{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 7.802729044834308,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.031189083820662766,
      "grad_norm": 126.5,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.5428,
      "step": 10
    },
    {
      "epoch": 0.06237816764132553,
      "grad_norm": 326.0,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.4758,
      "step": 20
    },
    {
      "epoch": 0.0935672514619883,
      "grad_norm": 524.0,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.5854,
      "step": 30
    },
    {
      "epoch": 0.12475633528265107,
      "grad_norm": 446.0,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.8299,
      "step": 40
    },
    {
      "epoch": 0.15594541910331383,
      "grad_norm": 268.0,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.0183,
      "step": 50
    },
    {
      "epoch": 0.1871345029239766,
      "grad_norm": 1592.0,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.316,
      "step": 60
    },
    {
      "epoch": 0.21832358674463936,
      "grad_norm": 139.0,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.5705,
      "step": 70
    },
    {
      "epoch": 0.24951267056530213,
      "grad_norm": 374.0,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 5.897,
      "step": 80
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 115.0,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.1907,
      "step": 90
    },
    {
      "epoch": 0.31189083820662766,
      "grad_norm": 824.0,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.602,
      "step": 100
    },
    {
      "epoch": 0.34307992202729043,
      "grad_norm": 74.5,
      "learning_rate": 8.8e-06,
      "loss": 1.3299,
      "step": 110
    },
    {
      "epoch": 0.3742690058479532,
      "grad_norm": 129.0,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.106,
      "step": 120
    },
    {
      "epoch": 0.40545808966861596,
      "grad_norm": 168.0,
      "learning_rate": 1.04e-05,
      "loss": 1.5673,
      "step": 130
    },
    {
      "epoch": 0.43664717348927873,
      "grad_norm": 52.75,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.9792,
      "step": 140
    },
    {
      "epoch": 0.4678362573099415,
      "grad_norm": 95.5,
      "learning_rate": 1.2e-05,
      "loss": 0.9398,
      "step": 150
    },
    {
      "epoch": 0.49902534113060426,
      "grad_norm": 66.0,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 0.8719,
      "step": 160
    },
    {
      "epoch": 0.530214424951267,
      "grad_norm": 46.75,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 1.0627,
      "step": 170
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 41.75,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 0.8263,
      "step": 180
    },
    {
      "epoch": 0.5925925925925926,
      "grad_norm": 45.5,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 0.8172,
      "step": 190
    },
    {
      "epoch": 0.6237816764132553,
      "grad_norm": 41.5,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.8474,
      "step": 200
    },
    {
      "epoch": 0.6549707602339181,
      "grad_norm": 41.75,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 0.8815,
      "step": 210
    },
    {
      "epoch": 0.6861598440545809,
      "grad_norm": 45.75,
      "learning_rate": 1.76e-05,
      "loss": 0.8097,
      "step": 220
    },
    {
      "epoch": 0.7173489278752436,
      "grad_norm": 412.0,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 1.0962,
      "step": 230
    },
    {
      "epoch": 0.7485380116959064,
      "grad_norm": 52.25,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 10.0622,
      "step": 240
    },
    {
      "epoch": 0.7797270955165692,
      "grad_norm": 38.0,
      "learning_rate": 2e-05,
      "loss": 1.358,
      "step": 250
    },
    {
      "epoch": 0.8109161793372319,
      "grad_norm": 47.25,
      "learning_rate": 1.9999025240093045e-05,
      "loss": 1.0034,
      "step": 260
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 22.5,
      "learning_rate": 1.9996101150403543e-05,
      "loss": 0.7018,
      "step": 270
    },
    {
      "epoch": 0.8732943469785575,
      "grad_norm": 42.0,
      "learning_rate": 1.9991228300988586e-05,
      "loss": 0.6776,
      "step": 280
    },
    {
      "epoch": 0.9044834307992202,
      "grad_norm": 29.0,
      "learning_rate": 1.9984407641819812e-05,
      "loss": 0.7348,
      "step": 290
    },
    {
      "epoch": 0.935672514619883,
      "grad_norm": 42.5,
      "learning_rate": 1.9975640502598243e-05,
      "loss": 0.6959,
      "step": 300
    },
    {
      "epoch": 0.9668615984405458,
      "grad_norm": 45.75,
      "learning_rate": 1.9964928592495046e-05,
      "loss": 0.7085,
      "step": 310
    },
    {
      "epoch": 0.9980506822612085,
      "grad_norm": 38.0,
      "learning_rate": 1.9952273999818312e-05,
      "loss": 0.6367,
      "step": 320
    },
    {
      "epoch": 1.0300194931773878,
      "grad_norm": 44.0,
      "learning_rate": 1.9937679191605964e-05,
