{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.004,
      "grad_norm": 115.85907745361328,
      "learning_rate": 4.997995991983968e-05,
      "loss": 3.7476,
      "step": 10
    },
    {
      "epoch": 0.008,
      "grad_norm": 66.54586029052734,
      "learning_rate": 4.977955911823648e-05,
      "loss": 2.8414,
      "step": 20
    },
    {
      "epoch": 0.012,
      "grad_norm": 55.67858123779297,
      "learning_rate": 4.957915831663327e-05,
      "loss": 2.3755,
      "step": 30
    },
    {
      "epoch": 0.016,
      "grad_norm": 49.31084060668945,
      "learning_rate": 4.937875751503006e-05,
      "loss": 1.9815,
      "step": 40
    },
    {
      "epoch": 0.02,
      "grad_norm": 50.2120475769043,
      "learning_rate": 4.917835671342685e-05,
      "loss": 2.2454,
      "step": 50
    },
    {
      "epoch": 0.024,
      "grad_norm": 33.266048431396484,
      "learning_rate": 4.897795591182365e-05,
      "loss": 2.4204,
      "step": 60
    },
    {
      "epoch": 0.028,
      "grad_norm": 25.996694564819336,
      "learning_rate": 4.877755511022044e-05,
      "loss": 1.4711,
      "step": 70
    },
    {
      "epoch": 0.032,
      "grad_norm": 43.97441864013672,
      "learning_rate": 4.8577154308617234e-05,
      "loss": 1.697,
      "step": 80
    },
    {
      "epoch": 0.036,
      "grad_norm": 37.83395767211914,
      "learning_rate": 4.8376753507014026e-05,
      "loss": 1.7633,
      "step": 90
    },
    {
      "epoch": 0.04,
      "grad_norm": 26.54341697692871,
      "learning_rate": 4.8176352705410824e-05,
      "loss": 1.3992,
      "step": 100
    },
    {
      "epoch": 0.044,
      "grad_norm": 42.20634841918945,
      "learning_rate": 4.797595190380762e-05,
      "loss": 1.6443,
      "step": 110
    },
    {
      "epoch": 0.048,
      "grad_norm": 24.657821655273438,
      "learning_rate": 4.7775551102204415e-05,
      "loss": 1.8471,
      "step": 120
    },
    {
      "epoch": 0.052,
      "grad_norm": 38.727420806884766,
      "learning_rate": 4.7575150300601207e-05,
      "loss": 1.5293,
      "step": 130
    },
    {
      "epoch": 0.056,
      "grad_norm": 31.97869873046875,
      "learning_rate": 4.7374749498998e-05,
      "loss": 1.6212,
      "step": 140
    },
    {
      "epoch": 0.06,
      "grad_norm": 31.056962966918945,
      "learning_rate": 4.717434869739479e-05,
      "loss": 1.4407,
      "step": 150
    },
    {
      "epoch": 0.064,
      "grad_norm": 29.63347053527832,
      "learning_rate": 4.697394789579159e-05,
      "loss": 1.1833,
      "step": 160
    },
    {
      "epoch": 0.068,
      "grad_norm": 44.844268798828125,
      "learning_rate": 4.677354709418838e-05,
      "loss": 1.5756,
      "step": 170
    },
    {
      "epoch": 0.072,
      "grad_norm": 31.4070987701416,
      "learning_rate": 4.657314629258517e-05,
      "loss": 1.4358,
      "step": 180
    },
    {
      "epoch": 0.076,
      "grad_norm": 26.982776641845703,
      "learning_rate": 4.6372745490981964e-05,
      "loss": 1.234,
      "step": 190
    },
    {
      "epoch": 0.08,
      "grad_norm": 19.802730560302734,
      "learning_rate": 4.617234468937876e-05,
      "loss": 1.3504,
      "step": 200
    },
    {
      "epoch": 0.084,
      "grad_norm": 33.88198471069336,
      "learning_rate": 4.5971943887775554e-05,
      "loss": 1.5029,
      "step": 210
    },
    {
      "epoch": 0.088,
      "grad_norm": 24.533716201782227,
      "learning_rate": 4.5771543086172346e-05,
      "loss": 1.2978,
      "step": 220
    },
    {
      "epoch": 0.092,
      "grad_norm": 27.563339233398438,
      "learning_rate": 4.557114228456914e-05,
      "loss": 1.7014,
      "step": 230
    },
    {
      "epoch": 0.096,
      "grad_norm": 29.428752899169922,
      "learning_rate": 4.5370741482965936e-05,
      "loss": 1.3845,
      "step": 240
    },
    {
      "epoch": 0.1,
      "grad_norm": 20.272520065307617,
      "learning_rate": 4.517034068136273e-05,
      "loss": 1.2192,
      "step": 250
    },
    {
      "epoch": 0.104,
      "grad_norm": 47.12469482421875,
      "learning_rate": 4.496993987975952e-05,
      "loss": 1.2814,
      "step": 260
    },
    {
      "epoch": 0.108,
      "grad_norm": 18.20330238342285,
      "learning_rate": 4.476953907815631e-05,
      "loss": 1.2717,
      "step": 270
    },
    {
      "epoch": 0.112,
      "grad_norm": 39.07451248168945,
      "learning_rate": 4.456913827655311e-05,
      "loss": 1.3291,
      "step": 280
    },
    {
      "epoch": 0.116,
      "grad_norm": 50.37272644042969,
      "learning_rate": 4.43687374749499e-05,
      "loss": 1.3691,
      "step": 290
    },
    {
      "epoch": 0.12,
      "grad_norm": 23.233367919921875,
      "learning_rate": 4.4168336673346694e-05,
      "loss": 1.4183,
      "step": 300
    },
    {
      "epoch": 0.124,
      "grad_norm": 22.46800422668457,
      "learning_rate": 4.3967935871743486e-05,
      "loss": 1.1226,
      "step": 310
    },
    {
      "epoch": 0.128,
      "grad_norm": 24.424856185913086,
