{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 500,
  "global_step": 2025,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.014814814814814815,
      "grad_norm": 26.017102109115793,
      "learning_rate": 4.926108374384237e-07,
      "loss": 0.5936,
      "step": 10
    },
    {
      "epoch": 0.02962962962962963,
      "grad_norm": 12.914864221350799,
      "learning_rate": 9.852216748768474e-07,
      "loss": 0.3977,
      "step": 20
    },
    {
      "epoch": 0.044444444444444446,
      "grad_norm": 7.447780113204344,
      "learning_rate": 1.4778325123152712e-06,
      "loss": 0.2758,
      "step": 30
    },
    {
      "epoch": 0.05925925925925926,
      "grad_norm": 7.2037494981971735,
      "learning_rate": 1.970443349753695e-06,
      "loss": 0.206,
      "step": 40
    },
    {
      "epoch": 0.07407407407407407,
      "grad_norm": 9.39085512125214,
      "learning_rate": 2.4630541871921186e-06,
      "loss": 0.2267,
      "step": 50
    },
    {
      "epoch": 0.08888888888888889,
      "grad_norm": 7.044937181975286,
      "learning_rate": 2.9556650246305424e-06,
      "loss": 0.2088,
      "step": 60
    },
    {
      "epoch": 0.1037037037037037,
      "grad_norm": 5.795730113429066,
      "learning_rate": 3.448275862068966e-06,
      "loss": 0.197,
      "step": 70
    },
    {
      "epoch": 0.11851851851851852,
      "grad_norm": 7.080218597087718,
      "learning_rate": 3.94088669950739e-06,
      "loss": 0.1934,
      "step": 80
    },
    {
      "epoch": 0.13333333333333333,
      "grad_norm": 5.53348676182737,
      "learning_rate": 4.4334975369458135e-06,
      "loss": 0.2117,
      "step": 90
    },
    {
      "epoch": 0.14814814814814814,
      "grad_norm": 5.286987282479059,
      "learning_rate": 4.926108374384237e-06,
      "loss": 0.2408,
      "step": 100
    },
    {
      "epoch": 0.16296296296296298,
      "grad_norm": 6.606219609047498,
      "learning_rate": 5.41871921182266e-06,
      "loss": 0.2066,
      "step": 110
    },
    {
      "epoch": 0.17777777777777778,
      "grad_norm": 4.701177482407085,
      "learning_rate": 5.911330049261085e-06,
      "loss": 0.2281,
      "step": 120
    },
    {
      "epoch": 0.1925925925925926,
      "grad_norm": 5.456763973680646,
      "learning_rate": 6.403940886699508e-06,
      "loss": 0.234,
      "step": 130
    },
    {
      "epoch": 0.2074074074074074,
      "grad_norm": 5.574928605824461,
      "learning_rate": 6.896551724137932e-06,
      "loss": 0.2239,
      "step": 140
    },
    {
      "epoch": 0.2222222222222222,
      "grad_norm": 4.2789352272007095,
      "learning_rate": 7.3891625615763555e-06,
      "loss": 0.2339,
      "step": 150
    },
    {
      "epoch": 0.23703703703703705,
      "grad_norm": 4.6444068197911585,
      "learning_rate": 7.88177339901478e-06,
      "loss": 0.2261,
      "step": 160
    },
    {
      "epoch": 0.2518518518518518,
      "grad_norm": 4.889863953460466,
      "learning_rate": 8.374384236453203e-06,
      "loss": 0.253,
      "step": 170
    },
    {
      "epoch": 0.26666666666666666,
      "grad_norm": 4.819379627675783,
      "learning_rate": 8.866995073891627e-06,
      "loss": 0.2592,
      "step": 180
    },
    {
      "epoch": 0.2814814814814815,
      "grad_norm": 4.783281320760556,
      "learning_rate": 9.359605911330049e-06,
      "loss": 0.2558,
      "step": 190
    },
    {
      "epoch": 0.2962962962962963,
      "grad_norm": 4.6210148664541055,
      "learning_rate": 9.852216748768475e-06,
      "loss": 0.2791,
      "step": 200
    },
    {
      "epoch": 0.3111111111111111,
      "grad_norm": 6.194129626142383,
      "learning_rate": 9.99963580513638e-06,
      "loss": 0.266,
      "step": 210
    },
    {
      "epoch": 0.32592592592592595,
      "grad_norm": 4.306028672789547,
      "learning_rate": 9.997852121279563e-06,
      "loss": 0.2644,
      "step": 220
    },
    {
      "epoch": 0.34074074074074073,
      "grad_norm": 4.550912215301358,
      "learning_rate": 9.994582585118449e-06,
      "loss": 0.2398,
      "step": 230
    },
    {
      "epoch": 0.35555555555555557,
      "grad_norm": 2.7873840728898416,
      "learning_rate": 9.989828168680164e-06,
      "loss": 0.2459,
      "step": 240
    },
    {
      "epoch": 0.37037037037037035,
      "grad_norm": 5.670238707619235,
      "learning_rate": 9.983590285444025e-06,
      "loss": 0.2874,
      "step": 250
    },
    {
      "epoch": 0.3851851851851852,
      "grad_norm": 3.666204083864708,