      "loss": 0.6069,
      "step": 330
    },
    {
      "epoch": 1.0612085769980506,
      "grad_norm": 17.625,
      "learning_rate": 1.9921147013144782e-05,
      "loss": 0.6435,
      "step": 340
    },
    {
      "epoch": 1.0923976608187134,
      "grad_norm": 26.875,
      "learning_rate": 1.9902680687415704e-05,
      "loss": 0.5493,
      "step": 350
    },
    {
      "epoch": 1.1235867446393761,
      "grad_norm": 33.5,
      "learning_rate": 1.988228381446553e-05,
      "loss": 0.5624,
      "step": 360
    },
    {
      "epoch": 1.154775828460039,
      "grad_norm": 40.75,
      "learning_rate": 1.985996037070505e-05,
      "loss": 0.5981,
      "step": 370
    },
    {
      "epoch": 1.1859649122807017,
      "grad_norm": 17.125,
      "learning_rate": 1.983571470813386e-05,
      "loss": 0.656,
      "step": 380
    },
    {
      "epoch": 1.2171539961013644,
      "grad_norm": 21.5,
      "learning_rate": 1.9809551553491918e-05,
      "loss": 0.4947,
      "step": 390
    },
    {
      "epoch": 1.2483430799220272,
      "grad_norm": 29.0,
      "learning_rate": 1.9781476007338058e-05,
      "loss": 0.562,
      "step": 400
    },
    {
      "epoch": 1.2795321637426902,
      "grad_norm": 8.1875,
      "learning_rate": 1.9751493543055634e-05,
      "loss": 0.4717,
      "step": 410
    },
    {
      "epoch": 1.310721247563353,
      "grad_norm": 41.5,
      "learning_rate": 1.9719610005785466e-05,
      "loss": 0.5865,
      "step": 420
    },
    {
      "epoch": 1.3419103313840157,
      "grad_norm": 23.375,
      "learning_rate": 1.9685831611286312e-05,
      "loss": 0.4994,
      "step": 430
    },
    {
      "epoch": 1.3730994152046785,
      "grad_norm": 18.375,
      "learning_rate": 1.9650164944723116e-05,
      "loss": 0.4635,
      "step": 440
    },
    {
      "epoch": 1.4042884990253413,
      "grad_norm": 25.0,
      "learning_rate": 1.961261695938319e-05,
      "loss": 0.5149,
      "step": 450
    },
    {
      "epoch": 1.435477582846004,
      "grad_norm": 37.5,
      "learning_rate": 1.9573194975320672e-05,
      "loss": 0.5077,
      "step": 460
    },
    {
      "epoch": 1.4666666666666668,
      "grad_norm": 21.0,
      "learning_rate": 1.9531906677929472e-05,
      "loss": 0.5361,
      "step": 470
    },
    {
      "epoch": 1.4978557504873296,
      "grad_norm": 11.3125,
      "learning_rate": 1.9488760116444966e-05,
      "loss": 0.5036,
      "step": 480
    },
    {
      "epoch": 1.5290448343079923,
      "grad_norm": 39.75,
      "learning_rate": 1.944376370237481e-05,
      "loss": 0.4732,
      "step": 490
    },
    {
      "epoch": 1.560233918128655,
      "grad_norm": 17.5,
      "learning_rate": 1.9396926207859085e-05,
      "loss": 0.4622,
      "step": 500
    },
    {
      "epoch": 1.5914230019493179,
      "grad_norm": 13.9375,
      "learning_rate": 1.9348256763960146e-05,
      "loss": 0.4519,
      "step": 510
    },
    {
      "epoch": 1.6226120857699806,
      "grad_norm": 21.375,
      "learning_rate": 1.9297764858882516e-05,
      "loss": 0.5471,
      "step": 520
    },
    {
      "epoch": 1.6538011695906434,
      "grad_norm": 34.25,
      "learning_rate": 1.9245460336123136e-05,
      "loss": 0.4767,
      "step": 530
    },
    {
      "epoch": 1.6849902534113061,
      "grad_norm": 23.625,
      "learning_rate": 1.9191353392552346e-05,
      "loss": 0.4921,
      "step": 540
    },
    {
      "epoch": 1.716179337231969,
      "grad_norm": 21.125,
      "learning_rate": 1.913545457642601e-05,
      "loss": 0.4537,
      "step": 550
    },
    {
      "epoch": 1.7473684210526317,
      "grad_norm": 28.5,
      "learning_rate": 1.907777478532909e-05,
      "loss": 0.466,
      "step": 560
    },
    {
      "epoch": 1.7785575048732944,
      "grad_norm": 15.375,
      "learning_rate": 1.901832526405114e-05,
      "loss": 0.4803,
      "step": 570
    },
    {
      "epoch": 1.8097465886939572,
      "grad_norm": 12.625,
      "learning_rate": 1.895711760239413e-05,
      "loss": 0.4508,
      "step": 580
    },
    {
      "epoch": 1.84093567251462,
      "grad_norm": 11.25,
      "learning_rate": 1.889416373291298e-05,
      "loss": 0.4905,
      "step": 590
    },
    {
      "epoch": 1.8721247563352827,
      "grad_norm": 11.375,
      "learning_rate": 1.8829475928589272e-05,
      "loss": 0.4482,
      "step": 600
    },
    {
      "epoch": 1.9033138401559455,
      "grad_norm": 43.25,
      "learning_rate": 1.8763066800438638e-05,
      "loss": 0.4531,
      "step": 610
    },
    {
      "epoch": 1.9345029239766083,
      "grad_norm": 24.75,
      "learning_rate": 1.869494929505219e-05,
      "loss": 0.4268,
      "step": 620
    },
    {
      "epoch": 1.965692007797271,
      "grad_norm": 49.5,
      "learning_rate": 1.8625136692072577e-05,
      "loss": 0.4737,
      "step": 630
    },
    {
      "epoch": 1.9968810916179338,
      "grad_norm": 22.5,
      "learning_rate": 1.855364260160507e-05,
      "loss": 0.487,
      "step": 640
    },
    {