      "learning_rate": 4.3767535070140284e-05,
      "loss": 1.3413,
      "step": 320
    },
    {
      "epoch": 0.132,
      "grad_norm": 14.698283195495605,
      "learning_rate": 4.3567134268537076e-05,
      "loss": 1.2009,
      "step": 330
    },
    {
      "epoch": 0.136,
      "grad_norm": 23.5820369720459,
      "learning_rate": 4.336673346693387e-05,
      "loss": 1.0799,
      "step": 340
    },
    {
      "epoch": 0.14,
      "grad_norm": 26.510631561279297,
      "learning_rate": 4.316633266533066e-05,
      "loss": 1.0801,
      "step": 350
    },
    {
      "epoch": 0.144,
      "grad_norm": 18.498275756835938,
      "learning_rate": 4.296593186372745e-05,
      "loss": 1.1631,
      "step": 360
    },
    {
      "epoch": 0.148,
      "grad_norm": 35.2937126159668,
      "learning_rate": 4.2765531062124256e-05,
      "loss": 1.3777,
      "step": 370
    },
    {
      "epoch": 0.152,
      "grad_norm": 40.05356216430664,
      "learning_rate": 4.256513026052105e-05,
      "loss": 1.1189,
      "step": 380
    },
    {
      "epoch": 0.156,
      "grad_norm": 18.918344497680664,
      "learning_rate": 4.236472945891784e-05,
      "loss": 1.0721,
      "step": 390
    },
    {
      "epoch": 0.16,
      "grad_norm": 20.29583168029785,
      "learning_rate": 4.216432865731463e-05,
      "loss": 0.9337,
      "step": 400
    },
    {
      "epoch": 0.164,
      "grad_norm": 19.447803497314453,
      "learning_rate": 4.2081993569131834e-05,
      "loss": 1.2047,
      "step": 410
    },
    {
      "epoch": 0.168,
      "grad_norm": 26.82716178894043,
      "learning_rate": 4.188102893890675e-05,
      "loss": 1.1294,
      "step": 420
    },
    {
      "epoch": 0.172,
      "grad_norm": 12.995594024658203,
      "learning_rate": 4.168006430868168e-05,
      "loss": 1.0033,
      "step": 430
    },
    {
      "epoch": 0.176,
      "grad_norm": 21.796598434448242,
      "learning_rate": 4.14790996784566e-05,
      "loss": 0.8864,
      "step": 440
    },
    {
      "epoch": 0.18,
      "grad_norm": 13.911988258361816,
      "learning_rate": 4.1278135048231516e-05,
      "loss": 0.8974,
      "step": 450
    },
    {
      "epoch": 0.184,
      "grad_norm": 25.945011138916016,
      "learning_rate": 4.1077170418006434e-05,
      "loss": 1.261,
      "step": 460
    },
    {
      "epoch": 0.188,
      "grad_norm": 19.943857192993164,
      "learning_rate": 4.087620578778135e-05,
      "loss": 1.0262,
      "step": 470
    },
    {
      "epoch": 0.192,
      "grad_norm": 23.558696746826172,
      "learning_rate": 4.067524115755627e-05,
      "loss": 0.9572,
      "step": 480
    },
    {
      "epoch": 0.196,
      "grad_norm": 42.70231628417969,
      "learning_rate": 4.047427652733119e-05,
      "loss": 1.044,
      "step": 490
    },
    {
      "epoch": 0.2,
      "grad_norm": 16.41856575012207,
      "learning_rate": 4.027331189710611e-05,
      "loss": 1.0562,
      "step": 500
    },
    {
      "epoch": 0.204,
      "grad_norm": 13.945029258728027,
      "learning_rate": 4.0072347266881035e-05,
      "loss": 0.9637,
      "step": 510
    },
    {
      "epoch": 0.208,
      "grad_norm": 16.55429458618164,
      "learning_rate": 3.9871382636655953e-05,
      "loss": 0.8104,
      "step": 520
    },
    {
      "epoch": 0.212,
      "grad_norm": 24.434778213500977,
      "learning_rate": 3.967041800643087e-05,
      "loss": 1.1942,
      "step": 530
    },
    {
      "epoch": 0.216,
      "grad_norm": 20.01283836364746,
      "learning_rate": 3.946945337620579e-05,
      "loss": 0.9209,
      "step": 540
    },
    {
      "epoch": 0.22,
      "grad_norm": 18.98524284362793,
      "learning_rate": 3.926848874598071e-05,
      "loss": 0.7445,
      "step": 550
    },
    {
      "epoch": 0.224,
      "grad_norm": 22.44414710998535,
      "learning_rate": 3.906752411575563e-05,
      "loss": 0.9528,
      "step": 560
    },
    {
      "epoch": 0.228,
      "grad_norm": 19.79057502746582,
      "learning_rate": 3.886655948553055e-05,
      "loss": 0.9216,
      "step": 570
    },
    {
      "epoch": 0.232,
      "grad_norm": 17.453460693359375,
      "learning_rate": 3.866559485530547e-05,
      "loss": 0.9834,
      "step": 580
    },
    {
      "epoch": 0.236,
      "grad_norm": 29.218969345092773,
      "learning_rate": 3.846463022508039e-05,
      "loss": 0.9945,
      "step": 590
    },
    {
      "epoch": 0.24,
      "grad_norm": 17.652963638305664,
      "learning_rate": 3.826366559485531e-05,
      "loss": 0.9308,
      "step": 600
    },
    {
      "epoch": 0.244,
      "grad_norm": 20.76468849182129,
      "learning_rate": 3.806270096463023e-05,
      "loss": 0.8735,
      "step": 610
    },
    {
      "epoch": 0.248,
      "grad_norm": 26.41815757751465,
      "learning_rate": 3.786173633440515e-05,
      "loss": 1.2878,
      "step": 620
    },
    {
      "epoch": 0.252,
      "grad_norm": 41.02421951293945,
      "learning_rate": 3.7660771704180066e-05,
      "loss": 1.1299,
      "step": 630
