{
  "best_metric": 0.5081329345703125,
  "best_model_checkpoint": "finetuned-cards-blackjack/checkpoint-2800",
  "epoch": 7.0,
  "eval_steps": 100,
  "global_step": 2891,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "grad_norm": 3.301379680633545,
      "learning_rate": 0.00019930819785541338,
      "loss": 1.3967,
      "step": 10
    },
    {
      "epoch": 0.05,
      "grad_norm": 3.8909592628479004,
      "learning_rate": 0.00019861639571082672,
      "loss": 1.3469,
      "step": 20
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.4435207843780518,
      "learning_rate": 0.00019792459356624006,
      "loss": 1.321,
      "step": 30
    },
    {
      "epoch": 0.1,
      "grad_norm": 4.026681900024414,
      "learning_rate": 0.0001972327914216534,
      "loss": 1.0812,
      "step": 40
    },
    {
      "epoch": 0.12,
      "grad_norm": 5.660863399505615,
      "learning_rate": 0.00019654098927706677,
      "loss": 1.2255,
      "step": 50
    },
    {
      "epoch": 0.15,
      "grad_norm": 4.201864719390869,
      "learning_rate": 0.00019584918713248014,
      "loss": 1.2845,
      "step": 60
    },
    {
      "epoch": 0.17,
      "grad_norm": 3.0525405406951904,
      "learning_rate": 0.00019515738498789345,
      "loss": 1.3223,
      "step": 70
    },
    {
      "epoch": 0.19,
      "grad_norm": 4.4655351638793945,
      "learning_rate": 0.00019446558284330682,
      "loss": 1.444,
      "step": 80
    },
    {
      "epoch": 0.22,
      "grad_norm": 5.199573993682861,
      "learning_rate": 0.00019377378069872016,
      "loss": 1.4201,
      "step": 90
    },
    {
      "epoch": 0.24,
      "grad_norm": 4.03210973739624,
      "learning_rate": 0.00019308197855413353,
      "loss": 1.3563,
      "step": 100
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.6749571183533448,
      "eval_loss": 1.1494646072387695,
      "eval_runtime": 6.2584,
      "eval_samples_per_second": 186.311,
      "eval_steps_per_second": 23.329,
      "step": 100
    },
    {
      "epoch": 0.27,
      "grad_norm": 4.278244495391846,
      "learning_rate": 0.00019239017640954688,
      "loss": 1.3488,
      "step": 110
    },
    {
      "epoch": 0.29,
      "grad_norm": 3.920788049697876,
      "learning_rate": 0.00019169837426496022,
      "loss": 1.451,
      "step": 120
    },
    {
      "epoch": 0.31,
      "grad_norm": 3.261601686477661,
      "learning_rate": 0.00019100657212037359,
      "loss": 1.2141,
      "step": 130
    },
    {
      "epoch": 0.34,
      "grad_norm": 5.404760837554932,
      "learning_rate": 0.00019031476997578695,
      "loss": 1.5746,
      "step": 140
    },
    {
      "epoch": 0.36,
      "grad_norm": 2.8406295776367188,
      "learning_rate": 0.0001896229678312003,
      "loss": 1.2777,
      "step": 150
    },
    {
      "epoch": 0.39,
      "grad_norm": 3.7745213508605957,
      "learning_rate": 0.00018893116568661364,
      "loss": 1.2075,
      "step": 160
    },
    {
      "epoch": 0.41,
      "grad_norm": 3.8692853450775146,
      "learning_rate": 0.00018823936354202698,
      "loss": 1.2082,
      "step": 170
    },
    {
      "epoch": 0.44,
      "grad_norm": 3.2764315605163574,
      "learning_rate": 0.00018754756139744035,
      "loss": 1.1009,
      "step": 180
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.6145033836364746,
      "learning_rate": 0.0001868557592528537,
      "loss": 1.2236,
      "step": 190
    },
    {
      "epoch": 0.48,
      "grad_norm": 4.715363502502441,
      "learning_rate": 0.00018616395710826703,
      "loss": 1.3393,
      "step": 200
    },
    {
      "epoch": 0.48,
      "eval_accuracy": 0.7204116638078902,
      "eval_loss": 1.0388233661651611,
      "eval_runtime": 6.0054,
      "eval_samples_per_second": 194.159,
      "eval_steps_per_second": 24.311,
      "step": 200
    },
    {
      "epoch": 0.51,
      "grad_norm": 3.9448986053466797,
      "learning_rate": 0.0001854721549636804,
      "loss": 1.1597,
      "step": 210
    },
    {
      "epoch": 0.53,
      "grad_norm": 3.357956886291504,
      "learning_rate": 0.00018478035281909374,
      "loss": 1.1734,
      "step": 220
    },
    {
      "epoch": 0.56,
      "grad_norm": 5.5605244636535645,
      "learning_rate": 0.0001840885506745071,
      "loss": 1.165,
      "step": 230
    },
    {
      "epoch": 0.58,
      "grad_norm": 4.48176908493042,
      "learning_rate": 0.00018339674852992045,
      "loss": 1.4025,
      "step": 240
    },
    {
      "epoch": 0.61,
      "grad_norm": 3.6814768314361572,
      "learning_rate": 0.0001827049463853338,
      "loss": 1.198,
      "step": 250
    },
    {
      "epoch": 0.63,
      "grad_norm": 4.112949848175049,
      "learning_rate": 0.00018201314424074716,
      "loss": 1.1062,
      "step": 260
    },
    {
      "epoch": 0.65,
      "grad_norm": 5.755402565002441,
      "learning_rate": 0.0001813213420961605,
      "loss": 1.144,
      "step": 270
    },
    {
      "epoch": 0.68,
      "grad_norm": 3.6004717350006104,
      "learning_rate": 0.00018062953995157384,
      "loss": 1.2527,
      "step": 280
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.4746742248535156,
      "learning_rate": 0.0001799377378069872,
      "loss": 1.1316,
      "step": 290
    },
    {
      "epoch": 0.73,
      "grad_norm": 2.231992483139038,
      "learning_rate": 0.00017924593566240055,
      "loss": 1.2033,
      "step": 300
    },
    {
      "epoch": 0.73,
      "eval_accuracy": 0.7547169811320755,
      "eval_loss": 0.9323562979698181,
      "eval_runtime": 7.9011,
      "eval_samples_per_second": 147.574,
      "eval_steps_per_second": 18.478,
      "step": 300
    },
    {
      "epoch": 0.75,
      "grad_norm": 4.073417663574219,
      "learning_rate": 0.00017855413351781392,
      "loss": 1.3702,
      "step": 310
    },
    {
      "epoch": 0.77,