      "epoch": 2.028849902534113,
      "grad_norm": 15.1875,
      "learning_rate": 1.848048096156426e-05,
      "loss": 0.4299,
      "step": 650
    },
    {
      "epoch": 2.0600389863547757,
      "grad_norm": 8.5,
      "learning_rate": 1.8405666034956842e-05,
      "loss": 0.4246,
      "step": 660
    },
    {
      "epoch": 2.0912280701754384,
      "grad_norm": 7.78125,
      "learning_rate": 1.8329212407100996e-05,
      "loss": 0.4017,
      "step": 670
    },
    {
      "epoch": 2.122417153996101,
      "grad_norm": 23.75,
      "learning_rate": 1.8251134982782952e-05,
      "loss": 0.4378,
      "step": 680
    },
    {
      "epoch": 2.153606237816764,
      "grad_norm": 17.125,
      "learning_rate": 1.8171448983351284e-05,
      "loss": 0.4024,
      "step": 690
    },
    {
      "epoch": 2.1847953216374267,
      "grad_norm": 19.125,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.395,
      "step": 700
    },
    {
      "epoch": 2.2159844054580895,
      "grad_norm": 21.125,
      "learning_rate": 1.8007313709487334e-05,
      "loss": 0.3597,
      "step": 710
    },
    {
      "epoch": 2.2471734892787523,
      "grad_norm": 4.6875,
      "learning_rate": 1.792289643355191e-05,
      "loss": 0.3673,
      "step": 720
    },
    {
      "epoch": 2.278362573099415,
      "grad_norm": 8.625,
      "learning_rate": 1.78369345732584e-05,
      "loss": 0.3961,
      "step": 730
    },
    {
      "epoch": 2.309551656920078,
      "grad_norm": 12.875,
      "learning_rate": 1.7749444887041797e-05,
      "loss": 0.3944,
      "step": 740
    },
    {
      "epoch": 2.3407407407407406,
      "grad_norm": 17.375,
      "learning_rate": 1.766044443118978e-05,
      "loss": 0.3777,
      "step": 750
    },
    {
      "epoch": 2.3719298245614033,
      "grad_norm": 12.375,
      "learning_rate": 1.7569950556517566e-05,
      "loss": 0.3799,
      "step": 760
    },
    {
      "epoch": 2.403118908382066,
      "grad_norm": 23.625,
      "learning_rate": 1.747798090498532e-05,
      "loss": 0.374,
      "step": 770
    },
    {
      "epoch": 2.434307992202729,
      "grad_norm": 23.125,
      "learning_rate": 1.7384553406258842e-05,
      "loss": 0.3891,
      "step": 780
    },
    {
      "epoch": 2.4654970760233916,
      "grad_norm": 15.1875,
      "learning_rate": 1.7289686274214116e-05,
      "loss": 0.3792,
      "step": 790
    },
    {
      "epoch": 2.4966861598440544,
      "grad_norm": 34.5,
      "learning_rate": 1.7193398003386514e-05,
      "loss": 0.374,
      "step": 800
    },
    {
      "epoch": 2.5278752436647176,
      "grad_norm": 15.875,
      "learning_rate": 1.709570736536521e-05,
      "loss": 0.3616,
      "step": 810
    },
    {
      "epoch": 2.5590643274853804,
      "grad_norm": 5.09375,
      "learning_rate": 1.6996633405133656e-05,
      "loss": 0.3745,
      "step": 820
    },
    {
      "epoch": 2.590253411306043,
      "grad_norm": 20.125,
      "learning_rate": 1.68961954373567e-05,
      "loss": 0.3756,
      "step": 830
    },
    {
      "epoch": 2.621442495126706,
      "grad_norm": 17.875,
      "learning_rate": 1.6794413042615168e-05,
      "loss": 0.3736,
      "step": 840
    },
    {
      "epoch": 2.6526315789473687,
      "grad_norm": 7.84375,
      "learning_rate": 1.6691306063588583e-05,
      "loss": 0.3827,
      "step": 850
    },
    {
      "epoch": 2.6838206627680314,
      "grad_norm": 16.625,
      "learning_rate": 1.6586894601186804e-05,
      "loss": 0.376,
      "step": 860
    },
    {
      "epoch": 2.715009746588694,
      "grad_norm": 20.125,
      "learning_rate": 1.6481199010631312e-05,
      "loss": 0.3706,
      "step": 870
    },
    {
      "epoch": 2.746198830409357,
      "grad_norm": 13.6875,
      "learning_rate": 1.63742398974869e-05,
      "loss": 0.3692,
      "step": 880
    },
    {
      "epoch": 2.7773879142300197,
      "grad_norm": 10.5625,
      "learning_rate": 1.6266038113644605e-05,
      "loss": 0.3232,
      "step": 890
    },
    {
      "epoch": 2.8085769980506825,
      "grad_norm": 11.75,
      "learning_rate": 1.6156614753256583e-05,
      "loss": 0.3514,
      "step": 900
    },
    {
      "epoch": 2.8397660818713453,
      "grad_norm": 16.625,
      "learning_rate": 1.6045991148623752e-05,
      "loss": 0.3483,
      "step": 910
    },
    {
      "epoch": 2.870955165692008,
      "grad_norm": 6.9375,
      "learning_rate": 1.5934188866037017e-05,
      "loss": 0.3546,
      "step": 920
    },
    {
      "epoch": 2.902144249512671,
      "grad_norm": 28.0,
      "learning_rate": 1.5821229701572897e-05,
      "loss": 0.3382,
      "step": 930
    },
    {
      "epoch": 2.9333333333333336,
      "grad_norm": 4.9375,
      "learning_rate": 1.570713567684432e-05,
      "loss": 0.327,
      "step": 940
    },
    {
      "epoch": 2.9645224171539963,
      "grad_norm": 6.71875,
      "learning_rate": 1.5591929034707468e-05,
      "loss": 0.3372,
      "step": 950
    },
    {
      "epoch": 2.995711500974659,
      "grad_norm": 14.375,