    },
    {
      "epoch": 0.256,
      "grad_norm": 17.34744644165039,
      "learning_rate": 3.7459807073954985e-05,
      "loss": 0.8315,
      "step": 640
    },
    {
      "epoch": 0.26,
      "grad_norm": 14.293941497802734,
      "learning_rate": 3.725884244372991e-05,
      "loss": 0.8405,
      "step": 650
    },
    {
      "epoch": 0.264,
      "grad_norm": 15.149956703186035,
      "learning_rate": 3.705787781350483e-05,
      "loss": 1.0297,
      "step": 660
    },
    {
      "epoch": 0.268,
      "grad_norm": 17.754810333251953,
      "learning_rate": 3.685691318327975e-05,
      "loss": 0.9322,
      "step": 670
    },
    {
      "epoch": 0.272,
      "grad_norm": 21.743669509887695,
      "learning_rate": 3.6655948553054666e-05,
      "loss": 0.9069,
      "step": 680
    },
    {
      "epoch": 0.276,
      "grad_norm": 29.161598205566406,
      "learning_rate": 3.6454983922829585e-05,
      "loss": 0.8633,
      "step": 690
    },
    {
      "epoch": 0.28,
      "grad_norm": 16.16539192199707,
      "learning_rate": 3.6254019292604503e-05,
      "loss": 0.7662,
      "step": 700
    },
    {
      "epoch": 0.284,
      "grad_norm": 25.36922264099121,
      "learning_rate": 3.605305466237942e-05,
      "loss": 0.7945,
      "step": 710
    },
    {
      "epoch": 0.288,
      "grad_norm": 24.251853942871094,
      "learning_rate": 3.585209003215435e-05,
      "loss": 0.9693,
      "step": 720
    },
    {
      "epoch": 0.292,
      "grad_norm": 15.235057830810547,
      "learning_rate": 3.5651125401929266e-05,
      "loss": 0.8969,
      "step": 730
    },
    {
      "epoch": 0.296,
      "grad_norm": 14.464040756225586,
      "learning_rate": 3.5450160771704185e-05,
      "loss": 0.7205,
      "step": 740
    },
    {
      "epoch": 0.3,
      "grad_norm": 23.044424057006836,
      "learning_rate": 3.5249196141479104e-05,
      "loss": 0.839,
      "step": 750
    },
    {
      "epoch": 0.304,
      "grad_norm": 25.620925903320312,
      "learning_rate": 3.504823151125402e-05,
      "loss": 0.8887,
      "step": 760
    },
    {
      "epoch": 0.308,
      "grad_norm": 10.347396850585938,
      "learning_rate": 3.484726688102894e-05,
      "loss": 0.645,
      "step": 770
    },
    {
      "epoch": 0.312,
      "grad_norm": 19.114471435546875,
      "learning_rate": 3.464630225080386e-05,
      "loss": 0.8341,
      "step": 780
    },
    {
      "epoch": 0.316,
      "grad_norm": 17.528043746948242,
      "learning_rate": 3.4445337620578785e-05,
      "loss": 0.7108,
      "step": 790
    },
    {
      "epoch": 0.32,
      "grad_norm": 13.186959266662598,
      "learning_rate": 3.4244372990353704e-05,
      "loss": 0.7991,
      "step": 800
    },
    {
      "epoch": 0.324,
      "grad_norm": 25.02106475830078,
      "learning_rate": 3.404340836012862e-05,
      "loss": 0.7284,
      "step": 810
    },
    {
      "epoch": 0.328,
      "grad_norm": 14.035198211669922,
      "learning_rate": 3.384244372990354e-05,
      "loss": 0.7589,
      "step": 820
    },
    {
      "epoch": 0.332,
      "grad_norm": 11.368013381958008,
      "learning_rate": 3.364147909967846e-05,
      "loss": 0.7638,
      "step": 830
    },
    {
      "epoch": 0.336,
      "grad_norm": 21.951080322265625,
      "learning_rate": 3.344051446945338e-05,
      "loss": 0.7869,
      "step": 840
    },
    {
      "epoch": 0.34,
      "grad_norm": 17.966073989868164,
      "learning_rate": 3.32395498392283e-05,
      "loss": 0.6792,
      "step": 850
    },
    {
      "epoch": 0.344,
      "grad_norm": 36.02198791503906,
      "learning_rate": 3.3038585209003216e-05,
      "loss": 0.6968,
      "step": 860
    },
    {
      "epoch": 0.348,
      "grad_norm": 32.43560791015625,
      "learning_rate": 3.283762057877814e-05,
      "loss": 0.7523,
      "step": 870
    },
    {
      "epoch": 0.352,
      "grad_norm": 30.29490852355957,
      "learning_rate": 3.263665594855306e-05,
      "loss": 0.6548,
      "step": 880
    },
    {
      "epoch": 0.356,
      "grad_norm": 8.957921981811523,
      "learning_rate": 3.243569131832798e-05,
      "loss": 0.7151,
      "step": 890
    },
    {
      "epoch": 0.36,
      "grad_norm": 15.583487510681152,
      "learning_rate": 3.22347266881029e-05,
      "loss": 0.625,
      "step": 900
    },
    {
      "epoch": 0.364,
      "grad_norm": 23.470478057861328,
      "learning_rate": 3.2033762057877816e-05,
      "loss": 0.9571,
      "step": 910
    },
    {
      "epoch": 0.368,
      "grad_norm": 31.515092849731445,
      "learning_rate": 3.1832797427652735e-05,
      "loss": 0.7395,
      "step": 920
    },
    {
      "epoch": 0.372,
      "grad_norm": 14.246073722839355,
      "learning_rate": 3.1631832797427654e-05,
      "loss": 0.6884,
      "step": 930
    },
    {
      "epoch": 0.376,
      "grad_norm": 25.352590560913086,
      "learning_rate": 3.143086816720258e-05,
      "loss": 0.772,
      "step": 940
    },
    {
      "epoch": 0.38,
      "grad_norm": 14.441354751586914,
