    },
    {
      "epoch": 0.7078080070780801,
      "grad_norm": 477.1028747558594,
      "learning_rate": 4.666666666666667e-05,
      "loss": 93.3721,
      "step": 3200
    },
    {
      "epoch": 0.7299270072992701,
      "grad_norm": 492.2008361816406,
      "learning_rate": 4.638888888888889e-05,
      "loss": 129.002,
      "step": 3300
    },
    {
      "epoch": 0.75204600752046,
      "grad_norm": 414.3881530761719,
      "learning_rate": 4.6111111111111115e-05,
      "loss": 88.6518,
      "step": 3400
    },
    {
      "epoch": 0.7741650077416501,
      "grad_norm": 170.11138916015625,
      "learning_rate": 4.5833333333333334e-05,
      "loss": 193.1568,
      "step": 3500
    },
    {
      "epoch": 0.79628400796284,
      "grad_norm": 787.8067016601562,
      "learning_rate": 4.555555555555556e-05,
      "loss": 153.1195,
      "step": 3600
    },
    {
      "epoch": 0.8184030081840301,
      "grad_norm": 940.7820434570312,
      "learning_rate": 4.527777777777778e-05,
      "loss": 312.0311,
      "step": 3700
    },
    {
      "epoch": 0.8405220084052201,
      "grad_norm": 719.3717041015625,
      "learning_rate": 4.5e-05,
      "loss": 160.2157,
      "step": 3800
    },
    {
      "epoch": 0.8626410086264101,
      "grad_norm": 710.9347534179688,
      "learning_rate": 4.472222222222223e-05,
      "loss": 149.8423,
      "step": 3900
    },
    {
      "epoch": 0.8847600088476001,
      "grad_norm": 812.0198364257812,
      "learning_rate": 4.4444444444444447e-05,
      "loss": 171.4201,
      "step": 4000
    },
    {
      "epoch": 0.90687900906879,
      "grad_norm": 347.59246826171875,
      "learning_rate": 4.4166666666666665e-05,
      "loss": 85.4954,
      "step": 4100
    },
    {
      "epoch": 0.9289980092899801,
      "grad_norm": 494.03570556640625,
      "learning_rate": 4.388888888888889e-05,
      "loss": 122.7032,
      "step": 4200
    },
    {
      "epoch": 0.9511170095111701,
      "grad_norm": 1107.2685546875,
      "learning_rate": 4.3611111111111116e-05,
      "loss": 110.0679,
      "step": 4300
    },
    {
      "epoch": 0.9732360097323601,
      "grad_norm": 554.3077392578125,
      "learning_rate": 4.3333333333333334e-05,
      "loss": 162.5302,
      "step": 4400
    },
    {
      "epoch": 0.9953550099535501,
      "grad_norm": 1022.9378662109375,
      "learning_rate": 4.305555555555556e-05,
      "loss": 123.3146,
      "step": 4500
    },
    {
      "epoch": 1.0,
      "eval_loss": 317.1508483886719,
      "eval_runtime": 20.7601,
      "eval_samples_per_second": 96.82,
      "eval_steps_per_second": 24.229,
      "step": 4521
    },
    {
      "epoch": 1.0174740101747402,
      "grad_norm": 604.2803344726562,
      "learning_rate": 4.277777777777778e-05,
      "loss": 87.0649,
      "step": 4600
    },
    {
      "epoch": 1.03959301039593,
      "grad_norm": 655.1663208007812,
      "learning_rate": 4.25e-05,
      "loss": 60.0608,
      "step": 4700
    },
    {
      "epoch": 1.06171201061712,
      "grad_norm": 613.4644165039062,
      "learning_rate": 4.222222222222222e-05,
      "loss": 152.5058,
      "step": 4800
    },
    {
      "epoch": 1.08383101083831,
      "grad_norm": 943.52392578125,
      "learning_rate": 4.194444444444445e-05,
      "loss": 112.7452,
      "step": 4900
    },
    {
      "epoch": 1.1059500110595002,
      "grad_norm": 741.4888916015625,
      "learning_rate": 4.166666666666667e-05,
      "loss": 180.4975,
      "step": 5000
    },
    {
      "epoch": 1.1280690112806901,
      "grad_norm": 984.3648071289062,
      "learning_rate": 4.138888888888889e-05,
      "loss": 109.8211,
      "step": 5100
    },
    {
      "epoch": 1.15018801150188,
      "grad_norm": 4328.1767578125,
      "learning_rate": 4.111111111111111e-05,