      "learning_rate": 1.5475632234925505e-05,
      "loss": 0.3392,
      "step": 960
    },
    {
      "epoch": 3.027680311890838,
      "grad_norm": 21.0,
      "learning_rate": 1.5358267949789968e-05,
      "loss": 0.3177,
      "step": 970
    },
    {
      "epoch": 3.058869395711501,
      "grad_norm": 17.0,
      "learning_rate": 1.5239859059700794e-05,
      "loss": 0.3119,
      "step": 980
    },
    {
      "epoch": 3.0900584795321637,
      "grad_norm": 8.1875,
      "learning_rate": 1.5120428648705716e-05,
      "loss": 0.304,
      "step": 990
    },
    {
      "epoch": 3.1212475633528265,
      "grad_norm": 10.75,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.2962,
      "step": 1000
    },
    {
      "epoch": 3.1524366471734893,
      "grad_norm": 17.25,
      "learning_rate": 1.4878596591387329e-05,
      "loss": 0.2975,
      "step": 1010
    },
    {
      "epoch": 3.183625730994152,
      "grad_norm": 2.25,
      "learning_rate": 1.4756242090702756e-05,
      "loss": 0.2932,
      "step": 1020
    },
    {
      "epoch": 3.214814814814815,
      "grad_norm": 16.75,
      "learning_rate": 1.463296035119862e-05,
      "loss": 0.2956,
      "step": 1030
    },
    {
      "epoch": 3.2460038986354776,
      "grad_norm": 17.75,
      "learning_rate": 1.4508775406894308e-05,
      "loss": 0.2861,
      "step": 1040
    },
    {
      "epoch": 3.2771929824561403,
      "grad_norm": 15.0,
      "learning_rate": 1.4383711467890776e-05,
      "loss": 0.2926,
      "step": 1050
    },
    {
      "epoch": 3.308382066276803,
      "grad_norm": 16.0,
      "learning_rate": 1.4257792915650728e-05,
      "loss": 0.3076,
      "step": 1060
    },
    {
      "epoch": 3.339571150097466,
      "grad_norm": 15.8125,
      "learning_rate": 1.413104429824542e-05,
      "loss": 0.2804,
      "step": 1070
    },
    {
      "epoch": 3.3707602339181286,
      "grad_norm": 3.75,
      "learning_rate": 1.4003490325568953e-05,
      "loss": 0.2651,
      "step": 1080
    },
    {
      "epoch": 3.4019493177387914,
      "grad_norm": 12.75,
      "learning_rate": 1.3875155864521031e-05,
      "loss": 0.2951,
      "step": 1090
    },
    {
      "epoch": 3.433138401559454,
      "grad_norm": 9.5625,
      "learning_rate": 1.3746065934159123e-05,
      "loss": 0.2963,
      "step": 1100
    },
    {
      "epoch": 3.464327485380117,
      "grad_norm": 14.0,
      "learning_rate": 1.3616245700820922e-05,
      "loss": 0.2999,
      "step": 1110
    },
    {
      "epoch": 3.4955165692007797,
      "grad_norm": 9.6875,
      "learning_rate": 1.3485720473218153e-05,
      "loss": 0.2867,
      "step": 1120
    },
    {
      "epoch": 3.5267056530214425,
      "grad_norm": 11.4375,
      "learning_rate": 1.3354515697502552e-05,
      "loss": 0.2846,
      "step": 1130
    },
    {
      "epoch": 3.557894736842105,
      "grad_norm": 10.875,
      "learning_rate": 1.3222656952305113e-05,
      "loss": 0.2934,
      "step": 1140
    },
    {
      "epoch": 3.589083820662768,
      "grad_norm": 12.1875,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.2819,
      "step": 1150
    },
    {
      "epoch": 3.6202729044834308,
      "grad_norm": 4.0625,
      "learning_rate": 1.2957080500440469e-05,
      "loss": 0.2896,
      "step": 1160
    },
    {
      "epoch": 3.6514619883040935,
      "grad_norm": 6.90625,
      "learning_rate": 1.2823414568428767e-05,
      "loss": 0.2913,
      "step": 1170
    },
    {
      "epoch": 3.6826510721247563,
      "grad_norm": 7.96875,
      "learning_rate": 1.2689198206152657e-05,
      "loss": 0.2945,
      "step": 1180
    },
    {
      "epoch": 3.713840155945419,
      "grad_norm": 6.25,
      "learning_rate": 1.2554457579357906e-05,
      "loss": 0.2989,
      "step": 1190
    },
    {
      "epoch": 3.745029239766082,
      "grad_norm": 4.9375,
      "learning_rate": 1.2419218955996677e-05,
      "loss": 0.3023,
      "step": 1200
    },
    {
      "epoch": 3.7762183235867446,
      "grad_norm": 14.0,
      "learning_rate": 1.2283508701106559e-05,
      "loss": 0.2751,
      "step": 1210
    },
    {
      "epoch": 3.8074074074074074,
      "grad_norm": 11.0625,
      "learning_rate": 1.2147353271670634e-05,
      "loss": 0.2741,
      "step": 1220
    },
    {
      "epoch": 3.83859649122807,
      "grad_norm": 6.875,
      "learning_rate": 1.2010779211459649e-05,
      "loss": 0.2744,
      "step": 1230
    },
    {
      "epoch": 3.869785575048733,
      "grad_norm": 4.71875,
      "learning_rate": 1.187381314585725e-05,
      "loss": 0.283,
      "step": 1240
    },
    {
      "epoch": 3.9009746588693957,
      "grad_norm": 6.78125,
      "learning_rate": 1.1736481776669307e-05,
      "loss": 0.2737,
      "step": 1250
    },
    {
      "epoch": 3.9321637426900584,
      "grad_norm": 4.625,
      "learning_rate": 1.159881187691835e-05,
      "loss": 0.2486,
      "step": 1260
    },
    {
      "epoch": 3.963352826510721,
      "grad_norm": 6.625,
      "learning_rate": 1.1460830285624119e-05,
      "loss": 0.2598,