      "grad_norm": 5.040902137756348,
      "learning_rate": 0.00017786233137322726,
      "loss": 1.1423,
      "step": 320
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.7068464756011963,
      "learning_rate": 0.0001771705292286406,
      "loss": 1.0609,
      "step": 330
    },
    {
      "epoch": 0.82,
      "grad_norm": 4.864231586456299,
      "learning_rate": 0.00017647872708405397,
      "loss": 1.201,
      "step": 340
    },
    {
      "epoch": 0.85,
      "grad_norm": 6.007138252258301,
      "learning_rate": 0.00017578692493946732,
      "loss": 1.0687,
      "step": 350
    },
    {
      "epoch": 0.87,
      "grad_norm": 4.0837225914001465,
      "learning_rate": 0.00017509512279488069,
      "loss": 1.1311,
      "step": 360
    },
    {
      "epoch": 0.9,
      "grad_norm": 4.566812992095947,
      "learning_rate": 0.00017440332065029403,
      "loss": 1.3071,
      "step": 370
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.3199901580810547,
      "learning_rate": 0.00017371151850570737,
      "loss": 1.0246,
      "step": 380
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.4883534908294678,
      "learning_rate": 0.00017301971636112074,
      "loss": 1.0215,
      "step": 390
    },
    {
      "epoch": 0.97,
      "grad_norm": 5.232284069061279,
      "learning_rate": 0.00017232791421653408,
      "loss": 0.9672,
      "step": 400
    },
    {
      "epoch": 0.97,
      "eval_accuracy": 0.7658662092624356,
      "eval_loss": 0.8557726144790649,
      "eval_runtime": 6.2462,
      "eval_samples_per_second": 186.674,
      "eval_steps_per_second": 23.374,
      "step": 400
    },
    {
      "epoch": 0.99,
      "grad_norm": 3.8225362300872803,
      "learning_rate": 0.00017163611207194742,
      "loss": 1.0908,
      "step": 410
    },
    {
      "epoch": 1.02,
      "grad_norm": 4.098091125488281,
      "learning_rate": 0.0001709443099273608,
      "loss": 0.8621,
      "step": 420
    },
    {
      "epoch": 1.04,
      "grad_norm": 4.027368068695068,
      "learning_rate": 0.00017025250778277413,
      "loss": 0.9868,
      "step": 430
    },
    {
      "epoch": 1.07,
      "grad_norm": 4.375247478485107,
      "learning_rate": 0.0001695607056381875,
      "loss": 1.0179,
      "step": 440
    },
    {
      "epoch": 1.09,
      "grad_norm": 8.204839706420898,
      "learning_rate": 0.00016886890349360084,
      "loss": 0.902,
      "step": 450
    },
    {
      "epoch": 1.11,
      "grad_norm": 3.1056785583496094,
      "learning_rate": 0.00016817710134901418,
      "loss": 0.9873,
      "step": 460
    },
    {
      "epoch": 1.14,
      "grad_norm": 6.811554908752441,
      "learning_rate": 0.00016748529920442755,
      "loss": 0.9035,
      "step": 470
    },
    {
      "epoch": 1.16,
      "grad_norm": 4.715181350708008,
      "learning_rate": 0.0001667934970598409,
      "loss": 1.0024,
      "step": 480
    },
    {
      "epoch": 1.19,
      "grad_norm": 5.355204105377197,
      "learning_rate": 0.00016610169491525423,
      "loss": 1.0384,
      "step": 490
    },
    {
      "epoch": 1.21,
      "grad_norm": 8.26843547821045,
      "learning_rate": 0.0001654098927706676,
      "loss": 0.8674,
      "step": 500
    },
    {
      "epoch": 1.21,
      "eval_accuracy": 0.7615780445969125,
      "eval_loss": 0.8456417322158813,
      "eval_runtime": 5.9836,
      "eval_samples_per_second": 194.865,
      "eval_steps_per_second": 24.4,
      "step": 500
    },
    {
      "epoch": 1.23,
      "grad_norm": 3.612718343734741,
      "learning_rate": 0.00016471809062608094,
      "loss": 1.035,
      "step": 510
    },
    {
      "epoch": 1.26,
      "grad_norm": 3.2531259059906006,
      "learning_rate": 0.0001640262884814943,
      "loss": 0.9591,
      "step": 520
    },
    {
      "epoch": 1.28,
      "grad_norm": 2.1132917404174805,
      "learning_rate": 0.00016333448633690765,
      "loss": 0.7013,
      "step": 530
    },
    {
      "epoch": 1.31,
      "grad_norm": 5.840766906738281,
      "learning_rate": 0.000162642684192321,
      "loss": 1.1066,
      "step": 540
    },
    {
      "epoch": 1.33,
      "grad_norm": 2.8128092288970947,
      "learning_rate": 0.00016195088204773436,
      "loss": 0.8851,
      "step": 550
    },
    {
      "epoch": 1.36,
      "grad_norm": 5.935888290405273,
      "learning_rate": 0.0001612590799031477,
      "loss": 0.9738,
      "step": 560
    },
    {
      "epoch": 1.38,
      "grad_norm": 4.2558488845825195,
      "learning_rate": 0.00016056727775856107,
      "loss": 1.1094,
      "step": 570
    },
    {
      "epoch": 1.4,
      "grad_norm": 3.7361583709716797,
      "learning_rate": 0.0001598754756139744,
      "loss": 1.0376,
      "step": 580
    },
    {
      "epoch": 1.43,
      "grad_norm": 3.6672043800354004,
      "learning_rate": 0.00015918367346938776,
      "loss": 0.9765,
      "step": 590
    },
    {
      "epoch": 1.45,
      "grad_norm": 2.8976941108703613,
      "learning_rate": 0.00015849187132480113,
      "loss": 0.8277,
      "step": 600
    },
    {
      "epoch": 1.45,
      "eval_accuracy": 0.7958833619210978,
      "eval_loss": 0.7562589049339294,
      "eval_runtime": 6.7504,
      "eval_samples_per_second": 172.731,
      "eval_steps_per_second": 21.628,
      "step": 600
    },
    {
      "epoch": 1.48,
      "grad_norm": 4.665554046630859,
      "learning_rate": 0.00015780006918021447,
      "loss": 0.8139,
      "step": 610
    },
    {
      "epoch": 1.5,
      "grad_norm": 4.166018486022949,
      "learning_rate": 0.0001571082670356278,
      "loss": 1.1314,
      "step": 620
    },
    {
      "epoch": 1.53,
      "grad_norm": 3.610258102416992,
      "learning_rate": 0.00015641646489104115,
      "loss": 0.9497,
      "step": 630
    },
    {
      "epoch": 1.55,
      "grad_norm": 4.610332489013672,