      "learning_rate": 9.975870789921322e-06,
      "loss": 0.243,
      "step": 260
    },
    {
      "epoch": 0.4,
      "grad_norm": 4.765419867995775,
      "learning_rate": 9.966671977103972e-06,
      "loss": 0.2418,
      "step": 270
    },
    {
      "epoch": 0.4148148148148148,
      "grad_norm": 4.266352929836881,
      "learning_rate": 9.955996581782218e-06,
      "loss": 0.3023,
      "step": 280
    },
    {
      "epoch": 0.42962962962962964,
      "grad_norm": 5.260394697457832,
      "learning_rate": 9.943847777731584e-06,
      "loss": 0.2765,
      "step": 290
    },
    {
      "epoch": 0.4444444444444444,
      "grad_norm": 8.072668515977384,
      "learning_rate": 9.93022917676932e-06,
      "loss": 0.2804,
      "step": 300
    },
    {
      "epoch": 0.45925925925925926,
      "grad_norm": 8.611241020880648,
      "learning_rate": 9.915144827680606e-06,
      "loss": 0.2703,
      "step": 310
    },
    {
      "epoch": 0.4740740740740741,
      "grad_norm": 5.269422886099256,
      "learning_rate": 9.898599215014868e-06,
      "loss": 0.2875,
      "step": 320
    },
    {
      "epoch": 0.4888888888888889,
      "grad_norm": 4.630321430071603,
      "learning_rate": 9.880597257752522e-06,
      "loss": 0.2639,
      "step": 330
    },
    {
      "epoch": 0.5037037037037037,
      "grad_norm": 6.86287315478417,
      "learning_rate": 9.861144307842574e-06,
      "loss": 0.2339,
      "step": 340
    },
    {
      "epoch": 0.5185185185185185,
      "grad_norm": 4.726853171099211,
      "learning_rate": 9.840246148611485e-06,
      "loss": 0.2552,
      "step": 350
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 3.24789768577898,
      "learning_rate": 9.817908993043819e-06,
      "loss": 0.2817,
      "step": 360
    },
    {
      "epoch": 0.5481481481481482,
      "grad_norm": 4.1350169337840965,
      "learning_rate": 9.794139481935108e-06,
      "loss": 0.2641,
      "step": 370
    },
    {
      "epoch": 0.562962962962963,
      "grad_norm": 2.730446652174673,
      "learning_rate": 9.768944681917582e-06,
      "loss": 0.2826,
      "step": 380
    },
    {
      "epoch": 0.5777777777777777,
      "grad_norm": 3.9971570005802266,
      "learning_rate": 9.742332083359252e-06,
      "loss": 0.2559,
      "step": 390
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 3.753217276080861,
      "learning_rate": 9.714309598137045e-06,
      "loss": 0.2306,
      "step": 400
    },
    {
      "epoch": 0.6074074074074074,
      "grad_norm": 15.424898203468624,
      "learning_rate": 9.68488555728462e-06,
      "loss": 0.2682,
      "step": 410
    },
    {
      "epoch": 0.6222222222222222,
      "grad_norm": 3.8975775295207318,
      "learning_rate": 9.654068708515564e-06,
      "loss": 0.2607,
      "step": 420
    },
    {
      "epoch": 0.6370370370370371,
      "grad_norm": 4.141551276129438,
      "learning_rate": 9.621868213622713e-06,
      "loss": 0.2383,
      "step": 430
    },
    {
      "epoch": 0.6518518518518519,
      "grad_norm": 3.9225855004559356,
      "learning_rate": 9.588293645754363e-06,
      "loss": 0.293,
      "step": 440
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 3.2761325115244624,
      "learning_rate": 9.553354986568201e-06,
      "loss": 0.2942,
      "step": 450
    },
    {
      "epoch": 0.6814814814814815,
      "grad_norm": 3.0682621048324945,
      "learning_rate": 9.517062623263768e-06,
      "loss": 0.2524,
      "step": 460
    },
    {
      "epoch": 0.6962962962962963,
      "grad_norm": 3.231493961817089,
      "learning_rate": 9.479427345494366e-06,
      "loss": 0.2623,
      "step": 470
    },
    {
      "epoch": 0.7111111111111111,
      "grad_norm": 5.3337328348116015,
      "learning_rate": 9.440460342159314e-06,
      "loss": 0.2786,
      "step": 480
    },
    {
      "epoch": 0.725925925925926,
      "grad_norm": 4.944222927733353,
      "learning_rate": 9.40017319807751e-06,
      "loss": 0.2276,
      "step": 490
    },
    {
      "epoch": 0.7407407407407407,
      "grad_norm": 4.898395138758685,
      "learning_rate": 9.358577890543277e-06,
      "loss": 0.2781,
      "step": 500
    },
    {
      "epoch": 0.7407407407407407,
      "eval_loss": 0.2680646777153015,
      "eval_runtime": 110.7868,
      "eval_samples_per_second": 2.708,
      "eval_steps_per_second": 1.354,
      "step": 500
    },
    {
      "epoch": 0.7555555555555555,
      "grad_norm": 3.853101145062431,
      "learning_rate": 9.315686785765556e-06,
      "loss": 0.2637,