      "loss": 86.7452,
      "step": 5200
    },
    {
      "epoch": 1.17230701172307,
      "grad_norm": 440.3537902832031,
      "learning_rate": 4.0833333333333334e-05,
      "loss": 145.0485,
      "step": 5300
    },
    {
      "epoch": 1.1944260119442602,
      "grad_norm": 1904.060302734375,
      "learning_rate": 4.055555555555556e-05,
      "loss": 137.598,
      "step": 5400
    },
    {
      "epoch": 1.2165450121654502,
      "grad_norm": 325.7207336425781,
      "learning_rate": 4.027777777777778e-05,
      "loss": 133.5432,
      "step": 5500
    },
    {
      "epoch": 1.2386640123866401,
      "grad_norm": 602.099609375,
      "learning_rate": 4e-05,
      "loss": 90.936,
      "step": 5600
    },
    {
      "epoch": 1.26078301260783,
      "grad_norm": 373.9248962402344,
      "learning_rate": 3.972222222222222e-05,
      "loss": 126.1215,
      "step": 5700
    },
    {
      "epoch": 1.28290201282902,
      "grad_norm": 1740.2845458984375,
      "learning_rate": 3.944444444444445e-05,
      "loss": 106.4278,
      "step": 5800
    },
    {
      "epoch": 1.3050210130502102,
      "grad_norm": 619.2306518554688,
      "learning_rate": 3.9166666666666665e-05,
      "loss": 69.9521,
      "step": 5900
    },
    {
      "epoch": 1.3271400132714002,
      "grad_norm": 397.7071838378906,
      "learning_rate": 3.888888888888889e-05,
      "loss": 92.6775,
      "step": 6000
    },
    {
      "epoch": 1.3492590134925901,
      "grad_norm": 261.9856872558594,
      "learning_rate": 3.8611111111111116e-05,
      "loss": 139.3951,
      "step": 6100
    },
    {
      "epoch": 1.37137801371378,
      "grad_norm": 326.20806884765625,
      "learning_rate": 3.8333333333333334e-05,
      "loss": 102.4056,
      "step": 6200
    },
    {
      "epoch": 1.39349701393497,
      "grad_norm": 5344.56005859375,
      "learning_rate": 3.805555555555555e-05,
      "loss": 108.2192,
      "step": 6300
    },
    {
      "epoch": 1.4156160141561602,
      "grad_norm": 1083.1983642578125,
      "learning_rate": 3.777777777777778e-05,
      "loss": 121.1852,
      "step": 6400
    },
    {
      "epoch": 1.4377350143773502,
      "grad_norm": 285.3344421386719,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 113.1963,
      "step": 6500
    },
    {
      "epoch": 1.4598540145985401,
      "grad_norm": 1590.1190185546875,
      "learning_rate": 3.722222222222222e-05,
      "loss": 101.6212,
      "step": 6600
    },
    {
      "epoch": 1.48197301481973,
      "grad_norm": 242.43016052246094,
      "learning_rate": 3.694444444444445e-05,
      "loss": 100.9942,
      "step": 6700
    },
    {
      "epoch": 1.50409201504092,
      "grad_norm": 419.2840576171875,
      "learning_rate": 3.6666666666666666e-05,
      "loss": 91.4742,
      "step": 6800
    },
    {
      "epoch": 1.5262110152621102,
      "grad_norm": 145.44712829589844,
      "learning_rate": 3.638888888888889e-05,
      "loss": 108.68,
      "step": 6900
    },
    {
      "epoch": 1.5483300154833002,
      "grad_norm": 866.0416259765625,
      "learning_rate": 3.611111111111111e-05,
      "loss": 139.873,
      "step": 7000
    },
    {
      "epoch": 1.5704490157044901,
      "grad_norm": 543.1744995117188,
      "learning_rate": 3.5833333333333335e-05,
      "loss": 84.9976,
      "step": 7100
    },
    {
      "epoch": 1.5925680159256803,
      "grad_norm": 961.9137573242188,
      "learning_rate": 3.555555555555556e-05,
      "loss": 91.9599,
      "step": 7200
    },
    {
      "epoch": 1.61468701614687,
      "grad_norm": 1136.900146484375,
      "learning_rate": 3.527777777777778e-05,
      "loss": 100.5749,
      "step": 7300
    },
    {
      "epoch": 1.6368060163680602,
      "grad_norm": 845.6568603515625,