      "learning_rate": 0.00015572466274645452,
      "loss": 1.0767,
      "step": 640
    },
    {
      "epoch": 1.57,
      "grad_norm": 3.796252965927124,
      "learning_rate": 0.0001550328606018679,
      "loss": 0.8486,
      "step": 650
    },
    {
      "epoch": 1.6,
      "grad_norm": 3.9809694290161133,
      "learning_rate": 0.00015434105845728123,
      "loss": 0.9211,
      "step": 660
    },
    {
      "epoch": 1.62,
      "grad_norm": 2.5232605934143066,
      "learning_rate": 0.00015364925631269457,
      "loss": 0.8843,
      "step": 670
    },
    {
      "epoch": 1.65,
      "grad_norm": 4.975670337677002,
      "learning_rate": 0.00015295745416810794,
      "loss": 0.9494,
      "step": 680
    },
    {
      "epoch": 1.67,
      "grad_norm": 5.420626163482666,
      "learning_rate": 0.00015226565202352128,
      "loss": 0.9786,
      "step": 690
    },
    {
      "epoch": 1.69,
      "grad_norm": 3.364365339279175,
      "learning_rate": 0.00015157384987893465,
      "loss": 0.8703,
      "step": 700
    },
    {
      "epoch": 1.69,
      "eval_accuracy": 0.7538593481989708,
      "eval_loss": 0.8465284109115601,
      "eval_runtime": 6.2814,
      "eval_samples_per_second": 185.628,
      "eval_steps_per_second": 23.243,
      "step": 700
    },
    {
      "epoch": 1.72,
      "grad_norm": 3.51340913772583,
      "learning_rate": 0.00015088204773434796,
      "loss": 0.9032,
      "step": 710
    },
    {
      "epoch": 1.74,
      "grad_norm": 3.7203245162963867,
      "learning_rate": 0.00015019024558976133,
      "loss": 0.7729,
      "step": 720
    },
    {
      "epoch": 1.77,
      "grad_norm": 3.596214771270752,
      "learning_rate": 0.0001494984434451747,
      "loss": 0.8151,
      "step": 730
    },
    {
      "epoch": 1.79,
      "grad_norm": 3.2724595069885254,
      "learning_rate": 0.00014880664130058804,
      "loss": 0.8064,
      "step": 740
    },
    {
      "epoch": 1.82,
      "grad_norm": 3.5748846530914307,
      "learning_rate": 0.00014811483915600139,
      "loss": 0.8419,
      "step": 750
    },
    {
      "epoch": 1.84,
      "grad_norm": 5.998478412628174,
      "learning_rate": 0.00014742303701141473,
      "loss": 0.8494,
      "step": 760
    },
    {
      "epoch": 1.86,
      "grad_norm": 3.545043706893921,
      "learning_rate": 0.0001467312348668281,
      "loss": 0.7695,
      "step": 770
    },
    {
      "epoch": 1.89,
      "grad_norm": 3.9944069385528564,
      "learning_rate": 0.00014603943272224146,
      "loss": 0.9405,
      "step": 780
    },
    {
      "epoch": 1.91,
      "grad_norm": 5.435621738433838,
      "learning_rate": 0.00014534763057765478,
      "loss": 0.8863,
      "step": 790
    },
    {
      "epoch": 1.94,
      "grad_norm": 7.365724086761475,
      "learning_rate": 0.00014465582843306815,
      "loss": 0.893,
      "step": 800
    },
    {
      "epoch": 1.94,
      "eval_accuracy": 0.8001715265866209,
      "eval_loss": 0.688121497631073,
      "eval_runtime": 6.0254,
      "eval_samples_per_second": 193.514,
      "eval_steps_per_second": 24.231,
      "step": 800
    },
    {
      "epoch": 1.96,
      "grad_norm": 6.192987442016602,
      "learning_rate": 0.0001439640262884815,
      "loss": 0.8819,
      "step": 810
    },
    {
      "epoch": 1.99,
      "grad_norm": 3.021066188812256,
      "learning_rate": 0.00014327222414389486,
      "loss": 0.6413,
      "step": 820
    },
    {
      "epoch": 2.01,
      "grad_norm": 4.522083759307861,
      "learning_rate": 0.0001425804219993082,
      "loss": 0.9185,
      "step": 830
    },
    {
      "epoch": 2.03,
      "grad_norm": 3.089639186859131,
      "learning_rate": 0.00014188861985472154,
      "loss": 0.7384,
      "step": 840
    },
    {
      "epoch": 2.06,
      "grad_norm": 4.491950988769531,
      "learning_rate": 0.0001411968177101349,
      "loss": 0.765,
      "step": 850
    },
    {
      "epoch": 2.08,
      "grad_norm": 3.618821144104004,
      "learning_rate": 0.00014050501556554828,
      "loss": 0.8237,
      "step": 860
    },
    {
      "epoch": 2.11,
      "grad_norm": 4.773688793182373,
      "learning_rate": 0.00013981321342096162,
      "loss": 0.7171,
      "step": 870
    },
    {
      "epoch": 2.13,
      "grad_norm": 1.607408881187439,
      "learning_rate": 0.00013912141127637496,
      "loss": 0.6438,
      "step": 880
    },
    {
      "epoch": 2.15,
      "grad_norm": 4.511462211608887,
      "learning_rate": 0.0001384296091317883,
      "loss": 0.7983,
      "step": 890
    },
    {
      "epoch": 2.18,
      "grad_norm": 4.259463787078857,
      "learning_rate": 0.00013773780698720167,
      "loss": 0.9454,
      "step": 900
    },
    {
      "epoch": 2.18,
      "eval_accuracy": 0.8027444253859348,
      "eval_loss": 0.7210972905158997,
      "eval_runtime": 5.9658,
      "eval_samples_per_second": 195.449,
      "eval_steps_per_second": 24.473,
      "step": 900
    },
    {
      "epoch": 2.2,
      "grad_norm": 3.810264825820923,
      "learning_rate": 0.00013704600484261504,
      "loss": 0.7729,
      "step": 910
    },
    {
      "epoch": 2.23,
      "grad_norm": 5.475677967071533,
      "learning_rate": 0.00013635420269802835,
      "loss": 0.7531,
      "step": 920
    },
    {
      "epoch": 2.25,
      "grad_norm": 2.9276745319366455,
      "learning_rate": 0.00013566240055344172,
      "loss": 0.725,
      "step": 930
    },
    {
      "epoch": 2.28,
      "grad_norm": 4.840962886810303,
      "learning_rate": 0.00013497059840885506,
      "loss": 0.6938,
      "step": 940
    },
    {
      "epoch": 2.3,
      "grad_norm": 5.3595194816589355,
      "learning_rate": 0.00013427879626426843,
      "loss": 0.6863,
      "step": 950
    },
    {
      "epoch": 2.32,
      "grad_norm": 7.755936145782471,
      "learning_rate": 0.00013358699411968177,
      "loss": 0.7146,