      "step": 510
    },
    {
      "epoch": 0.7703703703703704,
      "grad_norm": 4.4209909567001535,
      "learning_rate": 9.271512635191427e-06,
      "loss": 0.2707,
      "step": 520
    },
    {
      "epoch": 0.7851851851851852,
      "grad_norm": 4.336113654547109,
      "learning_rate": 9.22606857171515e-06,
      "loss": 0.2376,
      "step": 530
    },
    {
      "epoch": 0.8,
      "grad_norm": 5.107334588558504,
      "learning_rate": 9.179368105773768e-06,
      "loss": 0.2495,
      "step": 540
    },
    {
      "epoch": 0.8148148148148148,
      "grad_norm": 4.730641918906785,
      "learning_rate": 9.131425121330477e-06,
      "loss": 0.2379,
      "step": 550
    },
    {
      "epoch": 0.8296296296296296,
      "grad_norm": 19.960301317231036,
      "learning_rate": 9.082253871746962e-06,
      "loss": 0.2702,
      "step": 560
    },
    {
      "epoch": 0.8444444444444444,
      "grad_norm": 6.6215773453228115,
      "learning_rate": 9.031868975545884e-06,
      "loss": 0.2961,
      "step": 570
    },
    {
      "epoch": 0.8592592592592593,
      "grad_norm": 3.0739713217438274,
      "learning_rate": 8.980285412064827e-06,
      "loss": 0.2381,
      "step": 580
    },
    {
      "epoch": 0.8740740740740741,
      "grad_norm": 3.748817408124034,
      "learning_rate": 8.92751851700297e-06,
      "loss": 0.2258,
      "step": 590
    },
    {
      "epoch": 0.8888888888888888,
      "grad_norm": 3.6234404866008996,
      "learning_rate": 8.873583977861802e-06,
      "loss": 0.2395,
      "step": 600
    },
    {
      "epoch": 0.9037037037037037,
      "grad_norm": 4.04735847376876,
      "learning_rate": 8.818497829281272e-06,
      "loss": 0.2458,
      "step": 610
    },
    {
      "epoch": 0.9185185185185185,
      "grad_norm": 3.6442745029714296,
      "learning_rate": 8.762276448272709e-06,
      "loss": 0.2552,
      "step": 620
    },
    {
      "epoch": 0.9333333333333333,
      "grad_norm": 4.331703349382409,
      "learning_rate": 8.70493654934996e-06,
      "loss": 0.2909,
      "step": 630
    },
    {
      "epoch": 0.9481481481481482,
      "grad_norm": 6.051325269419405,
      "learning_rate": 8.646495179560221e-06,
      "loss": 0.2564,
      "step": 640
    },
    {
      "epoch": 0.9629629629629629,
      "grad_norm": 4.436797287629697,
      "learning_rate": 8.586969713415949e-06,
      "loss": 0.2374,
      "step": 650
    },
    {
      "epoch": 0.9777777777777777,
      "grad_norm": 2.756495280511311,
      "learning_rate": 8.526377847729475e-06,
      "loss": 0.2441,
      "step": 660
    },
    {
      "epoch": 0.9925925925925926,
      "grad_norm": 2.460062031773824,
      "learning_rate": 8.46473759635176e-06,
      "loss": 0.2566,
      "step": 670
    },
    {
      "epoch": 1.0074074074074073,
      "grad_norm": 2.8968674379949215,
      "learning_rate": 8.402067284816919e-06,
      "loss": 0.1853,
      "step": 680
    },
    {
      "epoch": 1.0222222222222221,
      "grad_norm": 3.1957903906863465,
      "learning_rate": 8.338385544894073e-06,
      "loss": 0.127,
      "step": 690
    },
    {
      "epoch": 1.037037037037037,
      "grad_norm": 2.6559772016064587,
      "learning_rate": 8.273711309048145e-06,
      "loss": 0.1254,
      "step": 700
    },
    {
      "epoch": 1.0518518518518518,
      "grad_norm": 3.0419408376224033,
      "learning_rate": 8.208063804811293e-06,
      "loss": 0.1385,
      "step": 710
    },
    {
      "epoch": 1.0666666666666667,
      "grad_norm": 2.324938116275548,
      "learning_rate": 8.141462549066581e-06,
      "loss": 0.1344,
      "step": 720
    },
    {
      "epoch": 1.0814814814814815,
      "grad_norm": 3.4270759875212233,
      "learning_rate": 8.073927342245663e-06,
      "loss": 0.1419,
      "step": 730
    },
    {
      "epoch": 1.0962962962962963,
      "grad_norm": 2.6055600373408105,
      "learning_rate": 8.005478262442132e-06,
      "loss": 0.1308,
      "step": 740
    },
    {
      "epoch": 1.1111111111111112,
      "grad_norm": 2.573600018786121,
      "learning_rate": 7.936135659442355e-06,
      "loss": 0.133,
      "step": 750
    },
    {
      "epoch": 1.125925925925926,
      "grad_norm": 1.3199535934067952,
      "learning_rate": 7.86592014867551e-06,
      "loss": 0.1099,
      "step": 760
    },
    {
      "epoch": 1.1407407407407408,
      "grad_norm": 2.5610850600680113,
      "learning_rate": 7.794852605084661e-06,
      "loss": 0.1429,
      "step": 770
    },
    {
      "epoch": 1.1555555555555554,