      "learning_rate": 3.12299035369775e-05,
      "loss": 0.7406,
      "step": 950
    },
    {
      "epoch": 0.384,
      "grad_norm": 29.33234405517578,
      "learning_rate": 3.102893890675242e-05,
      "loss": 0.7242,
      "step": 960
    },
    {
      "epoch": 0.388,
      "grad_norm": 16.018104553222656,
      "learning_rate": 3.0827974276527335e-05,
      "loss": 0.9183,
      "step": 970
    },
    {
      "epoch": 0.392,
      "grad_norm": 14.766180992126465,
      "learning_rate": 3.0627009646302254e-05,
      "loss": 0.8971,
      "step": 980
    },
    {
      "epoch": 0.396,
      "grad_norm": 10.733450889587402,
      "learning_rate": 3.042604501607717e-05,
      "loss": 0.5546,
      "step": 990
    },
    {
      "epoch": 0.4,
      "grad_norm": 15.318602561950684,
      "learning_rate": 3.0225080385852088e-05,
      "loss": 0.7582,
      "step": 1000
    },
    {
      "epoch": 0.404,
      "grad_norm": 16.220378875732422,
      "learning_rate": 3.0024115755627013e-05,
      "loss": 0.7034,
      "step": 1010
    },
    {
      "epoch": 0.408,
      "grad_norm": 24.005294799804688,
      "learning_rate": 2.9823151125401932e-05,
      "loss": 0.692,
      "step": 1020
    },
    {
      "epoch": 0.412,
      "grad_norm": 10.050074577331543,
      "learning_rate": 2.962218649517685e-05,
      "loss": 0.6043,
      "step": 1030
    },
    {
      "epoch": 0.416,
      "grad_norm": 14.405035972595215,
      "learning_rate": 2.942122186495177e-05,
      "loss": 0.779,
      "step": 1040
    },
    {
      "epoch": 0.42,
      "grad_norm": 21.141399383544922,
      "learning_rate": 2.9220257234726688e-05,
      "loss": 0.6058,
      "step": 1050
    },
    {
      "epoch": 0.424,
      "grad_norm": 12.982873916625977,
      "learning_rate": 2.9019292604501607e-05,
      "loss": 0.6479,
      "step": 1060
    },
    {
      "epoch": 0.428,
      "grad_norm": 11.213690757751465,
      "learning_rate": 2.8818327974276526e-05,
      "loss": 0.5357,
      "step": 1070
    },
    {
      "epoch": 0.432,
      "grad_norm": 9.400264739990234,
      "learning_rate": 2.861736334405145e-05,
      "loss": 0.7004,
      "step": 1080
    },
    {
      "epoch": 0.436,
      "grad_norm": 24.078365325927734,
      "learning_rate": 2.841639871382637e-05,
      "loss": 0.574,
      "step": 1090
    },
    {
      "epoch": 0.44,
      "grad_norm": 10.422481536865234,
      "learning_rate": 2.821543408360129e-05,
      "loss": 0.5758,
      "step": 1100
    },
    {
      "epoch": 0.444,
      "grad_norm": 14.57651424407959,
      "learning_rate": 2.8014469453376207e-05,
      "loss": 0.6732,
      "step": 1110
    },
    {
      "epoch": 0.448,
      "grad_norm": 11.885628700256348,
      "learning_rate": 2.7813504823151126e-05,
      "loss": 0.5472,
      "step": 1120
    },
    {
      "epoch": 0.452,
      "grad_norm": 14.696109771728516,
      "learning_rate": 2.7612540192926045e-05,
      "loss": 0.6875,
      "step": 1130
    },
    {
      "epoch": 0.456,
      "grad_norm": 10.769063949584961,
      "learning_rate": 2.7411575562700963e-05,
      "loss": 0.63,
      "step": 1140
    },
    {
      "epoch": 0.46,
      "grad_norm": 14.238389015197754,
      "learning_rate": 2.7210610932475882e-05,
      "loss": 0.6574,
      "step": 1150
    },
    {
      "epoch": 0.464,
      "grad_norm": 10.263813972473145,
      "learning_rate": 2.7009646302250807e-05,
      "loss": 0.6817,
      "step": 1160
    },
    {
      "epoch": 0.468,
      "grad_norm": 18.851377487182617,
      "learning_rate": 2.6808681672025726e-05,
      "loss": 0.49,
      "step": 1170
    },
    {
      "epoch": 0.472,
      "grad_norm": 17.136259078979492,
      "learning_rate": 2.6607717041800645e-05,
      "loss": 0.6844,
      "step": 1180
    },
    {
      "epoch": 0.476,
      "grad_norm": 12.520773887634277,
      "learning_rate": 2.6406752411575564e-05,
      "loss": 0.542,
      "step": 1190
    },
    {
      "epoch": 0.48,
      "grad_norm": 8.42679214477539,
      "learning_rate": 2.6205787781350482e-05,
      "loss": 0.517,
      "step": 1200
    },
    {
      "epoch": 0.484,
      "grad_norm": 16.89554214477539,
      "learning_rate": 2.60048231511254e-05,
      "loss": 0.5378,
      "step": 1210
    },
    {
      "epoch": 0.488,
      "grad_norm": 20.790712356567383,
      "learning_rate": 2.580385852090032e-05,
      "loss": 0.4598,
      "step": 1220
    },
    {
      "epoch": 0.492,
      "grad_norm": 13.592977523803711,
      "learning_rate": 2.5602893890675245e-05,
      "loss": 0.5201,
      "step": 1230
    },
    {
      "epoch": 0.496,
      "grad_norm": 17.89519500732422,
      "learning_rate": 2.5401929260450164e-05,
      "loss": 0.5422,
      "step": 1240
    },
    {
      "epoch": 0.5,
      "grad_norm": 12.296363830566406,
      "learning_rate": 2.5200964630225083e-05,
      "loss": 0.5155,
      "step": 1250
    },
    {
      "epoch": 0.504,
      "grad_norm": 14.820162773132324,
      "learning_rate": 2.5e-05,
      "loss": 0.5141,