      "learning_rate": 3.5e-05,
      "loss": 128.4079,
      "step": 7400
    },
    {
      "epoch": 1.6589250165892502,
      "grad_norm": 1455.4808349609375,
      "learning_rate": 3.472222222222222e-05,
      "loss": 124.5921,
      "step": 7500
    },
    {
      "epoch": 1.6810440168104401,
      "grad_norm": 370.6483459472656,
      "learning_rate": 3.444444444444445e-05,
      "loss": 103.7647,
      "step": 7600
    },
    {
      "epoch": 1.7031630170316303,
      "grad_norm": 9296.732421875,
      "learning_rate": 3.4166666666666666e-05,
      "loss": 104.9438,
      "step": 7700
    },
    {
      "epoch": 1.72528201725282,
      "grad_norm": 2471.158447265625,
      "learning_rate": 3.388888888888889e-05,
      "loss": 109.1481,
      "step": 7800
    },
    {
      "epoch": 1.7474010174740102,
      "grad_norm": 230.1200408935547,
      "learning_rate": 3.3611111111111116e-05,
      "loss": 87.4035,
      "step": 7900
    },
    {
      "epoch": 1.7695200176952002,
      "grad_norm": 875.7386474609375,
      "learning_rate": 3.3333333333333335e-05,
      "loss": 87.8452,
      "step": 8000
    },
    {
      "epoch": 1.7916390179163901,
      "grad_norm": 1371.0994873046875,
      "learning_rate": 3.3055555555555553e-05,
      "loss": 98.3798,
      "step": 8100
    },
    {
      "epoch": 1.8137580181375803,
      "grad_norm": 682.475341796875,
      "learning_rate": 3.277777777777778e-05,
      "loss": 104.5044,
      "step": 8200
    },
    {
      "epoch": 1.83587701835877,
      "grad_norm": 512.0708618164062,
      "learning_rate": 3.2500000000000004e-05,
      "loss": 64.1792,
      "step": 8300
    },
    {
      "epoch": 1.8579960185799602,
      "grad_norm": 1042.0284423828125,
      "learning_rate": 3.222222222222223e-05,
      "loss": 74.1599,
      "step": 8400
    },
    {
      "epoch": 1.8801150188011502,
      "grad_norm": 358.9875183105469,
      "learning_rate": 3.194444444444444e-05,
      "loss": 93.9574,
      "step": 8500
    },
    {
      "epoch": 1.9022340190223401,
      "grad_norm": 4613.166015625,
      "learning_rate": 3.1666666666666666e-05,
      "loss": 103.6234,
      "step": 8600
    },
    {
      "epoch": 1.9243530192435303,
      "grad_norm": 436.9840393066406,
      "learning_rate": 3.138888888888889e-05,
      "loss": 98.4847,
      "step": 8700
    },
    {
      "epoch": 1.94647201946472,
      "grad_norm": 29151.51171875,
      "learning_rate": 3.111111111111111e-05,
      "loss": 96.9473,
      "step": 8800
    },
    {
      "epoch": 1.9685910196859102,
      "grad_norm": 189.884033203125,
      "learning_rate": 3.0833333333333335e-05,
      "loss": 65.5306,
      "step": 8900
    },
    {
      "epoch": 1.9907100199071002,
      "grad_norm": 416.74969482421875,
      "learning_rate": 3.055555555555556e-05,
      "loss": 104.1101,
      "step": 9000
    },
    {
      "epoch": 2.0,
      "eval_loss": 227.4776611328125,
      "eval_runtime": 20.702,
      "eval_samples_per_second": 97.092,
      "eval_steps_per_second": 24.297,
      "step": 9042
    },
    {
      "epoch": 2.01282902012829,
      "grad_norm": 345.6188049316406,
      "learning_rate": 3.0277777777777776e-05,
      "loss": 103.7359,
      "step": 9100
    },
    {
      "epoch": 2.0349480203494803,
      "grad_norm": 2839.026611328125,
      "learning_rate": 3e-05,
      "loss": 83.4189,
      "step": 9200
    },
    {
      "epoch": 2.05706702057067,
      "grad_norm": 493.89093017578125,
      "learning_rate": 2.9722222222222223e-05,
      "loss": 93.5555,
      "step": 9300
    },
    {
      "epoch": 2.07918602079186,
      "grad_norm": 239.73707580566406,
      "learning_rate": 2.9444444444444448e-05,
      "loss": 102.3052,
      "step": 9400
    },
    {