      "grad_norm": 2.72955021912288,
      "learning_rate": 7.722954156920675e-06,
      "loss": 0.1227,
      "step": 780
    },
    {
      "epoch": 1.1703703703703703,
      "grad_norm": 3.447611029467879,
      "learning_rate": 7.650246179460826e-06,
      "loss": 0.1499,
      "step": 790
    },
    {
      "epoch": 1.1851851851851851,
      "grad_norm": 4.42799750544096,
      "learning_rate": 7.57675028865397e-06,
      "loss": 0.1564,
      "step": 800
    },
    {
      "epoch": 1.2,
      "grad_norm": 2.9245723364864586,
      "learning_rate": 7.502488334694167e-06,
      "loss": 0.1441,
      "step": 810
    },
    {
      "epoch": 1.2148148148148148,
      "grad_norm": 3.9787651452124524,
      "learning_rate": 7.427482395524646e-06,
      "loss": 0.1317,
      "step": 820
    },
    {
      "epoch": 1.2296296296296296,
      "grad_norm": 3.5447225202330745,
      "learning_rate": 7.35175477027408e-06,
      "loss": 0.1211,
      "step": 830
    },
    {
      "epoch": 1.2444444444444445,
      "grad_norm": 3.013749806838306,
      "learning_rate": 7.2753279726271e-06,
      "loss": 0.1284,
      "step": 840
    },
    {
      "epoch": 1.2592592592592593,
      "grad_norm": 2.8757278244677553,
      "learning_rate": 7.198224724131012e-06,
      "loss": 0.1436,
      "step": 850
    },
    {
      "epoch": 1.2740740740740741,
      "grad_norm": 3.813666156239548,
      "learning_rate": 7.120467947440719e-06,
      "loss": 0.1221,
      "step": 860
    },
    {
      "epoch": 1.2888888888888888,
      "grad_norm": 2.3407714617223885,
      "learning_rate": 7.042080759503866e-06,
      "loss": 0.1162,
      "step": 870
    },
    {
      "epoch": 1.3037037037037038,
      "grad_norm": 2.5559098463338397,
      "learning_rate": 6.963086464688209e-06,
      "loss": 0.1291,
      "step": 880
    },
    {
      "epoch": 1.3185185185185184,
      "grad_norm": 2.5329035033815637,
      "learning_rate": 6.883508547853268e-06,
      "loss": 0.1319,
      "step": 890
    },
    {
      "epoch": 1.3333333333333333,
      "grad_norm": 4.756367202905442,
      "learning_rate": 6.8033706673683276e-06,
      "loss": 0.125,
      "step": 900
    },
    {
      "epoch": 1.348148148148148,
      "grad_norm": 2.4135868103281446,
      "learning_rate": 6.722696648078838e-06,
      "loss": 0.1382,
      "step": 910
    },
    {
      "epoch": 1.362962962962963,
      "grad_norm": 2.1274954103320143,
      "learning_rate": 6.641510474223338e-06,
      "loss": 0.1311,
      "step": 920
    },
    {
      "epoch": 1.3777777777777778,
      "grad_norm": 5.925560885437899,
      "learning_rate": 6.559836282302984e-06,
      "loss": 0.1416,
      "step": 930
    },
    {
      "epoch": 1.3925925925925926,
      "grad_norm": 2.628591269898234,
      "learning_rate": 6.477698353905808e-06,
      "loss": 0.1519,
      "step": 940
    },
    {
      "epoch": 1.4074074074074074,
      "grad_norm": 2.7741467084148925,
      "learning_rate": 6.395121108487855e-06,
      "loss": 0.1179,
      "step": 950
    },
    {
      "epoch": 1.4222222222222223,
      "grad_norm": 2.4497960109998362,
      "learning_rate": 6.312129096113313e-06,
      "loss": 0.1329,
      "step": 960
    },
    {
      "epoch": 1.4370370370370371,
      "grad_norm": 3.192126852233065,
      "learning_rate": 6.228746990155831e-06,
      "loss": 0.0981,
      "step": 970
    },
    {
      "epoch": 1.4518518518518517,
      "grad_norm": 2.180408863260512,
      "learning_rate": 6.144999579963164e-06,
      "loss": 0.116,
      "step": 980
    },
    {
      "epoch": 1.4666666666666668,
      "grad_norm": 2.60239822124948,
      "learning_rate": 6.060911763487353e-06,
      "loss": 0.136,
      "step": 990
    },
    {
      "epoch": 1.4814814814814814,
      "grad_norm": 2.9763711731701745,
      "learning_rate": 5.976508539882604e-06,
      "loss": 0.1417,
      "step": 1000
    },
    {
      "epoch": 1.4814814814814814,
      "eval_loss": 0.23991718888282776,
      "eval_runtime": 110.0876,
      "eval_samples_per_second": 2.725,
      "eval_steps_per_second": 1.363,
      "step": 1000
    },
    {
      "epoch": 1.4962962962962962,
      "grad_norm": 1.5705062635477183,
      "learning_rate": 5.891815002073081e-06,
      "loss": 0.1118,
      "step": 1010
    },
    {
      "epoch": 1.511111111111111,
      "grad_norm": 1.8570339015498183,
      "learning_rate": 5.806856329292839e-06,
      "loss": 0.1202,
      "step": 1020
    },
    {
      "epoch": 1.525925925925926,