      "step": 1270
    },
    {
      "epoch": 3.994541910331384,
      "grad_norm": 9.6875,
      "learning_rate": 1.1322563902571227e-05,
      "loss": 0.2669,
      "step": 1280
    },
    {
      "epoch": 4.026510721247563,
      "grad_norm": 7.46875,
      "learning_rate": 1.1184039683065014e-05,
      "loss": 0.2389,
      "step": 1290
    },
    {
      "epoch": 4.057699805068226,
      "grad_norm": 6.46875,
      "learning_rate": 1.1045284632676535e-05,
      "loss": 0.2191,
      "step": 1300
    },
    {
      "epoch": 4.088888888888889,
      "grad_norm": 6.25,
      "learning_rate": 1.0906325801977804e-05,
      "loss": 0.2313,
      "step": 1310
    },
    {
      "epoch": 4.120077972709551,
      "grad_norm": 3.5,
      "learning_rate": 1.0767190281268187e-05,
      "loss": 0.2376,
      "step": 1320
    },
    {
      "epoch": 4.151267056530214,
      "grad_norm": 6.125,
      "learning_rate": 1.0627905195293135e-05,
      "loss": 0.2448,
      "step": 1330
    },
    {
      "epoch": 4.182456140350877,
      "grad_norm": 7.5,
      "learning_rate": 1.0488497697956134e-05,
      "loss": 0.2349,
      "step": 1340
    },
    {
      "epoch": 4.21364522417154,
      "grad_norm": 7.78125,
      "learning_rate": 1.0348994967025012e-05,
      "loss": 0.2252,
      "step": 1350
    },
    {
      "epoch": 4.244834307992202,
      "grad_norm": 2.578125,
      "learning_rate": 1.0209424198833571e-05,
      "loss": 0.2309,
      "step": 1360
    },
    {
      "epoch": 4.276023391812865,
      "grad_norm": 8.0,
      "learning_rate": 1.0069812602979617e-05,
      "loss": 0.2137,
      "step": 1370
    },
    {
      "epoch": 4.307212475633528,
      "grad_norm": 4.78125,
      "learning_rate": 9.930187397020385e-06,
      "loss": 0.2394,
      "step": 1380
    },
    {
      "epoch": 4.338401559454191,
      "grad_norm": 8.4375,
      "learning_rate": 9.790575801166432e-06,
      "loss": 0.2256,
      "step": 1390
    },
    {
      "epoch": 4.3695906432748535,
      "grad_norm": 5.4375,
      "learning_rate": 9.651005032974994e-06,
      "loss": 0.2273,
      "step": 1400
    },
    {
      "epoch": 4.400779727095516,
      "grad_norm": 5.25,
      "learning_rate": 9.511502302043867e-06,
      "loss": 0.2177,
      "step": 1410
    },
    {
      "epoch": 4.431968810916179,
      "grad_norm": 4.8125,
      "learning_rate": 9.372094804706867e-06,
      "loss": 0.2323,
      "step": 1420
    },
    {
      "epoch": 4.463157894736842,
      "grad_norm": 2.34375,
      "learning_rate": 9.232809718731815e-06,
      "loss": 0.2425,
      "step": 1430
    },
    {
      "epoch": 4.4943469785575045,
      "grad_norm": 6.53125,
      "learning_rate": 9.093674198022201e-06,
      "loss": 0.2248,
      "step": 1440
    },
    {
      "epoch": 4.525536062378167,
      "grad_norm": 2.671875,
      "learning_rate": 8.954715367323468e-06,
      "loss": 0.2083,
      "step": 1450
    },
    {
      "epoch": 4.55672514619883,
      "grad_norm": 5.15625,
      "learning_rate": 8.815960316934991e-06,
      "loss": 0.2291,
      "step": 1460
    },
    {
      "epoch": 4.587914230019493,
      "grad_norm": 7.28125,
      "learning_rate": 8.677436097428775e-06,
      "loss": 0.2425,
      "step": 1470
    },
    {
      "epoch": 4.619103313840156,
      "grad_norm": 3.546875,
      "learning_rate": 8.539169714375885e-06,
      "loss": 0.2285,
      "step": 1480
    },
    {
      "epoch": 4.650292397660818,
      "grad_norm": 6.25,
      "learning_rate": 8.401188123081653e-06,
      "loss": 0.2291,
      "step": 1490
    },
    {
      "epoch": 4.681481481481481,
      "grad_norm": 5.46875,
      "learning_rate": 8.263518223330698e-06,
      "loss": 0.2253,
      "step": 1500
    },
    {
      "epoch": 4.712670565302144,
      "grad_norm": 3.03125,
      "learning_rate": 8.126186854142752e-06,
      "loss": 0.2278,
      "step": 1510
    },
    {
      "epoch": 4.743859649122807,
      "grad_norm": 2.109375,
      "learning_rate": 7.989220788540356e-06,
      "loss": 0.2349,
      "step": 1520
    },
    {
      "epoch": 4.775048732943469,
      "grad_norm": 2.234375,
      "learning_rate": 7.852646728329368e-06,
      "loss": 0.2424,
      "step": 1530
    },
    {
      "epoch": 4.806237816764132,
      "grad_norm": 4.84375,
      "learning_rate": 7.716491298893443e-06,
      "loss": 0.2168,
      "step": 1540
    },
    {
      "epoch": 4.837426900584795,
      "grad_norm": 4.75,
      "learning_rate": 7.580781044003324e-06,
      "loss": 0.2063,
      "step": 1550
    },
    {
      "epoch": 4.868615984405458,
      "grad_norm": 2.578125,
      "learning_rate": 7.445542420642097e-06,
      "loss": 0.2291,
      "step": 1560
    },
    {
      "epoch": 4.8998050682261205,
      "grad_norm": 2.453125,
      "learning_rate": 7.310801793847344e-06,
      "loss": 0.2117,
      "step": 1570
    },
    {
      "epoch": 4.930994152046783,
      "grad_norm": 2.640625,
      "learning_rate": 7.176585431571235e-06,
      "loss": 0.2152,
      "step": 1580
    },
    {