      "step": 1260
    },
    {
      "epoch": 0.508,
      "grad_norm": 14.70346736907959,
      "learning_rate": 2.479903536977492e-05,
      "loss": 0.5812,
      "step": 1270
    },
    {
      "epoch": 0.512,
      "grad_norm": 18.85811424255371,
      "learning_rate": 2.4598070739549842e-05,
      "loss": 0.5159,
      "step": 1280
    },
    {
      "epoch": 0.516,
      "grad_norm": 12.222372055053711,
      "learning_rate": 2.439710610932476e-05,
      "loss": 0.4363,
      "step": 1290
    },
    {
      "epoch": 0.52,
      "grad_norm": 49.18086242675781,
      "learning_rate": 2.419614147909968e-05,
      "loss": 0.7475,
      "step": 1300
    },
    {
      "epoch": 0.524,
      "grad_norm": 12.499253273010254,
      "learning_rate": 2.3995176848874598e-05,
      "loss": 0.6813,
      "step": 1310
    },
    {
      "epoch": 0.528,
      "grad_norm": 14.978521347045898,
      "learning_rate": 2.379421221864952e-05,
      "loss": 0.5517,
      "step": 1320
    },
    {
      "epoch": 0.532,
      "grad_norm": 16.008121490478516,
      "learning_rate": 2.359324758842444e-05,
      "loss": 0.5561,
      "step": 1330
    },
    {
      "epoch": 0.536,
      "grad_norm": 17.769926071166992,
      "learning_rate": 2.3392282958199358e-05,
      "loss": 0.5745,
      "step": 1340
    },
    {
      "epoch": 0.54,
      "grad_norm": 11.014586448669434,
      "learning_rate": 2.3191318327974276e-05,
      "loss": 0.4567,
      "step": 1350
    },
    {
      "epoch": 0.544,
      "grad_norm": 10.055672645568848,
      "learning_rate": 2.29903536977492e-05,
      "loss": 0.5071,
      "step": 1360
    },
    {
      "epoch": 0.548,
      "grad_norm": 10.68797779083252,
      "learning_rate": 2.2789389067524117e-05,
      "loss": 0.6491,
      "step": 1370
    },
    {
      "epoch": 0.552,
      "grad_norm": 9.747801780700684,
      "learning_rate": 2.2588424437299036e-05,
      "loss": 0.4888,
      "step": 1380
    },
    {
      "epoch": 0.556,
      "grad_norm": 11.837730407714844,
      "learning_rate": 2.2387459807073958e-05,
      "loss": 0.4526,
      "step": 1390
    },
    {
      "epoch": 0.56,
      "grad_norm": 14.05400562286377,
      "learning_rate": 2.2186495176848876e-05,
      "loss": 0.6307,
      "step": 1400
    },
    {
      "epoch": 0.564,
      "grad_norm": 12.709571838378906,
      "learning_rate": 2.1985530546623795e-05,
      "loss": 0.427,
      "step": 1410
    },
    {
      "epoch": 0.568,
      "grad_norm": 10.895342826843262,
      "learning_rate": 2.1784565916398714e-05,
      "loss": 0.4546,
      "step": 1420
    },
    {
      "epoch": 0.572,
      "grad_norm": 13.848987579345703,
      "learning_rate": 2.1583601286173636e-05,
      "loss": 0.4159,
      "step": 1430
    },
    {
      "epoch": 0.576,
      "grad_norm": 16.61017608642578,
      "learning_rate": 2.1382636655948555e-05,
      "loss": 0.733,
      "step": 1440
    },
    {
      "epoch": 0.58,
      "grad_norm": 10.283616065979004,
      "learning_rate": 2.1181672025723473e-05,
      "loss": 0.4581,
      "step": 1450
    },
    {
      "epoch": 0.584,
      "grad_norm": 11.248019218444824,
      "learning_rate": 2.0980707395498395e-05,
      "loss": 0.462,
      "step": 1460
    },
    {
      "epoch": 0.588,
      "grad_norm": 10.817742347717285,
      "learning_rate": 2.0779742765273314e-05,
      "loss": 0.451,
      "step": 1470
    },
    {
      "epoch": 0.592,
      "grad_norm": 10.615836143493652,
      "learning_rate": 2.0578778135048233e-05,
      "loss": 0.5714,
      "step": 1480
    },
    {
      "epoch": 0.596,
      "grad_norm": 12.22169017791748,
      "learning_rate": 2.037781350482315e-05,
      "loss": 0.8718,
      "step": 1490
    },
    {
      "epoch": 0.6,
      "grad_norm": 13.216890335083008,
      "learning_rate": 2.0176848874598074e-05,
      "loss": 0.3766,
      "step": 1500
    },
    {
      "epoch": 0.604,
      "grad_norm": 13.472972869873047,
      "learning_rate": 1.9975884244372992e-05,
      "loss": 0.5891,
      "step": 1510
    },
    {
      "epoch": 0.608,
      "grad_norm": 20.179187774658203,
      "learning_rate": 1.977491961414791e-05,
      "loss": 0.3744,
      "step": 1520
    },
    {
      "epoch": 0.612,
      "grad_norm": 12.630617141723633,
      "learning_rate": 1.957395498392283e-05,
      "loss": 0.4564,
      "step": 1530
    },
    {
      "epoch": 0.616,
      "grad_norm": 20.459508895874023,
      "learning_rate": 1.9372990353697752e-05,
      "loss": 0.4566,
      "step": 1540
    },
    {
      "epoch": 0.62,
      "grad_norm": 16.580251693725586,
      "learning_rate": 1.917202572347267e-05,
      "loss": 0.447,
      "step": 1550
    },
    {
      "epoch": 0.624,
      "grad_norm": 13.90858268737793,
      "learning_rate": 1.897106109324759e-05,
      "loss": 0.4133,
      "step": 1560
    },
    {
      "epoch": 0.628,
      "grad_norm": 10.297750473022461,
      "learning_rate": 1.877009646302251e-05,
      "loss": 0.9703,
      "step": 1570
    },
    {