      "epoch": 2.1013050210130504,
      "grad_norm": 293.8266296386719,
      "learning_rate": 2.916666666666667e-05,
      "loss": 80.1726,
      "step": 9500
    },
    {
      "epoch": 2.12342402123424,
      "grad_norm": 251.94790649414062,
      "learning_rate": 2.8888888888888888e-05,
      "loss": 57.1753,
      "step": 9600
    },
    {
      "epoch": 2.1455430214554303,
      "grad_norm": 1789.1011962890625,
      "learning_rate": 2.861111111111111e-05,
      "loss": 68.5661,
      "step": 9700
    },
    {
      "epoch": 2.16766202167662,
      "grad_norm": 1088.1776123046875,
      "learning_rate": 2.8333333333333335e-05,
      "loss": 67.3807,
      "step": 9800
    },
    {
      "epoch": 2.18978102189781,
      "grad_norm": 511.173828125,
      "learning_rate": 2.8055555555555557e-05,
      "loss": 71.9351,
      "step": 9900
    },
    {
      "epoch": 2.2119000221190004,
      "grad_norm": 372.8301696777344,
      "learning_rate": 2.777777777777778e-05,
      "loss": 116.7822,
      "step": 10000
    },
    {
      "epoch": 2.23401902234019,
      "grad_norm": 673.3489379882812,
      "learning_rate": 2.7500000000000004e-05,
      "loss": 119.5753,
      "step": 10100
    },
    {
      "epoch": 2.2561380225613803,
      "grad_norm": 188.46298217773438,
      "learning_rate": 2.7222222222222223e-05,
      "loss": 77.327,
      "step": 10200
    },
    {
      "epoch": 2.2782570227825705,
      "grad_norm": 1344.5301513671875,
      "learning_rate": 2.6944444444444445e-05,
      "loss": 81.0025,
      "step": 10300
    },
    {
      "epoch": 2.30037602300376,
      "grad_norm": 950.5435791015625,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 77.1479,
      "step": 10400
    },
    {
      "epoch": 2.3224950232249504,
      "grad_norm": 253.42047119140625,
      "learning_rate": 2.6388888888888892e-05,
      "loss": 78.325,
      "step": 10500
    },
    {
      "epoch": 2.34461402344614,
      "grad_norm": 537.2485961914062,
      "learning_rate": 2.6111111111111114e-05,
      "loss": 114.1938,
      "step": 10600
    },
    {
      "epoch": 2.3667330236673303,
      "grad_norm": 391.37054443359375,
      "learning_rate": 2.5833333333333336e-05,
      "loss": 74.7277,
      "step": 10700
    },
    {
      "epoch": 2.3888520238885205,
      "grad_norm": 777.1848754882812,
      "learning_rate": 2.5555555555555554e-05,
      "loss": 87.6657,
      "step": 10800
    },
    {
      "epoch": 2.41097102410971,
      "grad_norm": 233.57789611816406,
      "learning_rate": 2.527777777777778e-05,
      "loss": 74.0286,
      "step": 10900
    },
    {
      "epoch": 2.4330900243309004,
      "grad_norm": 683.5059814453125,
      "learning_rate": 2.5e-05,
      "loss": 120.9458,
      "step": 11000
    },
    {
      "epoch": 2.45520902455209,
      "grad_norm": 495.2901916503906,
      "learning_rate": 2.4722222222222223e-05,
      "loss": 86.7892,
      "step": 11100
    },
    {
      "epoch": 2.4773280247732803,
      "grad_norm": 550.5516967773438,
      "learning_rate": 2.4444444444444445e-05,
      "loss": 89.1761,
      "step": 11200
    },
    {
      "epoch": 2.4994470249944705,
      "grad_norm": 403.9693298339844,
      "learning_rate": 2.4166666666666667e-05,
      "loss": 68.0416,
      "step": 11300
    },
    {
      "epoch": 2.52156602521566,
      "grad_norm": 156.51942443847656,
      "learning_rate": 2.3888888888888892e-05,
      "loss": 102.727,
      "step": 11400
    },
    {
      "epoch": 2.5436850254368504,
      "grad_norm": 1048.75439453125,
      "learning_rate": 2.361111111111111e-05,
      "loss": 101.3001,
      "step": 11500
    },
    {
      "epoch": 2.56580402565804,
      "grad_norm": 1137.298828125,
      "learning_rate": 2.3333333333333336e-05,
      "loss": 80.9905,