      "grad_norm": 2.9744617003579523,
      "learning_rate": 5.721657779600071e-06,
      "loss": 0.1256,
      "step": 1030
    },
    {
      "epoch": 1.5407407407407407,
      "grad_norm": 3.0753703391352056,
      "learning_rate": 5.636244682367937e-06,
      "loss": 0.1402,
      "step": 1040
    },
    {
      "epoch": 1.5555555555555556,
      "grad_norm": 2.464360253185267,
      "learning_rate": 5.5506424307541895e-06,
      "loss": 0.1034,
      "step": 1050
    },
    {
      "epoch": 1.5703703703703704,
      "grad_norm": 2.836329426651843,
      "learning_rate": 5.464876474151835e-06,
      "loss": 0.1302,
      "step": 1060
    },
    {
      "epoch": 1.585185185185185,
      "grad_norm": 4.20442553150441,
      "learning_rate": 5.3789723106230675e-06,
      "loss": 0.128,
      "step": 1070
    },
    {
      "epoch": 1.6,
      "grad_norm": 2.189788584493857,
      "learning_rate": 5.292955479318756e-06,
      "loss": 0.1265,
      "step": 1080
    },
    {
      "epoch": 1.6148148148148147,
      "grad_norm": 2.1857430787318273,
      "learning_rate": 5.206851552885691e-06,
      "loss": 0.1276,
      "step": 1090
    },
    {
      "epoch": 1.6296296296296298,
      "grad_norm": 2.0830677851745283,
      "learning_rate": 5.120686129863882e-06,
      "loss": 0.1131,
      "step": 1100
    },
    {
      "epoch": 1.6444444444444444,
      "grad_norm": 2.399753090852167,
      "learning_rate": 5.0344848270761635e-06,
      "loss": 0.1353,
      "step": 1110
    },
    {
      "epoch": 1.6592592592592592,
      "grad_norm": 2.972662069394894,
      "learning_rate": 4.948273272012363e-06,
      "loss": 0.1256,
      "step": 1120
    },
    {
      "epoch": 1.674074074074074,
      "grad_norm": 2.725058666193907,
      "learning_rate": 4.862077095210284e-06,
      "loss": 0.1223,
      "step": 1130
    },
    {
      "epoch": 1.6888888888888889,
      "grad_norm": 2.403455581216295,
      "learning_rate": 4.775921922635806e-06,
      "loss": 0.1101,
      "step": 1140
    },
    {
      "epoch": 1.7037037037037037,
      "grad_norm": 2.954298808597469,
      "learning_rate": 4.689833368064326e-06,
      "loss": 0.1136,
      "step": 1150
    },
    {
      "epoch": 1.7185185185185186,
      "grad_norm": 2.065869616587133,
      "learning_rate": 4.603837025465829e-06,
      "loss": 0.0936,
      "step": 1160
    },
    {
      "epoch": 1.7333333333333334,
      "grad_norm": 2.5225001244758434,
      "learning_rate": 4.517958461395846e-06,
      "loss": 0.1035,
      "step": 1170
    },
    {
      "epoch": 1.748148148148148,
      "grad_norm": 2.278719889103367,
      "learning_rate": 4.432223207394577e-06,
      "loss": 0.0978,
      "step": 1180
    },
    {
      "epoch": 1.762962962962963,
      "grad_norm": 3.0324188432282204,
      "learning_rate": 4.346656752396388e-06,
      "loss": 0.1234,
      "step": 1190
    },
    {
      "epoch": 1.7777777777777777,
      "grad_norm": 3.15770874304934,
      "learning_rate": 4.261284535152016e-06,
      "loss": 0.1117,
      "step": 1200
    },
    {
      "epoch": 1.7925925925925927,
      "grad_norm": 2.4288815684292007,
      "learning_rate": 4.176131936665669e-06,
      "loss": 0.1151,
      "step": 1210
    },
    {
      "epoch": 1.8074074074074074,
      "grad_norm": 2.2345406971177586,
      "learning_rate": 4.0912242726493e-06,
      "loss": 0.1087,
      "step": 1220
    },
    {
      "epoch": 1.8222222222222222,
      "grad_norm": 1.3631354507880038,
      "learning_rate": 4.006586785996285e-06,
      "loss": 0.1094,
      "step": 1230
    },
    {
      "epoch": 1.837037037037037,
      "grad_norm": 2.759382864621944,
      "learning_rate": 3.922244639276773e-06,
      "loss": 0.1077,
      "step": 1240
    },
    {
      "epoch": 1.8518518518518519,
      "grad_norm": 2.299484616218074,
      "learning_rate": 3.838222907256884e-06,
      "loss": 0.1106,
      "step": 1250
    },
    {
      "epoch": 1.8666666666666667,
      "grad_norm": 3.450427642587667,
      "learning_rate": 3.7545465694440363e-06,
      "loss": 0.1075,
      "step": 1260
    },
    {
      "epoch": 1.8814814814814815,
      "grad_norm": 1.946789383498192,
      "learning_rate": 3.6712405026605792e-06,
      "loss": 0.1097,
      "step": 1270
    },
    {
      "epoch": 1.8962962962962964,
      "grad_norm": 3.1898982494578663,
      "learning_rate": 3.5883294736479612e-06,
      "loss": 0.1004,
      "step": 1280
    },
    {
      "epoch": 1.911111111111111,
      "grad_norm": 2.180121020731452,