      "step": 960
    },
    {
      "epoch": 2.35,
      "grad_norm": 3.4426372051239014,
      "learning_rate": 0.00013289519197509512,
      "loss": 0.7144,
      "step": 970
    },
    {
      "epoch": 2.37,
      "grad_norm": 4.554823398590088,
      "learning_rate": 0.00013220338983050849,
      "loss": 0.6512,
      "step": 980
    },
    {
      "epoch": 2.4,
      "grad_norm": 2.8689632415771484,
      "learning_rate": 0.00013151158768592183,
      "loss": 0.7966,
      "step": 990
    },
    {
      "epoch": 2.42,
      "grad_norm": 3.4381957054138184,
      "learning_rate": 0.0001308197855413352,
      "loss": 0.8109,
      "step": 1000
    },
    {
      "epoch": 2.42,
      "eval_accuracy": 0.8284734133790738,
      "eval_loss": 0.6368530988693237,
      "eval_runtime": 6.2777,
      "eval_samples_per_second": 185.738,
      "eval_steps_per_second": 23.257,
      "step": 1000
    },
    {
      "epoch": 2.45,
      "grad_norm": 2.984152317047119,
      "learning_rate": 0.00013012798339674854,
      "loss": 0.6477,
      "step": 1010
    },
    {
      "epoch": 2.47,
      "grad_norm": 5.486266613006592,
      "learning_rate": 0.00012943618125216188,
      "loss": 0.6834,
      "step": 1020
    },
    {
      "epoch": 2.49,
      "grad_norm": 2.2987334728240967,
      "learning_rate": 0.00012874437910757525,
      "loss": 0.6415,
      "step": 1030
    },
    {
      "epoch": 2.52,
      "grad_norm": 7.007256507873535,
      "learning_rate": 0.0001280525769629886,
      "loss": 0.6464,
      "step": 1040
    },
    {
      "epoch": 2.54,
      "grad_norm": 1.817421555519104,
      "learning_rate": 0.00012736077481840193,
      "loss": 0.774,
      "step": 1050
    },
    {
      "epoch": 2.57,
      "grad_norm": 4.492701530456543,
      "learning_rate": 0.0001266689726738153,
      "loss": 0.9331,
      "step": 1060
    },
    {
      "epoch": 2.59,
      "grad_norm": 3.884744644165039,
      "learning_rate": 0.00012597717052922864,
      "loss": 0.6107,
      "step": 1070
    },
    {
      "epoch": 2.62,
      "grad_norm": 4.274733066558838,
      "learning_rate": 0.000125285368384642,
      "loss": 0.6268,
      "step": 1080
    },
    {
      "epoch": 2.64,
      "grad_norm": 4.432763576507568,
      "learning_rate": 0.00012459356624005535,
      "loss": 0.6326,
      "step": 1090
    },
    {
      "epoch": 2.66,
      "grad_norm": 4.16074275970459,
      "learning_rate": 0.0001239017640954687,
      "loss": 0.8762,
      "step": 1100
    },
    {
      "epoch": 2.66,
      "eval_accuracy": 0.839622641509434,
      "eval_loss": 0.6335619688034058,
      "eval_runtime": 6.1128,
      "eval_samples_per_second": 190.748,
      "eval_steps_per_second": 23.884,
      "step": 1100
    },
    {
      "epoch": 2.69,
      "grad_norm": 4.018909931182861,
      "learning_rate": 0.00012320996195088206,
      "loss": 0.8039,
      "step": 1110
    },
    {
      "epoch": 2.71,
      "grad_norm": 8.111551284790039,
      "learning_rate": 0.0001225181598062954,
      "loss": 0.6436,
      "step": 1120
    },
    {
      "epoch": 2.74,
      "grad_norm": 4.22373628616333,
      "learning_rate": 0.00012182635766170876,
      "loss": 0.6228,
      "step": 1130
    },
    {
      "epoch": 2.76,
      "grad_norm": 4.817978858947754,
      "learning_rate": 0.00012113455551712211,
      "loss": 0.7047,
      "step": 1140
    },
    {
      "epoch": 2.78,
      "grad_norm": 5.471624851226807,
      "learning_rate": 0.00012044275337253545,
      "loss": 0.8293,
      "step": 1150
    },
    {
      "epoch": 2.81,
      "grad_norm": 3.491068124771118,
      "learning_rate": 0.00011975095122794881,
      "loss": 0.7128,
      "step": 1160
    },
    {
      "epoch": 2.83,
      "grad_norm": 4.463800430297852,
      "learning_rate": 0.00011905914908336215,
      "loss": 0.7569,
      "step": 1170
    },
    {
      "epoch": 2.86,
      "grad_norm": 2.7582342624664307,
      "learning_rate": 0.00011836734693877552,
      "loss": 0.6774,
      "step": 1180
    },
    {
      "epoch": 2.88,
      "grad_norm": 4.606247901916504,
      "learning_rate": 0.00011767554479418887,
      "loss": 0.7384,
      "step": 1190
    },
    {
      "epoch": 2.91,
      "grad_norm": 4.3657660484313965,
      "learning_rate": 0.00011698374264960222,
      "loss": 0.8034,
      "step": 1200
    },
    {
      "epoch": 2.91,
      "eval_accuracy": 0.8164665523156089,
      "eval_loss": 0.657957911491394,
      "eval_runtime": 6.0796,
      "eval_samples_per_second": 191.79,
      "eval_steps_per_second": 24.015,
      "step": 1200
    },
    {
      "epoch": 2.93,
      "grad_norm": 3.4329655170440674,
      "learning_rate": 0.00011629194050501557,
      "loss": 0.733,
      "step": 1210
    },
    {
      "epoch": 2.95,
      "grad_norm": 4.565850257873535,
      "learning_rate": 0.00011560013836042894,
      "loss": 0.7215,
      "step": 1220
    },
    {
      "epoch": 2.98,
      "grad_norm": 3.225835084915161,
      "learning_rate": 0.00011490833621584227,
      "loss": 0.6895,
      "step": 1230
    },
    {
      "epoch": 3.0,
      "grad_norm": 5.188159942626953,
      "learning_rate": 0.00011421653407125564,
      "loss": 0.6821,
      "step": 1240
    },
    {
      "epoch": 3.03,
      "grad_norm": 1.3620127439498901,
      "learning_rate": 0.00011352473192666896,
      "loss": 0.5604,
      "step": 1250
    },
    {
      "epoch": 3.05,
      "grad_norm": 3.446960687637329,
      "learning_rate": 0.00011283292978208233,
      "loss": 0.5554,
      "step": 1260
    },
    {
      "epoch": 3.08,
      "grad_norm": 5.016846656799316,
      "learning_rate": 0.00011214112763749569,
      "loss": 0.7156,
      "step": 1270
    },
    {
      "epoch": 3.1,
      "grad_norm": 5.600184440612793,
      "learning_rate": 0.00011144932549290903,
      "loss": 0.745,
      "step": 1280