      "step": 11600
    },
    {
      "epoch": 2.5879230258792303,
      "grad_norm": 600.23681640625,
      "learning_rate": 2.3055555555555558e-05,
      "loss": 68.7155,
      "step": 11700
    },
    {
      "epoch": 2.6100420261004205,
      "grad_norm": 840.565185546875,
      "learning_rate": 2.277777777777778e-05,
      "loss": 97.0005,
      "step": 11800
    },
    {
      "epoch": 2.63216102632161,
      "grad_norm": 722.6634521484375,
      "learning_rate": 2.25e-05,
      "loss": 43.6906,
      "step": 11900
    },
    {
      "epoch": 2.6542800265428004,
      "grad_norm": 399.0193176269531,
      "learning_rate": 2.2222222222222223e-05,
      "loss": 87.6618,
      "step": 12000
    },
    {
      "epoch": 2.67639902676399,
      "grad_norm": 1078.0657958984375,
      "learning_rate": 2.1944444444444445e-05,
      "loss": 90.9304,
      "step": 12100
    },
    {
      "epoch": 2.6985180269851803,
      "grad_norm": 719.8499145507812,
      "learning_rate": 2.1666666666666667e-05,
      "loss": 60.1098,
      "step": 12200
    },
    {
      "epoch": 2.7206370272063705,
      "grad_norm": 2423.7578125,
      "learning_rate": 2.138888888888889e-05,
      "loss": 80.944,
      "step": 12300
    },
    {
      "epoch": 2.74275602742756,
      "grad_norm": 1538.58447265625,
      "learning_rate": 2.111111111111111e-05,
      "loss": 84.306,
      "step": 12400
    },
    {
      "epoch": 2.7648750276487504,
      "grad_norm": 1219.7099609375,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 61.9064,
      "step": 12500
    },
    {
      "epoch": 2.78699402786994,
      "grad_norm": 419.0851135253906,
      "learning_rate": 2.0555555555555555e-05,
      "loss": 58.863,
      "step": 12600
    },
    {
      "epoch": 2.8091130280911303,
      "grad_norm": 10715.943359375,
      "learning_rate": 2.027777777777778e-05,
      "loss": 74.7694,
      "step": 12700
    },
    {
      "epoch": 2.8312320283123205,
      "grad_norm": 201.3819122314453,
      "learning_rate": 2e-05,
      "loss": 83.1654,
      "step": 12800
    },
    {
      "epoch": 2.85335102853351,
      "grad_norm": 3191.744873046875,
      "learning_rate": 1.9722222222222224e-05,
      "loss": 79.8109,
      "step": 12900
    },
    {
      "epoch": 2.8754700287547004,
      "grad_norm": 284.8264465332031,
      "learning_rate": 1.9444444444444445e-05,
      "loss": 91.2385,
      "step": 13000
    },
    {
      "epoch": 2.89758902897589,
      "grad_norm": 355.34674072265625,
      "learning_rate": 1.9166666666666667e-05,
      "loss": 53.8721,
      "step": 13100
    },
    {
      "epoch": 2.9197080291970803,
      "grad_norm": 760.095703125,
      "learning_rate": 1.888888888888889e-05,
      "loss": 70.6495,
      "step": 13200
    },
    {
      "epoch": 2.9418270294182705,
      "grad_norm": 163.60043334960938,
      "learning_rate": 1.861111111111111e-05,
      "loss": 61.0286,
      "step": 13300
    },
    {
      "epoch": 2.96394602963946,
      "grad_norm": 694.377197265625,
      "learning_rate": 1.8333333333333333e-05,
      "loss": 70.8661,
      "step": 13400
    },
    {
      "epoch": 2.9860650298606504,
      "grad_norm": 789.5768432617188,
      "learning_rate": 1.8055555555555555e-05,
      "loss": 51.5301,
      "step": 13500
    },
    {
      "epoch": 3.0,
      "eval_loss": 166.22483825683594,
      "eval_runtime": 20.711,
      "eval_samples_per_second": 97.05,
      "eval_steps_per_second": 24.287,
      "step": 13563
    },
    {
      "epoch": 3.00818403008184,
      "grad_norm": 588.47265625,
      "learning_rate": 1.777777777777778e-05,
      "loss": 60.6116,
      "step": 13600
    },
    {
      "epoch": 3.0303030303030303,
      "grad_norm": 637.5720825195312,
      "learning_rate": 1.75e-05,