      "epoch": 0.632,
      "grad_norm": 19.886884689331055,
      "learning_rate": 1.856913183279743e-05,
      "loss": 0.6253,
      "step": 1580
    },
    {
      "epoch": 0.636,
      "grad_norm": 10.709681510925293,
      "learning_rate": 1.836816720257235e-05,
      "loss": 0.4657,
      "step": 1590
    },
    {
      "epoch": 0.64,
      "grad_norm": 16.986331939697266,
      "learning_rate": 1.8167202572347267e-05,
      "loss": 0.3986,
      "step": 1600
    },
    {
      "epoch": 0.644,
      "grad_norm": 13.649085998535156,
      "learning_rate": 1.796623794212219e-05,
      "loss": 0.4337,
      "step": 1610
    },
    {
      "epoch": 0.648,
      "grad_norm": 7.645134449005127,
      "learning_rate": 1.7765273311897108e-05,
      "loss": 0.3901,
      "step": 1620
    },
    {
      "epoch": 0.652,
      "grad_norm": 11.727263450622559,
      "learning_rate": 1.7564308681672027e-05,
      "loss": 0.3545,
      "step": 1630
    },
    {
      "epoch": 0.656,
      "grad_norm": 6.705881595611572,
      "learning_rate": 1.736334405144695e-05,
      "loss": 0.3471,
      "step": 1640
    },
    {
      "epoch": 0.66,
      "grad_norm": 12.363304138183594,
      "learning_rate": 1.7162379421221868e-05,
      "loss": 0.4351,
      "step": 1650
    },
    {
      "epoch": 0.664,
      "grad_norm": 20.208723068237305,
      "learning_rate": 1.6961414790996786e-05,
      "loss": 0.4284,
      "step": 1660
    },
    {
      "epoch": 0.668,
      "grad_norm": 10.82363224029541,
      "learning_rate": 1.6760450160771705e-05,
      "loss": 0.3369,
      "step": 1670
    },
    {
      "epoch": 0.672,
      "grad_norm": 9.544486045837402,
      "learning_rate": 1.6559485530546627e-05,
      "loss": 0.4059,
      "step": 1680
    },
    {
      "epoch": 0.676,
      "grad_norm": 8.426627159118652,
      "learning_rate": 1.6358520900321546e-05,
      "loss": 0.4494,
      "step": 1690
    },
    {
      "epoch": 0.68,
      "grad_norm": 8.424084663391113,
      "learning_rate": 1.6157556270096464e-05,
      "loss": 0.4807,
      "step": 1700
    },
    {
      "epoch": 0.684,
      "grad_norm": 9.49954891204834,
      "learning_rate": 1.5956591639871383e-05,
      "loss": 0.3647,
      "step": 1710
    },
    {
      "epoch": 0.688,
      "grad_norm": 14.690208435058594,
      "learning_rate": 1.5755627009646305e-05,
      "loss": 0.3715,
      "step": 1720
    },
    {
      "epoch": 0.692,
      "grad_norm": 12.074922561645508,
      "learning_rate": 1.5554662379421224e-05,
      "loss": 0.491,
      "step": 1730
    },
    {
      "epoch": 0.696,
      "grad_norm": 13.278485298156738,
      "learning_rate": 1.5353697749196143e-05,
      "loss": 0.4185,
      "step": 1740
    },
    {
      "epoch": 0.7,
      "grad_norm": 12.987263679504395,
      "learning_rate": 1.5152733118971063e-05,
      "loss": 0.5613,
      "step": 1750
    },
    {
      "epoch": 0.704,
      "grad_norm": 6.863049030303955,
      "learning_rate": 1.4951768488745982e-05,
      "loss": 0.3245,
      "step": 1760
    },
    {
      "epoch": 0.708,
      "grad_norm": 11.087668418884277,
      "learning_rate": 1.47508038585209e-05,
      "loss": 0.4174,
      "step": 1770
    },
    {
      "epoch": 0.712,
      "grad_norm": 5.16309118270874,
      "learning_rate": 1.4549839228295819e-05,
      "loss": 0.3233,
      "step": 1780
    },
    {
      "epoch": 0.716,
      "grad_norm": 12.031776428222656,
      "learning_rate": 1.4348874598070741e-05,
      "loss": 0.3574,
      "step": 1790
    },
    {
      "epoch": 0.72,
      "grad_norm": 13.569413185119629,
      "learning_rate": 1.414790996784566e-05,
      "loss": 0.5619,
      "step": 1800
    },
    {
      "epoch": 0.724,
      "grad_norm": 5.905683517456055,
      "learning_rate": 1.3946945337620579e-05,
      "loss": 0.3286,
      "step": 1810
    },
    {
      "epoch": 0.728,
      "grad_norm": 10.061610221862793,
      "learning_rate": 1.3745980707395497e-05,
      "loss": 0.3899,
      "step": 1820
    },
    {
      "epoch": 0.732,
      "grad_norm": 12.293854713439941,
      "learning_rate": 1.354501607717042e-05,
      "loss": 0.4059,
      "step": 1830
    },
    {
      "epoch": 0.736,
      "grad_norm": 13.248871803283691,
      "learning_rate": 1.3344051446945338e-05,
      "loss": 0.4339,
      "step": 1840
    },
    {
      "epoch": 0.74,
      "grad_norm": 9.589434623718262,
      "learning_rate": 1.3143086816720257e-05,
      "loss": 0.4178,
      "step": 1850
    },
    {
      "epoch": 0.744,
      "grad_norm": 8.538604736328125,
      "learning_rate": 1.2942122186495179e-05,
      "loss": 0.3152,
      "step": 1860
    },
    {
      "epoch": 0.748,
      "grad_norm": 18.58129119873047,
      "learning_rate": 1.2741157556270097e-05,
      "loss": 0.4276,
      "step": 1870
    },
    {
      "epoch": 0.752,
      "grad_norm": 8.69501781463623,
      "learning_rate": 1.2540192926045016e-05,
      "loss": 0.4304,
      "step": 1880
    },
    {
      "epoch": 0.756,