    },
    {
      "epoch": 3.12,
      "grad_norm": 3.7640082836151123,
      "learning_rate": 0.00011075752334832239,
      "loss": 0.6316,
      "step": 1290
    },
    {
      "epoch": 3.15,
      "grad_norm": 4.837277889251709,
      "learning_rate": 0.00011006572120373573,
      "loss": 0.5833,
      "step": 1300
    },
    {
      "epoch": 3.15,
      "eval_accuracy": 0.8439108061749572,
      "eval_loss": 0.5827564597129822,
      "eval_runtime": 6.3009,
      "eval_samples_per_second": 185.052,
      "eval_steps_per_second": 23.171,
      "step": 1300
    },
    {
      "epoch": 3.17,
      "grad_norm": 5.6047797203063965,
      "learning_rate": 0.00010937391905914908,
      "loss": 0.7184,
      "step": 1310
    },
    {
      "epoch": 3.2,
      "grad_norm": 4.177833080291748,
      "learning_rate": 0.0001087512971290211,
      "loss": 0.7094,
      "step": 1320
    },
    {
      "epoch": 3.22,
      "grad_norm": 2.0531811714172363,
      "learning_rate": 0.00010805949498443447,
      "loss": 0.6679,
      "step": 1330
    },
    {
      "epoch": 3.24,
      "grad_norm": 2.919313430786133,
      "learning_rate": 0.0001073676928398478,
      "loss": 0.4368,
      "step": 1340
    },
    {
      "epoch": 3.27,
      "grad_norm": 5.47185754776001,
      "learning_rate": 0.00010667589069526116,
      "loss": 0.5666,
      "step": 1350
    },
    {
      "epoch": 3.29,
      "grad_norm": 5.082462310791016,
      "learning_rate": 0.00010598408855067452,
      "loss": 0.7758,
      "step": 1360
    },
    {
      "epoch": 3.32,
      "grad_norm": 3.3282408714294434,
      "learning_rate": 0.00010529228640608786,
      "loss": 0.6055,
      "step": 1370
    },
    {
      "epoch": 3.34,
      "grad_norm": 5.19661808013916,
      "learning_rate": 0.00010460048426150121,
      "loss": 0.6678,
      "step": 1380
    },
    {
      "epoch": 3.37,
      "grad_norm": 2.5423412322998047,
      "learning_rate": 0.00010390868211691456,
      "loss": 0.5333,
      "step": 1390
    },
    {
      "epoch": 3.39,
      "grad_norm": 9.068185806274414,
      "learning_rate": 0.00010321687997232792,
      "loss": 0.8811,
      "step": 1400
    },
    {
      "epoch": 3.39,
      "eval_accuracy": 0.8259005145797599,
      "eval_loss": 0.6564387679100037,
      "eval_runtime": 5.9782,
      "eval_samples_per_second": 195.041,
      "eval_steps_per_second": 24.422,
      "step": 1400
    },
    {
      "epoch": 3.41,
      "grad_norm": 5.794499397277832,
      "learning_rate": 0.00010252507782774128,
      "loss": 0.8535,
      "step": 1410
    },
    {
      "epoch": 3.44,
      "grad_norm": 4.731594562530518,
      "learning_rate": 0.00010183327568315462,
      "loss": 0.6556,
      "step": 1420
    },
    {
      "epoch": 3.46,
      "grad_norm": 3.823868751525879,
      "learning_rate": 0.00010114147353856798,
      "loss": 0.687,
      "step": 1430
    },
    {
      "epoch": 3.49,
      "grad_norm": 7.72351598739624,
      "learning_rate": 0.00010044967139398133,
      "loss": 0.6479,
      "step": 1440
    },
    {
      "epoch": 3.51,
      "grad_norm": 5.026217937469482,
      "learning_rate": 9.975786924939467e-05,
      "loss": 0.64,
      "step": 1450
    },
    {
      "epoch": 3.54,
      "grad_norm": 2.873476028442383,
      "learning_rate": 9.906606710480803e-05,
      "loss": 0.6692,
      "step": 1460
    },
    {
      "epoch": 3.56,
      "grad_norm": 3.9772098064422607,
      "learning_rate": 9.837426496022138e-05,
      "loss": 0.8156,
      "step": 1470
    },
    {
      "epoch": 3.58,
      "grad_norm": 5.044854164123535,
      "learning_rate": 9.768246281563474e-05,
      "loss": 0.7261,
      "step": 1480
    },
    {
      "epoch": 3.61,
      "grad_norm": 2.8663127422332764,
      "learning_rate": 9.699066067104808e-05,
      "loss": 0.7608,
      "step": 1490
    },
    {
      "epoch": 3.63,
      "grad_norm": 7.623239040374756,
      "learning_rate": 9.629885852646143e-05,
      "loss": 0.5639,
      "step": 1500
    },
    {
      "epoch": 3.63,
      "eval_accuracy": 0.8439108061749572,
      "eval_loss": 0.5736597180366516,
      "eval_runtime": 6.1394,
      "eval_samples_per_second": 189.92,
      "eval_steps_per_second": 23.781,
      "step": 1500
    },
    {
      "epoch": 3.66,
      "grad_norm": 4.9297966957092285,
      "learning_rate": 9.560705638187479e-05,
      "loss": 0.6672,
      "step": 1510
    },
    {
      "epoch": 3.68,
      "grad_norm": 2.0875468254089355,
      "learning_rate": 9.491525423728815e-05,
      "loss": 0.5335,
      "step": 1520
    },
    {
      "epoch": 3.7,
      "grad_norm": 10.350793838500977,
      "learning_rate": 9.422345209270149e-05,
      "loss": 0.7163,
      "step": 1530
    },
    {
      "epoch": 3.73,
      "grad_norm": 4.162230014801025,
      "learning_rate": 9.353164994811484e-05,
      "loss": 0.7528,
      "step": 1540
    },
    {
      "epoch": 3.75,
      "grad_norm": 5.249913215637207,
      "learning_rate": 9.28398478035282e-05,
      "loss": 0.7946,
      "step": 1550
    },
    {
      "epoch": 3.78,
      "grad_norm": 3.5188651084899902,
      "learning_rate": 9.214804565894155e-05,
      "loss": 0.7108,
      "step": 1560
    },
    {
      "epoch": 3.8,
      "grad_norm": 5.497143268585205,
      "learning_rate": 9.14562435143549e-05,
      "loss": 0.8685,
      "step": 1570
    },
    {
      "epoch": 3.83,
      "grad_norm": 4.383511066436768,
      "learning_rate": 9.076444136976825e-05,
      "loss": 0.7093,
      "step": 1580
    },
    {
      "epoch": 3.85,
      "grad_norm": 4.953444957733154,
      "learning_rate": 9.00726392251816e-05,
      "loss": 0.7253,
      "step": 1590
    },
    {
      "epoch": 3.87,
      "grad_norm": 3.61757230758667,
      "learning_rate": 8.938083708059496e-05,
      "loss": 0.639,
      "step": 1600
    },
    {
      "epoch": 3.87,