      "learning_rate": 3.5058381317036285e-06,
      "loss": 0.0906,
      "step": 1290
    },
    {
      "epoch": 1.925925925925926,
      "grad_norm": 2.4521692799587593,
      "learning_rate": 3.423791001352823e-06,
      "loss": 0.1209,
      "step": 1300
    },
    {
      "epoch": 1.9407407407407407,
      "grad_norm": 2.578764727071429,
      "learning_rate": 3.3422124750574902e-06,
      "loss": 0.1084,
      "step": 1310
    },
    {
      "epoch": 1.9555555555555557,
      "grad_norm": 2.9192101640040526,
      "learning_rate": 3.2611268059644535e-06,
      "loss": 0.0974,
      "step": 1320
    },
    {
      "epoch": 1.9703703703703703,
      "grad_norm": 3.1900819524103254,
      "learning_rate": 3.1805581006949856e-06,
      "loss": 0.1291,
      "step": 1330
    },
    {
      "epoch": 1.9851851851851852,
      "grad_norm": 3.5372332007233585,
      "learning_rate": 3.100530312177956e-06,
      "loss": 0.1028,
      "step": 1340
    },
    {
      "epoch": 2.0,
      "grad_norm": 1.682319850341937,
      "learning_rate": 3.0210672325286806e-06,
      "loss": 0.1026,
      "step": 1350
    },
    {
      "epoch": 2.0148148148148146,
      "grad_norm": 1.3372912715867262,
      "learning_rate": 2.9421924859755525e-06,
      "loss": 0.0428,
      "step": 1360
    },
    {
      "epoch": 2.0296296296296297,
      "grad_norm": 1.3386975885602668,
      "learning_rate": 2.8639295218366115e-06,
      "loss": 0.0373,
      "step": 1370
    },
    {
      "epoch": 2.0444444444444443,
      "grad_norm": 1.2812593118990734,
      "learning_rate": 2.78630160754811e-06,
      "loss": 0.0354,
      "step": 1380
    },
    {
      "epoch": 2.0592592592592593,
      "grad_norm": 1.290388617260846,
      "learning_rate": 2.709331821747133e-06,
      "loss": 0.0265,
      "step": 1390
    },
    {
      "epoch": 2.074074074074074,
      "grad_norm": 1.573529577676281,
      "learning_rate": 2.63304304741037e-06,
      "loss": 0.0278,
      "step": 1400
    },
    {
      "epoch": 2.088888888888889,
      "grad_norm": 2.419667316602129,
      "learning_rate": 2.55745796505105e-06,
      "loss": 0.0487,
      "step": 1410
    },
    {
      "epoch": 2.1037037037037036,
      "grad_norm": 1.0785142520976503,
      "learning_rate": 2.482599045976059e-06,
      "loss": 0.0364,
      "step": 1420
    },
    {
      "epoch": 2.1185185185185187,
      "grad_norm": 2.0952689701078784,
      "learning_rate": 2.408488545605265e-06,
      "loss": 0.045,
      "step": 1430
    },
    {
      "epoch": 2.1333333333333333,
      "grad_norm": 2.823990301006105,
      "learning_rate": 2.3351484968550264e-06,
      "loss": 0.0368,
      "step": 1440
    },
    {
      "epoch": 2.148148148148148,
      "grad_norm": 1.7955041398866707,
      "learning_rate": 2.2626007035878377e-06,
      "loss": 0.0421,
      "step": 1450
    },
    {
      "epoch": 2.162962962962963,
      "grad_norm": 0.8109338878062605,
      "learning_rate": 2.1908667341300923e-06,
      "loss": 0.0315,
      "step": 1460
    },
    {
      "epoch": 2.1777777777777776,
      "grad_norm": 1.9133299721589396,
      "learning_rate": 2.1199679148598434e-06,
      "loss": 0.0348,
      "step": 1470
    },
    {
      "epoch": 2.1925925925925926,
      "grad_norm": 1.2071834999832143,
      "learning_rate": 2.0499253238665284e-06,
      "loss": 0.0496,
      "step": 1480
    },
    {
      "epoch": 2.2074074074074073,
      "grad_norm": 1.0217372762938706,
      "learning_rate": 1.9807597846844737e-06,
      "loss": 0.0287,
      "step": 1490
    },
    {
      "epoch": 2.2222222222222223,
      "grad_norm": 1.0023764675817943,
      "learning_rate": 1.9124918601021124e-06,
      "loss": 0.0362,
      "step": 1500
    },
    {
      "epoch": 2.2222222222222223,
      "eval_loss": 0.25548920035362244,
      "eval_runtime": 109.9471,
      "eval_samples_per_second": 2.729,
      "eval_steps_per_second": 1.364,
      "step": 1500
    },
    {
      "epoch": 2.237037037037037,
      "grad_norm": 3.3820011756317867,
      "learning_rate": 1.845141846048691e-06,
      "loss": 0.0327,
      "step": 1510
    },
    {
      "epoch": 2.251851851851852,
      "grad_norm": 1.8639592123431572,
      "learning_rate": 1.778729765560337e-06,
      "loss": 0.0275,
      "step": 1520
    },
    {
      "epoch": 2.2666666666666666,
      "grad_norm": 1.924019925176325,
      "learning_rate": 1.7132753628272403e-06,
      "loss": 0.0465,
      "step": 1530
    },
    {
      "epoch": 2.2814814814814817,