      "loss": 75.5694,
      "step": 13700
    },
    {
      "epoch": 3.0524220305242205,
      "grad_norm": 271.2422180175781,
      "learning_rate": 1.7222222222222224e-05,
      "loss": 67.7535,
      "step": 13800
    },
    {
      "epoch": 3.07454103074541,
      "grad_norm": 1230.089111328125,
      "learning_rate": 1.6944444444444446e-05,
      "loss": 84.6008,
      "step": 13900
    },
    {
      "epoch": 3.0966600309666004,
      "grad_norm": 493.8188781738281,
      "learning_rate": 1.6666666666666667e-05,
      "loss": 57.6848,
      "step": 14000
    },
    {
      "epoch": 3.11877903118779,
      "grad_norm": 927.6559448242188,
      "learning_rate": 1.638888888888889e-05,
      "loss": 77.8659,
      "step": 14100
    },
    {
      "epoch": 3.1408980314089803,
      "grad_norm": 1547.560302734375,
      "learning_rate": 1.6111111111111115e-05,
      "loss": 56.6124,
      "step": 14200
    },
    {
      "epoch": 3.1630170316301705,
      "grad_norm": 1353.518798828125,
      "learning_rate": 1.5833333333333333e-05,
      "loss": 52.2098,
      "step": 14300
    },
    {
      "epoch": 3.18513603185136,
      "grad_norm": 20034.76171875,
      "learning_rate": 1.5555555555555555e-05,
      "loss": 47.6291,
      "step": 14400
    },
    {
      "epoch": 3.2072550320725504,
      "grad_norm": 295.83868408203125,
      "learning_rate": 1.527777777777778e-05,
      "loss": 76.8827,
      "step": 14500
    },
    {
      "epoch": 3.22937403229374,
      "grad_norm": 1203.1685791015625,
      "learning_rate": 1.5e-05,
      "loss": 75.1167,
      "step": 14600
    },
    {
      "epoch": 3.2514930325149303,
      "grad_norm": 514.1155395507812,
      "learning_rate": 1.4722222222222224e-05,
      "loss": 56.7415,
      "step": 14700
    },
    {
      "epoch": 3.2736120327361204,
      "grad_norm": 254.90188598632812,
      "learning_rate": 1.4444444444444444e-05,
      "loss": 65.6232,
      "step": 14800
    },
    {
      "epoch": 3.29573103295731,
      "grad_norm": 2475.375,
      "learning_rate": 1.4166666666666668e-05,
      "loss": 82.0212,
      "step": 14900
    },
    {
      "epoch": 3.3178500331785004,
      "grad_norm": 695.0991821289062,
      "learning_rate": 1.388888888888889e-05,
      "loss": 70.9937,
      "step": 15000
    },
    {
      "epoch": 3.33996903339969,
      "grad_norm": 3679.786376953125,
      "learning_rate": 1.3611111111111111e-05,
      "loss": 87.9886,
      "step": 15100
    },
    {
      "epoch": 3.3620880336208803,
      "grad_norm": 335.57244873046875,
      "learning_rate": 1.3333333333333333e-05,
      "loss": 81.2132,
      "step": 15200
    },
    {
      "epoch": 3.3842070338420704,
      "grad_norm": 3056.2080078125,
      "learning_rate": 1.3055555555555557e-05,
      "loss": 84.2817,
      "step": 15300
    },
    {
      "epoch": 3.40632603406326,
      "grad_norm": 633.110107421875,
      "learning_rate": 1.2777777777777777e-05,
      "loss": 91.3568,
      "step": 15400
    },
    {
      "epoch": 3.4284450342844504,
      "grad_norm": 1400.1513671875,
      "learning_rate": 1.25e-05,
      "loss": 64.4974,
      "step": 15500
    },
    {
      "epoch": 3.4505640345056405,
      "grad_norm": 366.1932678222656,
      "learning_rate": 1.2222222222222222e-05,
      "loss": 65.3457,
      "step": 15600
    },
    {
      "epoch": 3.4726830347268303,
      "grad_norm": 77.0812759399414,
      "learning_rate": 1.1944444444444446e-05,
      "loss": 94.2989,
      "step": 15700
    },
    {
      "epoch": 3.4948020349480204,
      "grad_norm": 517.035888671875,
      "learning_rate": 1.1666666666666668e-05,
      "loss": 54.1213,
      "step": 15800
    },
    {
      "epoch": 3.5169210351692106,
      "grad_norm": 582.3343505859375,