      "eval_accuracy": 0.8379073756432247,
      "eval_loss": 0.560886561870575,
      "eval_runtime": 5.9489,
      "eval_samples_per_second": 196.002,
      "eval_steps_per_second": 24.542,
      "step": 1600
    },
    {
      "epoch": 3.9,
      "grad_norm": 5.650317668914795,
      "learning_rate": 8.868903493600831e-05,
      "loss": 0.8287,
      "step": 1610
    },
    {
      "epoch": 3.92,
      "grad_norm": 6.012249946594238,
      "learning_rate": 8.799723279142166e-05,
      "loss": 0.6976,
      "step": 1620
    },
    {
      "epoch": 3.95,
      "grad_norm": 5.186240196228027,
      "learning_rate": 8.730543064683501e-05,
      "loss": 0.7452,
      "step": 1630
    },
    {
      "epoch": 3.97,
      "grad_norm": 5.2669572830200195,
      "learning_rate": 8.661362850224835e-05,
      "loss": 0.8586,
      "step": 1640
    },
    {
      "epoch": 4.0,
      "grad_norm": 2.4105117321014404,
      "learning_rate": 8.592182635766172e-05,
      "loss": 0.6194,
      "step": 1650
    },
    {
      "epoch": 4.02,
      "grad_norm": 1.2886375188827515,
      "learning_rate": 8.523002421307506e-05,
      "loss": 0.4969,
      "step": 1660
    },
    {
      "epoch": 4.04,
      "grad_norm": 1.915207862854004,
      "learning_rate": 8.453822206848842e-05,
      "loss": 0.571,
      "step": 1670
    },
    {
      "epoch": 4.07,
      "grad_norm": 3.7422375679016113,
      "learning_rate": 8.384641992390176e-05,
      "loss": 0.6791,
      "step": 1680
    },
    {
      "epoch": 4.09,
      "grad_norm": 5.4421467781066895,
      "learning_rate": 8.315461777931513e-05,
      "loss": 0.6829,
      "step": 1690
    },
    {
      "epoch": 4.12,
      "grad_norm": 1.9872852563858032,
      "learning_rate": 8.246281563472847e-05,
      "loss": 0.6455,
      "step": 1700
    },
    {
      "epoch": 4.12,
      "eval_accuracy": 0.8370497427101201,
      "eval_loss": 0.5820054411888123,
      "eval_runtime": 6.2231,
      "eval_samples_per_second": 187.366,
      "eval_steps_per_second": 23.461,
      "step": 1700
    },
    {
      "epoch": 4.14,
      "grad_norm": 3.1257195472717285,
      "learning_rate": 8.177101349014182e-05,
      "loss": 0.6619,
      "step": 1710
    },
    {
      "epoch": 4.16,
      "grad_norm": 3.6743292808532715,
      "learning_rate": 8.107921134555517e-05,
      "loss": 0.8357,
      "step": 1720
    },
    {
      "epoch": 4.19,
      "grad_norm": 3.7856836318969727,
      "learning_rate": 8.038740920096852e-05,
      "loss": 0.6017,
      "step": 1730
    },
    {
      "epoch": 4.21,
      "grad_norm": 4.6526970863342285,
      "learning_rate": 7.969560705638188e-05,
      "loss": 0.6805,
      "step": 1740
    },
    {
      "epoch": 4.24,
      "grad_norm": 3.4002761840820312,
      "learning_rate": 7.900380491179523e-05,
      "loss": 0.5558,
      "step": 1750
    },
    {
      "epoch": 4.26,
      "grad_norm": 3.9795327186584473,
      "learning_rate": 7.831200276720859e-05,
      "loss": 0.6921,
      "step": 1760
    },
    {
      "epoch": 4.29,
      "grad_norm": 3.5085155963897705,
      "learning_rate": 7.762020062262193e-05,
      "loss": 0.5476,
      "step": 1770
    },
    {
      "epoch": 4.31,
      "grad_norm": 5.0314412117004395,
      "learning_rate": 7.69283984780353e-05,
      "loss": 0.8566,
      "step": 1780
    },
    {
      "epoch": 4.33,
      "grad_norm": 2.536855697631836,
      "learning_rate": 7.623659633344864e-05,
      "loss": 0.5743,
      "step": 1790
    },
    {
      "epoch": 4.36,
      "grad_norm": 4.995050430297852,
      "learning_rate": 7.5544794188862e-05,
      "loss": 0.5402,
      "step": 1800
    },
    {
      "epoch": 4.36,
      "eval_accuracy": 0.8344768439108061,
      "eval_loss": 0.5796906352043152,
      "eval_runtime": 6.1279,
      "eval_samples_per_second": 190.278,
      "eval_steps_per_second": 23.826,
      "step": 1800
    },
    {
      "epoch": 4.38,
      "grad_norm": 4.143467903137207,
      "learning_rate": 7.485299204427533e-05,
      "loss": 0.6715,
      "step": 1810
    },
    {
      "epoch": 4.41,
      "grad_norm": 1.8152028322219849,
      "learning_rate": 7.416118989968869e-05,
      "loss": 0.6965,
      "step": 1820
    },
    {
      "epoch": 4.43,
      "grad_norm": 3.699620485305786,
      "learning_rate": 7.346938775510205e-05,
      "loss": 0.5758,
      "step": 1830
    },
    {
      "epoch": 4.46,
      "grad_norm": 2.2266180515289307,
      "learning_rate": 7.27775856105154e-05,
      "loss": 0.6802,
      "step": 1840
    },
    {
      "epoch": 4.48,
      "grad_norm": 4.586669445037842,
      "learning_rate": 7.208578346592874e-05,
      "loss": 0.5885,
      "step": 1850
    },
    {
      "epoch": 4.5,
      "grad_norm": 4.72069787979126,
      "learning_rate": 7.13939813213421e-05,
      "loss": 0.6404,
      "step": 1860
    },
    {
      "epoch": 4.53,
      "grad_norm": 5.436990261077881,
      "learning_rate": 7.070217917675545e-05,
      "loss": 0.7781,
      "step": 1870
    },
    {
      "epoch": 4.55,
      "grad_norm": 4.715204238891602,
      "learning_rate": 7.001037703216881e-05,
      "loss": 0.7109,
      "step": 1880
    },
    {
      "epoch": 4.58,
      "grad_norm": 3.261801242828369,
      "learning_rate": 6.931857488758215e-05,
      "loss": 0.5707,
      "step": 1890
    },
    {
      "epoch": 4.6,
      "grad_norm": 3.516954183578491,
      "learning_rate": 6.86267727429955e-05,
      "loss": 0.5311,
      "step": 1900
    },
    {
      "epoch": 4.6,
      "eval_accuracy": 0.8456260720411664,
      "eval_loss": 0.55106520652771,
      "eval_runtime": 6.3501,
      "eval_samples_per_second": 183.618,
      "eval_steps_per_second": 22.992,
      "step": 1900
    },
    {
      "epoch": 4.62,
      "grad_norm": 4.697694778442383,
      "learning_rate": 6.793497059840886e-05,
      "loss": 0.6169,