      "grad_norm": 1.152773684003778,
      "learning_rate": 1.6487980973237434e-06,
      "loss": 0.0308,
      "step": 1540
    },
    {
      "epoch": 2.2962962962962963,
      "grad_norm": 1.894824044203695,
      "learning_rate": 1.5853171380230791e-06,
      "loss": 0.0345,
      "step": 1550
    },
    {
      "epoch": 2.311111111111111,
      "grad_norm": 1.3328611183809627,
      "learning_rate": 1.5228513576984633e-06,
      "loss": 0.0281,
      "step": 1560
    },
    {
      "epoch": 2.325925925925926,
      "grad_norm": 3.750377643295275,
      "learning_rate": 1.4614193273122562e-06,
      "loss": 0.0445,
      "step": 1570
    },
    {
      "epoch": 2.3407407407407406,
      "grad_norm": 3.8759939460354915,
      "learning_rate": 1.401039310494855e-06,
      "loss": 0.0435,
      "step": 1580
    },
    {
      "epoch": 2.3555555555555556,
      "grad_norm": 3.0989225284077917,
      "learning_rate": 1.3417292581149388e-06,
      "loss": 0.0353,
      "step": 1590
    },
    {
      "epoch": 2.3703703703703702,
      "grad_norm": 0.9919766228227254,
      "learning_rate": 1.2835068029427188e-06,
      "loss": 0.0299,
      "step": 1600
    },
    {
      "epoch": 2.3851851851851853,
      "grad_norm": 1.1007285921225456,
      "learning_rate": 1.2263892544077439e-06,
      "loss": 0.0351,
      "step": 1610
    },
    {
      "epoch": 2.4,
      "grad_norm": 1.8350934062145996,
      "learning_rate": 1.1703935934528327e-06,
      "loss": 0.0449,
      "step": 1620
    },
    {
      "epoch": 2.414814814814815,
      "grad_norm": 1.0737399447700786,
      "learning_rate": 1.1155364674856834e-06,
      "loss": 0.0279,
      "step": 1630
    },
    {
      "epoch": 2.4296296296296296,
      "grad_norm": 1.1914051412570683,
      "learning_rate": 1.0618341854296176e-06,
      "loss": 0.0443,
      "step": 1640
    },
    {
      "epoch": 2.4444444444444446,
      "grad_norm": 1.4625234311241986,
      "learning_rate": 1.0093027128749722e-06,
      "loss": 0.0398,
      "step": 1650
    },
    {
      "epoch": 2.4592592592592593,
      "grad_norm": 2.809214532722792,
      "learning_rate": 9.57957667332562e-07,
      "loss": 0.0392,
      "step": 1660
    },
    {
      "epoch": 2.474074074074074,
      "grad_norm": 2.048944419289513,
      "learning_rate": 9.078143135906154e-07,
      "loss": 0.0361,
      "step": 1670
    },
    {
      "epoch": 2.488888888888889,
      "grad_norm": 2.2346039939090168,
      "learning_rate": 8.588875591765838e-07,
      "loss": 0.0424,
      "step": 1680
    },
    {
      "epoch": 2.5037037037037035,
      "grad_norm": 0.8685048120192506,
      "learning_rate": 8.111919499251653e-07,
      "loss": 0.0265,
      "step": 1690
    },
    {
      "epoch": 2.5185185185185186,
      "grad_norm": 5.531554078682163,
      "learning_rate": 7.647416656538464e-07,
      "loss": 0.0379,
      "step": 1700
    },
    {
      "epoch": 2.533333333333333,
      "grad_norm": 2.567773799038788,
      "learning_rate": 7.195505159472726e-07,
      "loss": 0.0269,
      "step": 1710
    },
    {
      "epoch": 2.5481481481481483,
      "grad_norm": 1.5337559661912539,
      "learning_rate": 6.756319360516856e-07,
      "loss": 0.0276,
      "step": 1720
    },
    {
      "epoch": 2.562962962962963,
      "grad_norm": 2.401685833878839,
      "learning_rate": 6.329989828806482e-07,
      "loss": 0.0283,
      "step": 1730
    },
    {
      "epoch": 2.5777777777777775,
      "grad_norm": 1.682913358837789,
      "learning_rate": 5.916643311332438e-07,
      "loss": 0.0391,
      "step": 1740
    },
    {
      "epoch": 2.5925925925925926,
      "grad_norm": 1.689757637438205,
      "learning_rate": 5.516402695259165e-07,
      "loss": 0.0324,
      "step": 1750
    },
    {
      "epoch": 2.6074074074074076,
      "grad_norm": 1.2197797882219985,
      "learning_rate": 5.12938697139056e-07,
      "loss": 0.0298,
      "step": 1760
    },
    {
      "epoch": 2.6222222222222222,
      "grad_norm": 2.191377307550078,
      "learning_rate": 4.755711198794233e-07,
      "loss": 0.0323,
      "step": 1770
    },
    {
      "epoch": 2.637037037037037,
      "grad_norm": 1.103714063244568,
      "learning_rate": 4.395486470594645e-07,
      "loss": 0.0256,
      "step": 1780
    },
    {
      "epoch": 2.651851851851852,
      "grad_norm": 1.6229391835726705,
      "learning_rate": 4.048819880945337e-07,
      "loss": 0.0291,
      "step": 1790
    },
    {
      "epoch": 2.6666666666666665,
      "grad_norm": 1.6366466447258576,