      "epoch": 4.962183235867446,
      "grad_norm": 6.90625,
      "learning_rate": 7.042919499559538e-06,
      "loss": 0.2092,
      "step": 1590
    },
    {
      "epoch": 4.993372319688109,
      "grad_norm": 9.0625,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.2142,
      "step": 1600
    },
    {
      "epoch": 5.025341130604288,
      "grad_norm": 4.53125,
      "learning_rate": 6.777343047694891e-06,
      "loss": 0.1957,
      "step": 1610
    },
    {
      "epoch": 5.056530214424951,
      "grad_norm": 2.984375,
      "learning_rate": 6.645484302497452e-06,
      "loss": 0.1895,
      "step": 1620
    },
    {
      "epoch": 5.087719298245614,
      "grad_norm": 2.671875,
      "learning_rate": 6.5142795267818505e-06,
      "loss": 0.1966,
      "step": 1630
    },
    {
      "epoch": 5.118908382066277,
      "grad_norm": 4.25,
      "learning_rate": 6.383754299179079e-06,
      "loss": 0.1837,
      "step": 1640
    },
    {
      "epoch": 5.150097465886939,
      "grad_norm": 3.359375,
      "learning_rate": 6.25393406584088e-06,
      "loss": 0.1886,
      "step": 1650
    },
    {
      "epoch": 5.181286549707602,
      "grad_norm": 4.46875,
      "learning_rate": 6.124844135478971e-06,
      "loss": 0.2011,
      "step": 1660
    },
    {
      "epoch": 5.212475633528265,
      "grad_norm": 6.03125,
      "learning_rate": 5.996509674431053e-06,
      "loss": 0.194,
      "step": 1670
    },
    {
      "epoch": 5.243664717348928,
      "grad_norm": 3.875,
      "learning_rate": 5.868955701754584e-06,
      "loss": 0.1921,
      "step": 1680
    },
    {
      "epoch": 5.2748538011695905,
      "grad_norm": 1.390625,
      "learning_rate": 5.742207084349274e-06,
      "loss": 0.1889,
      "step": 1690
    },
    {
      "epoch": 5.306042884990253,
      "grad_norm": 4.5,
      "learning_rate": 5.616288532109225e-06,
      "loss": 0.2059,
      "step": 1700
    },
    {
      "epoch": 5.337231968810916,
      "grad_norm": 2.671875,
      "learning_rate": 5.491224593105695e-06,
      "loss": 0.1957,
      "step": 1710
    },
    {
      "epoch": 5.368421052631579,
      "grad_norm": 3.875,
      "learning_rate": 5.367039648801386e-06,
      "loss": 0.201,
      "step": 1720
    },
    {
      "epoch": 5.3996101364522415,
      "grad_norm": 2.4375,
      "learning_rate": 5.243757909297247e-06,
      "loss": 0.2037,
      "step": 1730
    },
    {
      "epoch": 5.430799220272904,
      "grad_norm": 2.859375,
      "learning_rate": 5.121403408612672e-06,
      "loss": 0.1872,
      "step": 1740
    },
    {
      "epoch": 5.461988304093567,
      "grad_norm": 1.640625,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.182,
      "step": 1750
    },
    {
      "epoch": 5.49317738791423,
      "grad_norm": 1.984375,
      "learning_rate": 4.879571351294287e-06,
      "loss": 0.1959,
      "step": 1760
    },
    {
      "epoch": 5.524366471734893,
      "grad_norm": 1.328125,
      "learning_rate": 4.76014094029921e-06,
      "loss": 0.1848,
      "step": 1770
    },
    {
      "epoch": 5.555555555555555,
      "grad_norm": 1.65625,
      "learning_rate": 4.641732050210032e-06,
      "loss": 0.1791,
      "step": 1780
    },
    {
      "epoch": 5.586744639376218,
      "grad_norm": 2.734375,
      "learning_rate": 4.524367765074499e-06,
      "loss": 0.1885,
      "step": 1790
    },
    {
      "epoch": 5.617933723196881,
      "grad_norm": 4.5,
      "learning_rate": 4.408070965292534e-06,
      "loss": 0.2202,
      "step": 1800
    },
    {
      "epoch": 5.649122807017544,
      "grad_norm": 1.65625,
      "learning_rate": 4.292864323155684e-06,
      "loss": 0.1871,
      "step": 1810
    },
    {
      "epoch": 5.680311890838206,
      "grad_norm": 1.28125,
      "learning_rate": 4.178770298427107e-06,
      "loss": 0.1998,
      "step": 1820
    },
    {
      "epoch": 5.711500974658869,
      "grad_norm": 3.359375,
      "learning_rate": 4.065811133962987e-06,
      "loss": 0.1949,
      "step": 1830
    },
    {
      "epoch": 5.742690058479532,
      "grad_norm": 1.8203125,
      "learning_rate": 3.954008851376252e-06,
      "loss": 0.1873,
      "step": 1840
    },
    {
      "epoch": 5.773879142300195,
      "grad_norm": 1.359375,
      "learning_rate": 3.8433852467434175e-06,
      "loss": 0.1836,
      "step": 1850
    },
    {
      "epoch": 5.8050682261208575,
      "grad_norm": 2.375,
      "learning_rate": 3.7339618863553983e-06,
      "loss": 0.1755,
      "step": 1860
    },
    {
      "epoch": 5.83625730994152,
      "grad_norm": 1.5703125,
      "learning_rate": 3.625760102513103e-06,
      "loss": 0.1909,
      "step": 1870
    },
    {
      "epoch": 5.867446393762183,
      "grad_norm": 1.84375,
      "learning_rate": 3.5188009893686916e-06,
      "loss": 0.1815,
      "step": 1880
    },
    {
      "epoch": 5.898635477582846,
      "grad_norm": 1.703125,
      "learning_rate": 3.4131053988131947e-06,
      "loss": 0.1897,
      "step": 1890
    },
    {
      "epoch": 5.9298245614035086,
      "grad_norm": 1.2265625,