      "step": 1910
    },
    {
      "epoch": 4.65,
      "grad_norm": 4.627555847167969,
      "learning_rate": 6.724316845382221e-05,
      "loss": 0.649,
      "step": 1920
    },
    {
      "epoch": 4.67,
      "grad_norm": 3.16441011428833,
      "learning_rate": 6.662054652369423e-05,
      "loss": 0.6538,
      "step": 1930
    },
    {
      "epoch": 4.7,
      "grad_norm": 3.6413562297821045,
      "learning_rate": 6.592874437910757e-05,
      "loss": 0.5963,
      "step": 1940
    },
    {
      "epoch": 4.72,
      "grad_norm": 4.7628960609436035,
      "learning_rate": 6.523694223452093e-05,
      "loss": 0.5106,
      "step": 1950
    },
    {
      "epoch": 4.75,
      "grad_norm": 3.3812217712402344,
      "learning_rate": 6.454514008993428e-05,
      "loss": 0.6908,
      "step": 1960
    },
    {
      "epoch": 4.77,
      "grad_norm": 3.9284725189208984,
      "learning_rate": 6.385333794534764e-05,
      "loss": 0.4863,
      "step": 1970
    },
    {
      "epoch": 4.79,
      "grad_norm": 3.633194923400879,
      "learning_rate": 6.316153580076098e-05,
      "loss": 0.4383,
      "step": 1980
    },
    {
      "epoch": 4.82,
      "grad_norm": 6.324495792388916,
      "learning_rate": 6.246973365617433e-05,
      "loss": 0.8217,
      "step": 1990
    },
    {
      "epoch": 4.84,
      "grad_norm": 5.055554389953613,
      "learning_rate": 6.177793151158769e-05,
      "loss": 0.5734,
      "step": 2000
    },
    {
      "epoch": 4.84,
      "eval_accuracy": 0.8507718696397941,
      "eval_loss": 0.5443547964096069,
      "eval_runtime": 5.9845,
      "eval_samples_per_second": 194.837,
      "eval_steps_per_second": 24.396,
      "step": 2000
    },
    {
      "epoch": 4.87,
      "grad_norm": 3.0936367511749268,
      "learning_rate": 6.108612936700104e-05,
      "loss": 0.6563,
      "step": 2010
    },
    {
      "epoch": 4.89,
      "grad_norm": 3.478715181350708,
      "learning_rate": 6.039432722241439e-05,
      "loss": 0.595,
      "step": 2020
    },
    {
      "epoch": 4.92,
      "grad_norm": 3.4817001819610596,
      "learning_rate": 5.970252507782774e-05,
      "loss": 0.3782,
      "step": 2030
    },
    {
      "epoch": 4.94,
      "grad_norm": 6.603343963623047,
      "learning_rate": 5.901072293324109e-05,
      "loss": 0.5735,
      "step": 2040
    },
    {
      "epoch": 4.96,
      "grad_norm": 1.6107616424560547,
      "learning_rate": 5.831892078865445e-05,
      "loss": 0.6045,
      "step": 2050
    },
    {
      "epoch": 4.99,
      "grad_norm": 4.367840766906738,
      "learning_rate": 5.76271186440678e-05,
      "loss": 0.5533,
      "step": 2060
    },
    {
      "epoch": 5.01,
      "grad_norm": 6.6618266105651855,
      "learning_rate": 5.6935316499481154e-05,
      "loss": 0.4114,
      "step": 2070
    },
    {
      "epoch": 5.04,
      "grad_norm": 7.852776050567627,
      "learning_rate": 5.62435143548945e-05,
      "loss": 0.642,
      "step": 2080
    },
    {
      "epoch": 5.06,
      "grad_norm": 5.769771099090576,
      "learning_rate": 5.555171221030785e-05,
      "loss": 0.5934,
      "step": 2090
    },
    {
      "epoch": 5.08,
      "grad_norm": 2.4635396003723145,
      "learning_rate": 5.485991006572121e-05,
      "loss": 0.5206,
      "step": 2100
    },
    {
      "epoch": 5.08,
      "eval_accuracy": 0.8636363636363636,
      "eval_loss": 0.5326434969902039,
      "eval_runtime": 6.1606,
      "eval_samples_per_second": 189.267,
      "eval_steps_per_second": 23.699,
      "step": 2100
    },
    {
      "epoch": 5.11,
      "grad_norm": 4.699713706970215,
      "learning_rate": 5.416810792113456e-05,
      "loss": 0.6311,
      "step": 2110
    },
    {
      "epoch": 5.13,
      "grad_norm": 3.2119288444519043,
      "learning_rate": 5.347630577654791e-05,
      "loss": 0.4781,
      "step": 2120
    },
    {
      "epoch": 5.16,
      "grad_norm": 5.083879470825195,
      "learning_rate": 5.278450363196126e-05,
      "loss": 0.5123,
      "step": 2130
    },
    {
      "epoch": 5.18,
      "grad_norm": 3.2444283962249756,
      "learning_rate": 5.209270148737462e-05,
      "loss": 0.39,
      "step": 2140
    },
    {
      "epoch": 5.21,
      "grad_norm": 2.9540326595306396,
      "learning_rate": 5.140089934278797e-05,
      "loss": 0.5081,
      "step": 2150
    },
    {
      "epoch": 5.23,
      "grad_norm": 4.055675029754639,
      "learning_rate": 5.0709097198201316e-05,
      "loss": 0.5884,
      "step": 2160
    },
    {
      "epoch": 5.25,
      "grad_norm": 2.7214150428771973,
      "learning_rate": 5.0017295053614664e-05,
      "loss": 0.5241,
      "step": 2170
    },
    {
      "epoch": 5.28,
      "grad_norm": 1.249835729598999,
      "learning_rate": 4.932549290902802e-05,
      "loss": 0.3994,
      "step": 2180
    },
    {
      "epoch": 5.3,
      "grad_norm": 5.9494829177856445,
      "learning_rate": 4.863369076444137e-05,
      "loss": 0.5295,
      "step": 2190
    },
    {
      "epoch": 5.33,
      "grad_norm": 3.318251371383667,
      "learning_rate": 4.794188861985472e-05,
      "loss": 0.6272,
      "step": 2200
    },
    {
      "epoch": 5.33,
      "eval_accuracy": 0.8524871355060034,
      "eval_loss": 0.5477628707885742,
      "eval_runtime": 6.3218,
      "eval_samples_per_second": 184.442,
      "eval_steps_per_second": 23.095,
      "step": 2200
    },
    {
      "epoch": 5.35,
      "grad_norm": 4.580955505371094,
      "learning_rate": 4.725008647526807e-05,
      "loss": 0.5204,
      "step": 2210
    },
    {
      "epoch": 5.38,
      "grad_norm": 1.7544004917144775,
      "learning_rate": 4.6558284330681426e-05,
      "loss": 0.5388,
      "step": 2220
    },
    {
      "epoch": 5.4,
      "grad_norm": 4.454672336578369,
      "learning_rate": 4.586648218609478e-05,
      "loss": 0.4773,
      "step": 2230
    },