      "learning_rate": 1.138888888888889e-05,
      "loss": 80.1348,
      "step": 15900
    },
    {
      "epoch": 3.5390400353904004,
      "grad_norm": 134.23880004882812,
      "learning_rate": 1.1111111111111112e-05,
      "loss": 76.6385,
      "step": 16000
    },
    {
      "epoch": 3.56115903561159,
      "grad_norm": 341.63604736328125,
      "learning_rate": 1.0833333333333334e-05,
      "loss": 78.0171,
      "step": 16100
    },
    {
      "epoch": 3.5832780358327803,
      "grad_norm": 1244.135009765625,
      "learning_rate": 1.0555555555555555e-05,
      "loss": 51.3064,
      "step": 16200
    },
    {
      "epoch": 3.6053970360539704,
      "grad_norm": 916.8475952148438,
      "learning_rate": 1.0277777777777777e-05,
      "loss": 79.2429,
      "step": 16300
    },
    {
      "epoch": 3.6275160362751606,
      "grad_norm": 269.9622802734375,
      "learning_rate": 1e-05,
      "loss": 68.2658,
      "step": 16400
    },
    {
      "epoch": 3.6496350364963503,
      "grad_norm": 272.8192138671875,
      "learning_rate": 9.722222222222223e-06,
      "loss": 54.9957,
      "step": 16500
    },
    {
      "epoch": 3.6717540367175405,
      "grad_norm": 529.631591796875,
      "learning_rate": 9.444444444444445e-06,
      "loss": 55.4425,
      "step": 16600
    },
    {
      "epoch": 3.6938730369387303,
      "grad_norm": 478.92327880859375,
      "learning_rate": 9.166666666666666e-06,
      "loss": 49.7692,
      "step": 16700
    },
    {
      "epoch": 3.7159920371599204,
      "grad_norm": 499.35406494140625,
      "learning_rate": 8.88888888888889e-06,
      "loss": 46.7543,
      "step": 16800
    },
    {
      "epoch": 3.7381110373811106,
      "grad_norm": 179.26344299316406,
      "learning_rate": 8.611111111111112e-06,
      "loss": 64.0389,
      "step": 16900
    },
    {
      "epoch": 3.7602300376023003,
      "grad_norm": 1287.7215576171875,
      "learning_rate": 8.333333333333334e-06,
      "loss": 59.5381,
      "step": 17000
    },
    {
      "epoch": 3.7823490378234905,
      "grad_norm": 916.1105346679688,
      "learning_rate": 8.055555555555557e-06,
      "loss": 60.6414,
      "step": 17100
    },
    {
      "epoch": 3.8044680380446803,
      "grad_norm": 398.42724609375,
      "learning_rate": 7.777777777777777e-06,
      "loss": 56.4996,
      "step": 17200
    },
    {
      "epoch": 3.8265870382658704,
      "grad_norm": 466.2370910644531,
      "learning_rate": 7.5e-06,
      "loss": 55.1325,
      "step": 17300
    },
    {
      "epoch": 3.8487060384870606,
      "grad_norm": 426.53271484375,
      "learning_rate": 7.222222222222222e-06,
      "loss": 74.4013,
      "step": 17400
    },
    {
      "epoch": 3.8708250387082503,
      "grad_norm": 595.925537109375,
      "learning_rate": 6.944444444444445e-06,
      "loss": 65.174,
      "step": 17500
    },
    {
      "epoch": 3.8929440389294405,
      "grad_norm": 960.3365478515625,
      "learning_rate": 6.666666666666667e-06,
      "loss": 61.204,
      "step": 17600
    },
    {
      "epoch": 3.9150630391506303,
      "grad_norm": 488.77978515625,
      "learning_rate": 6.3888888888888885e-06,
      "loss": 64.5583,
      "step": 17700
    },
    {
      "epoch": 3.9371820393718204,
      "grad_norm": 386.27911376953125,
      "learning_rate": 6.111111111111111e-06,
      "loss": 53.6574,
      "step": 17800
    },
    {
      "epoch": 3.9593010395930106,
      "grad_norm": 741.3096313476562,
      "learning_rate": 5.833333333333334e-06,
      "loss": 56.0408,
      "step": 17900
    },
    {
      "epoch": 3.9814200398142003,
      "grad_norm": 125.60317993164062,
      "learning_rate": 5.555555555555556e-06,
      "loss": 51.7872,
      "step": 18000
    },
    {
      "epoch": 4.0,