      "learning_rate": 3.308693936411421e-06,
      "loss": 0.183,
      "step": 1900
    },
    {
      "epoch": 5.961013645224171,
      "grad_norm": 1.2109375,
      "learning_rate": 3.2055869573848374e-06,
      "loss": 0.1865,
      "step": 1910
    },
    {
      "epoch": 5.992202729044834,
      "grad_norm": 1.4921875,
      "learning_rate": 3.103804562643302e-06,
      "loss": 0.197,
      "step": 1920
    },
    {
      "epoch": 6.024171539961014,
      "grad_norm": 0.9921875,
      "learning_rate": 3.003366594866345e-06,
      "loss": 0.1904,
      "step": 1930
    },
    {
      "epoch": 6.055360623781676,
      "grad_norm": 1.46875,
      "learning_rate": 2.9042926346347932e-06,
      "loss": 0.178,
      "step": 1940
    },
    {
      "epoch": 6.086549707602339,
      "grad_norm": 1.734375,
      "learning_rate": 2.8066019966134907e-06,
      "loss": 0.1877,
      "step": 1950
    },
    {
      "epoch": 6.117738791423002,
      "grad_norm": 1.0546875,
      "learning_rate": 2.7103137257858867e-06,
      "loss": 0.1877,
      "step": 1960
    },
    {
      "epoch": 6.148927875243665,
      "grad_norm": 1.5390625,
      "learning_rate": 2.615446593741161e-06,
      "loss": 0.185,
      "step": 1970
    },
    {
      "epoch": 6.1801169590643275,
      "grad_norm": 1.5,
      "learning_rate": 2.522019095014683e-06,
      "loss": 0.1695,
      "step": 1980
    },
    {
      "epoch": 6.21130604288499,
      "grad_norm": 0.90234375,
      "learning_rate": 2.4300494434824373e-06,
      "loss": 0.1652,
      "step": 1990
    },
    {
      "epoch": 6.242495126705653,
      "grad_norm": 1.7734375,
      "learning_rate": 2.339555568810221e-06,
      "loss": 0.1775,
      "step": 2000
    },
    {
      "epoch": 6.273684210526316,
      "grad_norm": 3.609375,
      "learning_rate": 2.2505551129582047e-06,
      "loss": 0.1856,
      "step": 2010
    },
    {
      "epoch": 6.3048732943469785,
      "grad_norm": 2.640625,
      "learning_rate": 2.163065426741603e-06,
      "loss": 0.1714,
      "step": 2020
    },
    {
      "epoch": 6.336062378167641,
      "grad_norm": 1.125,
      "learning_rate": 2.0771035664480944e-06,
      "loss": 0.1763,
      "step": 2030
    },
    {
      "epoch": 6.367251461988304,
      "grad_norm": 1.4375,
      "learning_rate": 1.9926862905126663e-06,
      "loss": 0.178,
      "step": 2040
    },
    {
      "epoch": 6.398440545808967,
      "grad_norm": 1.09375,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.1843,
      "step": 2050
    },
    {
      "epoch": 6.42962962962963,
      "grad_norm": 2.78125,
      "learning_rate": 1.8285510166487154e-06,
      "loss": 0.183,
      "step": 2060
    },
    {
      "epoch": 6.460818713450292,
      "grad_norm": 0.98046875,
      "learning_rate": 1.7488650172170496e-06,
      "loss": 0.1708,
      "step": 2070
    },
    {
      "epoch": 6.492007797270955,
      "grad_norm": 1.453125,
      "learning_rate": 1.6707875928990059e-06,
      "loss": 0.179,
      "step": 2080
    },
    {
      "epoch": 6.523196881091618,
      "grad_norm": 1.0703125,
      "learning_rate": 1.5943339650431578e-06,
      "loss": 0.1779,
      "step": 2090
    },
    {
      "epoch": 6.554385964912281,
      "grad_norm": 0.921875,
      "learning_rate": 1.5195190384357405e-06,
      "loss": 0.1668,
      "step": 2100
    },
    {
      "epoch": 6.585575048732943,
      "grad_norm": 1.2890625,
      "learning_rate": 1.446357398394934e-06,
      "loss": 0.1707,
      "step": 2110
    },
    {
      "epoch": 6.616764132553606,
      "grad_norm": 1.015625,
      "learning_rate": 1.3748633079274254e-06,
      "loss": 0.1726,
      "step": 2120
    },
    {
      "epoch": 6.647953216374269,
      "grad_norm": 1.5,
      "learning_rate": 1.30505070494781e-06,
      "loss": 0.1827,
      "step": 2130
    },
    {
      "epoch": 6.679142300194932,
      "grad_norm": 2.25,
      "learning_rate": 1.2369331995613664e-06,
      "loss": 0.1832,
      "step": 2140
    },
    {
      "epoch": 6.7103313840155945,
      "grad_norm": 1.5078125,
      "learning_rate": 1.1705240714107301e-06,
      "loss": 0.1884,
      "step": 2150
    },
    {
      "epoch": 6.741520467836257,
      "grad_norm": 1.9921875,
      "learning_rate": 1.1058362670870248e-06,
      "loss": 0.1884,
      "step": 2160
    },
    {
      "epoch": 6.77270955165692,
      "grad_norm": 1.0703125,
      "learning_rate": 1.042882397605871e-06,
      "loss": 0.1829,
      "step": 2170
    },
    {
      "epoch": 6.803898635477583,
      "grad_norm": 2.875,
      "learning_rate": 9.816747359488632e-07,
      "loss": 0.1791,
      "step": 2180
    },
    {
      "epoch": 6.8350877192982455,
      "grad_norm": 1.2578125,
      "learning_rate": 9.222252146709143e-07,
      "loss": 0.1748,
      "step": 2190
    },
    {
      "epoch": 6.866276803118908,
      "grad_norm": 1.171875,
      "learning_rate": 8.645454235739903e-07,
      "loss": 0.1666,
      "step": 2200
    },
    {
      "epoch": 6.897465886939571,
      "grad_norm": 2.546875,