      "learning_rate": 3.7158144931900395e-07,
      "loss": 0.0262,
      "step": 1800
    },
    {
      "epoch": 2.6814814814814816,
      "grad_norm": 1.3157936293009576,
      "learning_rate": 3.396569309222114e-07,
      "loss": 0.0282,
      "step": 1810
    },
    {
      "epoch": 2.696296296296296,
      "grad_norm": 1.3484419903044749,
      "learning_rate": 3.091179240051462e-07,
      "loss": 0.0282,
      "step": 1820
    },
    {
      "epoch": 2.7111111111111112,
      "grad_norm": 2.820403024559756,
      "learning_rate": 2.799735077587695e-07,
      "loss": 0.036,
      "step": 1830
    },
    {
      "epoch": 2.725925925925926,
      "grad_norm": 1.108338803655822,
      "learning_rate": 2.5223234676478193e-07,
      "loss": 0.0236,
      "step": 1840
    },
    {
      "epoch": 2.7407407407407405,
      "grad_norm": 1.1593020570108006,
      "learning_rate": 2.2590268841966357e-07,
      "loss": 0.0306,
      "step": 1850
    },
    {
      "epoch": 2.7555555555555555,
      "grad_norm": 0.9513752742372911,
      "learning_rate": 2.0099236048273407e-07,
      "loss": 0.0306,
      "step": 1860
    },
    {
      "epoch": 2.7703703703703706,
      "grad_norm": 2.6444015892408417,
      "learning_rate": 1.7750876874897627e-07,
      "loss": 0.0317,
      "step": 1870
    },
    {
      "epoch": 2.785185185185185,
      "grad_norm": 2.028461793078949,
      "learning_rate": 1.554588948473068e-07,
      "loss": 0.0303,
      "step": 1880
    },
    {
      "epoch": 2.8,
      "grad_norm": 2.9681404257688637,
      "learning_rate": 1.3484929416495096e-07,
      "loss": 0.0318,
      "step": 1890
    },
    {
      "epoch": 2.814814814814815,
      "grad_norm": 1.7073934684848604,
      "learning_rate": 1.1568609389853546e-07,
      "loss": 0.0403,
      "step": 1900
    },
    {
      "epoch": 2.8296296296296295,
      "grad_norm": 1.5684215370418764,
      "learning_rate": 9.7974991232489e-08,
      "loss": 0.0284,
      "step": 1910
    },
    {
      "epoch": 2.8444444444444446,
      "grad_norm": 1.742009123872279,
      "learning_rate": 8.172125164527312e-08,
      "loss": 0.0408,
      "step": 1920
    },
    {
      "epoch": 2.859259259259259,
      "grad_norm": 1.7024512023509764,
      "learning_rate": 6.692970734397176e-08,
      "loss": 0.038,
      "step": 1930
    },
    {
      "epoch": 2.8740740740740742,
      "grad_norm": 1.0455017584864013,
      "learning_rate": 5.360475582768088e-08,
      "loss": 0.0362,
      "step": 1940
    },
    {
      "epoch": 2.888888888888889,
      "grad_norm": 1.3244853973367365,
      "learning_rate": 4.175035858013987e-08,
      "loss": 0.0294,
      "step": 1950
    },
    {
      "epoch": 2.9037037037037035,
      "grad_norm": 1.3623719555386105,
      "learning_rate": 3.13700398919925e-08,
      "loss": 0.0303,
      "step": 1960
    },
    {
      "epoch": 2.9185185185185185,
      "grad_norm": 1.417988437962572,
      "learning_rate": 2.2466885813018925e-08,
      "loss": 0.0284,
      "step": 1970
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 0.8742047293170354,
      "learning_rate": 1.504354323466073e-08,
      "loss": 0.0235,
      "step": 1980
    },
    {
      "epoch": 2.948148148148148,
      "grad_norm": 1.9231464337238604,
      "learning_rate": 9.102219103103161e-09,
      "loss": 0.0383,
      "step": 1990
    },
    {
      "epoch": 2.962962962962963,
      "grad_norm": 2.1190416620342303,
      "learning_rate": 4.644679763155524e-09,
      "loss": 0.0244,
      "step": 2000
    },
    {
      "epoch": 2.962962962962963,
      "eval_loss": 0.2506597936153412,
      "eval_runtime": 109.9607,
      "eval_samples_per_second": 2.728,
      "eval_steps_per_second": 1.364,
      "step": 2000
    },
    {
      "epoch": 2.977777777777778,
      "grad_norm": 1.0361197122874597,
      "learning_rate": 1.6722504331195822e-09,
      "loss": 0.024,
      "step": 2010
    },
    {
      "epoch": 2.9925925925925925,
      "grad_norm": 1.8532792834690945,
      "learning_rate": 1.8581481080415242e-10,
      "loss": 0.0264,
      "step": 2020
    },
    {
      "epoch": 3.0,
      "step": 2025,
      "total_flos": 15334618890240.0,
      "train_loss": 0.13858298796930432,
      "train_runtime": 10686.6427,
      "train_samples_per_second": 0.758,
      "train_steps_per_second": 0.189
    }
  ],
  "logging_steps": 10,
  "max_steps": 2025,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 15334618890240.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}