      "grad_norm": 14.74836254119873,
      "learning_rate": 1.2339228295819937e-05,
      "loss": 0.3541,
      "step": 1890
    },
    {
      "epoch": 0.76,
      "grad_norm": 7.415429592132568,
      "learning_rate": 1.2138263665594855e-05,
      "loss": 0.3713,
      "step": 1900
    },
    {
      "epoch": 0.764,
      "grad_norm": 8.78702163696289,
      "learning_rate": 1.1937299035369776e-05,
      "loss": 0.3025,
      "step": 1910
    },
    {
      "epoch": 0.768,
      "grad_norm": 2.6222591400146484,
      "learning_rate": 1.1736334405144696e-05,
      "loss": 0.2279,
      "step": 1920
    },
    {
      "epoch": 0.772,
      "grad_norm": 8.457213401794434,
      "learning_rate": 1.1535369774919615e-05,
      "loss": 0.3841,
      "step": 1930
    },
    {
      "epoch": 0.776,
      "grad_norm": 9.097604751586914,
      "learning_rate": 1.1334405144694535e-05,
      "loss": 0.3436,
      "step": 1940
    },
    {
      "epoch": 0.78,
      "grad_norm": 9.933280944824219,
      "learning_rate": 1.1133440514469454e-05,
      "loss": 0.4276,
      "step": 1950
    },
    {
      "epoch": 0.784,
      "grad_norm": 9.58340072631836,
      "learning_rate": 1.0932475884244374e-05,
      "loss": 0.281,
      "step": 1960
    },
    {
      "epoch": 0.788,
      "grad_norm": 13.846723556518555,
      "learning_rate": 1.0731511254019293e-05,
      "loss": 0.2836,
      "step": 1970
    },
    {
      "epoch": 0.792,
      "grad_norm": 30.122060775756836,
      "learning_rate": 1.0530546623794213e-05,
      "loss": 0.3722,
      "step": 1980
    },
    {
      "epoch": 0.796,
      "grad_norm": 8.666303634643555,
      "learning_rate": 1.0329581993569132e-05,
      "loss": 0.2778,
      "step": 1990
    },
    {
      "epoch": 0.8,
      "grad_norm": 7.968908786773682,
      "learning_rate": 1.0128617363344052e-05,
      "loss": 0.2778,
      "step": 2000
    },
    {
      "epoch": 0.804,
      "grad_norm": 6.987481594085693,
      "learning_rate": 9.927652733118971e-06,
      "loss": 0.2778,
      "step": 2010
    },
    {
      "epoch": 0.808,
      "grad_norm": 11.547746658325195,
      "learning_rate": 9.726688102893891e-06,
      "loss": 0.3181,
      "step": 2020
    },
    {
      "epoch": 0.812,
      "grad_norm": 7.187608242034912,
      "learning_rate": 9.525723472668812e-06,
      "loss": 0.3211,
      "step": 2030
    },
    {
      "epoch": 0.816,
      "grad_norm": 14.975872039794922,
      "learning_rate": 9.32475884244373e-06,
      "loss": 0.2335,
      "step": 2040
    },
    {
      "epoch": 0.82,
      "grad_norm": 5.20744514465332,
      "learning_rate": 9.123794212218651e-06,
      "loss": 0.3012,
      "step": 2050
    },
    {
      "epoch": 0.824,
      "grad_norm": 9.876429557800293,
      "learning_rate": 8.92282958199357e-06,
      "loss": 0.3095,
      "step": 2060
    },
    {
      "epoch": 0.828,
      "grad_norm": 7.847969055175781,
      "learning_rate": 8.72186495176849e-06,
      "loss": 0.3336,
      "step": 2070
    },
    {
      "epoch": 0.832,
      "grad_norm": 5.847342014312744,
      "learning_rate": 8.520900321543409e-06,
      "loss": 0.3471,
      "step": 2080
    },
    {
      "epoch": 0.836,
      "grad_norm": 12.866349220275879,
      "learning_rate": 8.319935691318329e-06,
      "loss": 0.5096,
      "step": 2090
    },
    {
      "epoch": 0.84,
      "grad_norm": 5.676148891448975,
      "learning_rate": 8.118971061093248e-06,
      "loss": 0.3028,
      "step": 2100
    },
    {
      "epoch": 0.844,
      "grad_norm": 10.639591217041016,
      "learning_rate": 7.918006430868168e-06,
      "loss": 0.3065,
      "step": 2110
    },
    {
      "epoch": 0.848,
      "grad_norm": 4.760974884033203,
      "learning_rate": 7.717041800643089e-06,
      "loss": 0.2966,
      "step": 2120
    },
    {
      "epoch": 0.852,
      "grad_norm": 11.585098266601562,
      "learning_rate": 7.516077170418006e-06,
      "loss": 0.3466,
      "step": 2130
    },
    {
      "epoch": 0.856,
      "grad_norm": 4.685300827026367,
      "learning_rate": 7.315112540192927e-06,
      "loss": 0.397,
      "step": 2140
    },
    {
      "epoch": 0.86,
      "grad_norm": 7.504604339599609,
      "learning_rate": 7.1141479099678455e-06,
      "loss": 0.2216,
      "step": 2150
    },
    {
      "epoch": 0.864,
      "grad_norm": 16.2221736907959,
      "learning_rate": 6.913183279742766e-06,
      "loss": 0.3863,
      "step": 2160
    },
    {
      "epoch": 0.868,
      "grad_norm": 9.542070388793945,
      "learning_rate": 6.732315112540192e-06,
      "loss": 0.2478,
      "step": 2170
    },
    {
      "epoch": 0.872,
      "grad_norm": 12.575300216674805,
      "learning_rate": 6.531350482315113e-06,
      "loss": 0.3366,
      "step": 2180
    },
    {
      "epoch": 0.876,
      "grad_norm": 12.5275297164917,
      "learning_rate": 6.330385852090033e-06,
      "loss": 0.42,
      "step": 2190
    },
    {
      "epoch": 0.88,
      "grad_norm": 7.2454514503479,
      "learning_rate": 6.129421221864952e-06,