      "eval_loss": 174.72474670410156,
      "eval_runtime": 20.9561,
      "eval_samples_per_second": 95.915,
      "eval_steps_per_second": 24.003,
      "step": 18084
    },
    {
      "epoch": 4.00353904003539,
      "grad_norm": 206.36895751953125,
      "learning_rate": 5.277777777777778e-06,
      "loss": 40.9699,
      "step": 18100
    },
    {
      "epoch": 4.02565804025658,
      "grad_norm": 399.98455810546875,
      "learning_rate": 5e-06,
      "loss": 56.6913,
      "step": 18200
    },
    {
      "epoch": 4.04777704047777,
      "grad_norm": 1047.0919189453125,
      "learning_rate": 4.722222222222222e-06,
      "loss": 48.5654,
      "step": 18300
    },
    {
      "epoch": 4.069896040698961,
      "grad_norm": 882.0901489257812,
      "learning_rate": 4.444444444444445e-06,
      "loss": 47.7457,
      "step": 18400
    },
    {
      "epoch": 4.092015040920151,
      "grad_norm": 515.3323364257812,
      "learning_rate": 4.166666666666667e-06,
      "loss": 66.938,
      "step": 18500
    },
    {
      "epoch": 4.11413404114134,
      "grad_norm": 1115.885009765625,
      "learning_rate": 3.888888888888889e-06,
      "loss": 58.0705,
      "step": 18600
    },
    {
      "epoch": 4.13625304136253,
      "grad_norm": 672.346435546875,
      "learning_rate": 3.611111111111111e-06,
      "loss": 68.3134,
      "step": 18700
    },
    {
      "epoch": 4.15837204158372,
      "grad_norm": 592.4734497070312,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 55.2237,
      "step": 18800
    },
    {
      "epoch": 4.180491041804911,
      "grad_norm": 642.23583984375,
      "learning_rate": 3.0555555555555556e-06,
      "loss": 77.2789,
      "step": 18900
    },
    {
      "epoch": 4.202610042026101,
      "grad_norm": 595.8189086914062,
      "learning_rate": 2.777777777777778e-06,
      "loss": 54.4822,
      "step": 19000
    },
    {
      "epoch": 4.22472904224729,
      "grad_norm": 440.1129150390625,
      "learning_rate": 2.5e-06,
      "loss": 50.1236,
      "step": 19100
    },
    {
      "epoch": 4.24684804246848,
      "grad_norm": 8661.6494140625,
      "learning_rate": 2.2222222222222225e-06,
      "loss": 81.7971,
      "step": 19200
    },
    {
      "epoch": 4.26896704268967,
      "grad_norm": 214.70748901367188,
      "learning_rate": 1.9444444444444444e-06,
      "loss": 51.6579,
      "step": 19300
    },
    {
      "epoch": 4.291086042910861,
      "grad_norm": 724.3253173828125,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 68.406,
      "step": 19400
    },
    {
      "epoch": 4.313205043132051,
      "grad_norm": 1145.943115234375,
      "learning_rate": 1.388888888888889e-06,
      "loss": 62.2609,
      "step": 19500
    },
    {
      "epoch": 4.33532404335324,
      "grad_norm": 1727.2354736328125,
      "learning_rate": 1.1111111111111112e-06,
      "loss": 67.2163,
      "step": 19600
    },
    {
      "epoch": 4.35744304357443,
      "grad_norm": 734.8032836914062,
      "learning_rate": 8.333333333333333e-07,
      "loss": 57.1625,
      "step": 19700
    },
    {
      "epoch": 4.37956204379562,
      "grad_norm": 1056.7657470703125,
      "learning_rate": 5.555555555555556e-07,
      "loss": 61.3986,
      "step": 19800
    },
    {
      "epoch": 4.401681044016811,
      "grad_norm": 550.4881591796875,
      "learning_rate": 2.777777777777778e-07,
      "loss": 60.5436,
      "step": 19900
    },
    {
      "epoch": 4.423800044238001,
      "grad_norm": 435.5983581542969,
      "learning_rate": 0.0,
      "loss": 60.0424,
      "step": 20000
    }
  ],
  "logging_steps": 100,
  "max_steps": 20000,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 5,
  "save_steps": 10000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}