      "learning_rate": 8.086466074476562e-07,
      "loss": 0.161,
      "step": 2210
    },
    {
      "epoch": 6.928654970760234,
      "grad_norm": 1.09375,
      "learning_rate": 7.545396638768698e-07,
      "loss": 0.1811,
      "step": 2220
    },
    {
      "epoch": 6.959844054580897,
      "grad_norm": 2.28125,
      "learning_rate": 7.022351411174866e-07,
      "loss": 0.1817,
      "step": 2230
    },
    {
      "epoch": 6.991033138401559,
      "grad_norm": 1.921875,
      "learning_rate": 6.517432360398556e-07,
      "loss": 0.1775,
      "step": 2240
    },
    {
      "epoch": 7.023001949317739,
      "grad_norm": 1.21875,
      "learning_rate": 6.030737921409169e-07,
      "loss": 0.1643,
      "step": 2250
    },
    {
      "epoch": 7.054191033138402,
      "grad_norm": 1.296875,
      "learning_rate": 5.562362976251901e-07,
      "loss": 0.1754,
      "step": 2260
    },
    {
      "epoch": 7.0853801169590644,
      "grad_norm": 1.25,
      "learning_rate": 5.112398835550348e-07,
      "loss": 0.1782,
      "step": 2270
    },
    {
      "epoch": 7.116569200779727,
      "grad_norm": 1.515625,
      "learning_rate": 4.6809332207053083e-07,
      "loss": 0.1734,
      "step": 2280
    },
    {
      "epoch": 7.14775828460039,
      "grad_norm": 1.0625,
      "learning_rate": 4.268050246793276e-07,
      "loss": 0.1805,
      "step": 2290
    },
    {
      "epoch": 7.178947368421053,
      "grad_norm": 1.484375,
      "learning_rate": 3.8738304061681107e-07,
      "loss": 0.176,
      "step": 2300
    },
    {
      "epoch": 7.2101364522417155,
      "grad_norm": 1.40625,
      "learning_rate": 3.498350552768859e-07,
      "loss": 0.1778,
      "step": 2310
    },
    {
      "epoch": 7.241325536062378,
      "grad_norm": 1.953125,
      "learning_rate": 3.1416838871368925e-07,
      "loss": 0.1722,
      "step": 2320
    },
    {
      "epoch": 7.272514619883041,
      "grad_norm": 1.4296875,
      "learning_rate": 2.8038999421453827e-07,
      "loss": 0.1726,
      "step": 2330
    },
    {
      "epoch": 7.303703703703704,
      "grad_norm": 1.7265625,
      "learning_rate": 2.4850645694436736e-07,
      "loss": 0.1867,
      "step": 2340
    },
    {
      "epoch": 7.334892787524367,
      "grad_norm": 1.0078125,
      "learning_rate": 2.1852399266194312e-07,
      "loss": 0.1828,
      "step": 2350
    },
    {
      "epoch": 7.366081871345029,
      "grad_norm": 1.4609375,
      "learning_rate": 1.9044844650808468e-07,
      "loss": 0.1706,
      "step": 2360
    },
    {
      "epoch": 7.397270955165692,
      "grad_norm": 0.9765625,
      "learning_rate": 1.6428529186614195e-07,
      "loss": 0.18,
      "step": 2370
    },
    {
      "epoch": 7.428460038986355,
      "grad_norm": 0.99609375,
      "learning_rate": 1.400396292949513e-07,
      "loss": 0.1686,
      "step": 2380
    },
    {
      "epoch": 7.459649122807018,
      "grad_norm": 1.3828125,
      "learning_rate": 1.1771618553447217e-07,
      "loss": 0.1755,
      "step": 2390
    },
    {
      "epoch": 7.49083820662768,
      "grad_norm": 1.2890625,
      "learning_rate": 9.731931258429638e-08,
      "loss": 0.1805,
      "step": 2400
    },
    {
      "epoch": 7.522027290448343,
      "grad_norm": 1.390625,
      "learning_rate": 7.885298685522235e-08,
      "loss": 0.1749,
      "step": 2410
    },
    {
      "epoch": 7.553216374269006,
      "grad_norm": 1.0390625,
      "learning_rate": 6.232080839403631e-08,
      "loss": 0.1756,
      "step": 2420
    },
    {
      "epoch": 7.584405458089669,
      "grad_norm": 0.91015625,
      "learning_rate": 4.772600018168816e-08,
      "loss": 0.1692,
      "step": 2430
    },
    {
      "epoch": 7.6155945419103315,
      "grad_norm": 1.5625,
      "learning_rate": 3.50714075049563e-08,
      "loss": 0.1676,
      "step": 2440
    },
    {
      "epoch": 7.646783625730994,
      "grad_norm": 2.953125,
      "learning_rate": 2.4359497401758026e-08,
      "loss": 0.1811,
      "step": 2450
    },
    {
      "epoch": 7.677972709551657,
      "grad_norm": 1.3046875,
      "learning_rate": 1.5592358180189782e-08,
      "loss": 0.18,
      "step": 2460
    },
    {
      "epoch": 7.70916179337232,
      "grad_norm": 1.4453125,
      "learning_rate": 8.771699011416169e-09,
      "loss": 0.1801,
      "step": 2470
    },
    {
      "epoch": 7.7403508771929825,
      "grad_norm": 1.0625,
      "learning_rate": 3.898849596456477e-09,
      "loss": 0.1835,
      "step": 2480
    },
    {
      "epoch": 7.771539961013645,
      "grad_norm": 1.4140625,
      "learning_rate": 9.74759906957612e-10,
      "loss": 0.1731,
      "step": 2490
    },
    {
      "epoch": 7.802729044834308,
      "grad_norm": 1.59375,
      "learning_rate": 0.0,
      "loss": 0.1847,
      "step": 2500
    }
  ],
  "logging_steps": 10,
  "max_steps": 2500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 8,
  "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": 9.5664155394048e+17,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}