      "loss": 0.2943,
      "step": 2200
    },
    {
      "epoch": 0.884,
      "grad_norm": 5.587230682373047,
      "learning_rate": 5.928456591639871e-06,
      "loss": 0.2124,
      "step": 2210
    },
    {
      "epoch": 0.888,
      "grad_norm": 12.111313819885254,
      "learning_rate": 5.727491961414791e-06,
      "loss": 0.4733,
      "step": 2220
    },
    {
      "epoch": 0.892,
      "grad_norm": 13.287424087524414,
      "learning_rate": 5.526527331189711e-06,
      "loss": 0.3033,
      "step": 2230
    },
    {
      "epoch": 0.896,
      "grad_norm": 16.977447509765625,
      "learning_rate": 5.325562700964631e-06,
      "loss": 0.3256,
      "step": 2240
    },
    {
      "epoch": 0.9,
      "grad_norm": 6.201746463775635,
      "learning_rate": 5.12459807073955e-06,
      "loss": 0.1995,
      "step": 2250
    },
    {
      "epoch": 0.904,
      "grad_norm": 10.706707954406738,
      "learning_rate": 4.92363344051447e-06,
      "loss": 0.3303,
      "step": 2260
    },
    {
      "epoch": 0.908,
      "grad_norm": 14.989971160888672,
      "learning_rate": 4.7226688102893895e-06,
      "loss": 0.2842,
      "step": 2270
    },
    {
      "epoch": 0.912,
      "grad_norm": 15.570676803588867,
      "learning_rate": 4.521704180064309e-06,
      "loss": 0.212,
      "step": 2280
    },
    {
      "epoch": 0.916,
      "grad_norm": 3.9141690731048584,
      "learning_rate": 4.320739549839229e-06,
      "loss": 0.263,
      "step": 2290
    },
    {
      "epoch": 0.92,
      "grad_norm": 6.531361103057861,
      "learning_rate": 4.119774919614148e-06,
      "loss": 0.2503,
      "step": 2300
    },
    {
      "epoch": 0.924,
      "grad_norm": 8.718388557434082,
      "learning_rate": 3.918810289389068e-06,
      "loss": 0.2793,
      "step": 2310
    },
    {
      "epoch": 0.928,
      "grad_norm": 11.070220947265625,
      "learning_rate": 3.7178456591639876e-06,
      "loss": 0.2585,
      "step": 2320
    },
    {
      "epoch": 0.932,
      "grad_norm": 7.7603759765625,
      "learning_rate": 3.516881028938907e-06,
      "loss": 0.5099,
      "step": 2330
    },
    {
      "epoch": 0.936,
      "grad_norm": 15.190449714660645,
      "learning_rate": 3.3159163987138267e-06,
      "loss": 0.2601,
      "step": 2340
    },
    {
      "epoch": 0.94,
      "grad_norm": 9.578422546386719,
      "learning_rate": 3.1149517684887463e-06,
      "loss": 0.264,
      "step": 2350
    },
    {
      "epoch": 0.944,
      "grad_norm": 10.097108840942383,
      "learning_rate": 2.913987138263666e-06,
      "loss": 0.6162,
      "step": 2360
    },
    {
      "epoch": 0.948,
      "grad_norm": 8.5523042678833,
      "learning_rate": 2.7130225080385853e-06,
      "loss": 0.2649,
      "step": 2370
    },
    {
      "epoch": 0.952,
      "grad_norm": 8.127545356750488,
      "learning_rate": 2.512057877813505e-06,
      "loss": 0.4394,
      "step": 2380
    },
    {
      "epoch": 0.956,
      "grad_norm": 8.142045021057129,
      "learning_rate": 2.311093247588425e-06,
      "loss": 0.2339,
      "step": 2390
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.302300930023193,
      "learning_rate": 2.1101286173633444e-06,
      "loss": 0.2135,
      "step": 2400
    },
    {
      "epoch": 0.964,
      "grad_norm": 4.838590621948242,
      "learning_rate": 1.909163987138264e-06,
      "loss": 0.3181,
      "step": 2410
    },
    {
      "epoch": 0.968,
      "grad_norm": 6.775151252746582,
      "learning_rate": 1.7081993569131833e-06,
      "loss": 0.2516,
      "step": 2420
    },
    {
      "epoch": 0.972,
      "grad_norm": 13.491608619689941,
      "learning_rate": 1.507234726688103e-06,
      "loss": 0.2294,
      "step": 2430
    },
    {
      "epoch": 0.976,
      "grad_norm": 7.067889213562012,
      "learning_rate": 1.3062700964630226e-06,
      "loss": 0.2206,
      "step": 2440
    },
    {
      "epoch": 0.98,
      "grad_norm": 6.7473530769348145,
      "learning_rate": 1.1053054662379423e-06,
      "loss": 0.2003,
      "step": 2450
    },
    {
      "epoch": 0.984,
      "grad_norm": 9.074109077453613,
      "learning_rate": 9.043408360128617e-07,
      "loss": 0.2343,
      "step": 2460
    },
    {
      "epoch": 0.988,
      "grad_norm": 10.73699951171875,
      "learning_rate": 7.033762057877814e-07,
      "loss": 0.6721,
      "step": 2470
    },
    {
      "epoch": 0.992,
      "grad_norm": 5.29847526550293,
      "learning_rate": 5.02411575562701e-07,
      "loss": 0.236,
      "step": 2480
    },
    {
      "epoch": 0.996,
      "grad_norm": 10.172593116760254,
      "learning_rate": 3.014469453376206e-07,
      "loss": 0.216,
      "step": 2490
    },
    {
      "epoch": 1.0,
      "grad_norm": 16.507993698120117,
      "learning_rate": 1.0048231511254019e-07,
      "loss": 0.2836,
      "step": 2500
    }
  ],
  "logging_steps": 10,
  "max_steps": 2500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 4.513321594098893e+16,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}