    {
      "epoch": 5.42,
      "grad_norm": 7.039458274841309,
      "learning_rate": 4.517468004150813e-05,
      "loss": 0.5821,
      "step": 2240
    },
    {
      "epoch": 5.45,
      "grad_norm": 4.6715006828308105,
      "learning_rate": 4.4482877896921485e-05,
      "loss": 0.5881,
      "step": 2250
    },
    {
      "epoch": 5.47,
      "grad_norm": 3.3161721229553223,
      "learning_rate": 4.379107575233484e-05,
      "loss": 0.5587,
      "step": 2260
    },
    {
      "epoch": 5.5,
      "grad_norm": 5.245398044586182,
      "learning_rate": 4.309927360774819e-05,
      "loss": 0.4973,
      "step": 2270
    },
    {
      "epoch": 5.52,
      "grad_norm": 3.1016721725463867,
      "learning_rate": 4.2407471463161536e-05,
      "loss": 0.4004,
      "step": 2280
    },
    {
      "epoch": 5.54,
      "grad_norm": 4.883015155792236,
      "learning_rate": 4.171566931857489e-05,
      "loss": 0.4267,
      "step": 2290
    },
    {
      "epoch": 5.57,
      "grad_norm": 4.550380229949951,
      "learning_rate": 4.102386717398824e-05,
      "loss": 0.5124,
      "step": 2300
    },
    {
      "epoch": 5.57,
      "eval_accuracy": 0.8687821612349914,
      "eval_loss": 0.5295912623405457,
      "eval_runtime": 6.0432,
      "eval_samples_per_second": 192.944,
      "eval_steps_per_second": 24.159,
      "step": 2300
    },
    {
      "epoch": 5.59,
      "grad_norm": 3.2469303607940674,
      "learning_rate": 4.0332065029401595e-05,
      "loss": 0.4733,
      "step": 2310
    },
    {
      "epoch": 5.62,
      "grad_norm": 5.656401634216309,
      "learning_rate": 3.964026288481494e-05,
      "loss": 0.5978,
      "step": 2320
    },
    {
      "epoch": 5.64,
      "grad_norm": 2.7476541996002197,
      "learning_rate": 3.89484607402283e-05,
      "loss": 0.3513,
      "step": 2330
    },
    {
      "epoch": 5.67,
      "grad_norm": 4.047815322875977,
      "learning_rate": 3.825665859564165e-05,
      "loss": 0.5287,
      "step": 2340
    },
    {
      "epoch": 5.69,
      "grad_norm": 3.4885923862457275,
      "learning_rate": 3.7564856451055e-05,
      "loss": 0.4946,
      "step": 2350
    },
    {
      "epoch": 5.71,
      "grad_norm": 7.513520240783691,
      "learning_rate": 3.687305430646835e-05,
      "loss": 0.6233,
      "step": 2360
    },
    {
      "epoch": 5.74,
      "grad_norm": 2.3985989093780518,
      "learning_rate": 3.61812521618817e-05,
      "loss": 0.5149,
      "step": 2370
    },
    {
      "epoch": 5.76,
      "grad_norm": 5.046018123626709,
      "learning_rate": 3.5489450017295054e-05,
      "loss": 0.4948,
      "step": 2380
    },
    {
      "epoch": 5.79,
      "grad_norm": 2.6082875728607178,
      "learning_rate": 3.479764787270841e-05,
      "loss": 0.6084,
      "step": 2390
    },
    {
      "epoch": 5.81,
      "grad_norm": 2.541283369064331,
      "learning_rate": 3.410584572812176e-05,
      "loss": 0.5659,
      "step": 2400
    },
    {
      "epoch": 5.81,
      "eval_accuracy": 0.8704974271012007,
      "eval_loss": 0.5180826783180237,
      "eval_runtime": 6.1391,
      "eval_samples_per_second": 189.929,
      "eval_steps_per_second": 23.782,
      "step": 2400
    },
    {
      "epoch": 5.84,
      "grad_norm": 2.853994846343994,
      "learning_rate": 3.341404358353511e-05,
      "loss": 0.6081,
      "step": 2410
    },
    {
      "epoch": 5.86,
      "grad_norm": 4.628828525543213,
      "learning_rate": 3.272224143894847e-05,
      "loss": 0.4588,
      "step": 2420
    },
    {
      "epoch": 5.88,
      "grad_norm": 3.1006319522857666,
      "learning_rate": 3.2030439294361816e-05,
      "loss": 0.4341,
      "step": 2430
    },
    {
      "epoch": 5.91,
      "grad_norm": 2.395719528198242,
      "learning_rate": 3.133863714977517e-05,
      "loss": 0.442,
      "step": 2440
    },
    {
      "epoch": 5.93,
      "grad_norm": 3.238839864730835,
      "learning_rate": 3.064683500518852e-05,
      "loss": 0.4359,
      "step": 2450
    },
    {
      "epoch": 5.96,
      "grad_norm": 5.706843852996826,
      "learning_rate": 2.9955032860601867e-05,
      "loss": 0.5139,
      "step": 2460
    },
    {
      "epoch": 5.98,
      "grad_norm": 6.059083461761475,
      "learning_rate": 2.9263230716015223e-05,
      "loss": 0.4459,
      "step": 2470
    },
    {
      "epoch": 6.0,
      "grad_norm": 4.164783954620361,
      "learning_rate": 2.857142857142857e-05,
      "loss": 0.5037,
      "step": 2480
    },
    {
      "epoch": 6.03,
      "grad_norm": 3.230203151702881,
      "learning_rate": 2.7879626426841926e-05,
      "loss": 0.4225,
      "step": 2490
    },
    {
      "epoch": 6.05,
      "grad_norm": 5.467190742492676,
      "learning_rate": 2.7187824282255274e-05,
      "loss": 0.4212,
      "step": 2500
    },
    {
      "epoch": 6.05,
      "eval_accuracy": 0.8610634648370498,
      "eval_loss": 0.5200443267822266,
      "eval_runtime": 6.2608,
      "eval_samples_per_second": 186.239,
      "eval_steps_per_second": 23.32,
      "step": 2500
    },
    {
      "epoch": 6.08,
      "grad_norm": 3.7668442726135254,
      "learning_rate": 2.649602213766863e-05,
      "loss": 0.4042,
      "step": 2510
    },
    {
      "epoch": 6.1,
      "grad_norm": 3.094477415084839,
      "learning_rate": 2.580421999308198e-05,
      "loss": 0.4338,
      "step": 2520
    },
    {
      "epoch": 6.13,
      "grad_norm": 5.538024425506592,
      "learning_rate": 2.5112417848495333e-05,
      "loss": 0.3269,
      "step": 2530
    },
    {
      "epoch": 6.15,
      "grad_norm": 5.658746719360352,
      "learning_rate": 2.4420615703908685e-05,
      "loss": 0.4719,
      "step": 2540
    },
    {
      "epoch": 6.17,
      "grad_norm": 1.6886987686157227,
      "learning_rate": 2.3728813559322036e-05,
      "loss": 0.395,
      "step": 2550
    },
    {
      "epoch": 6.2,
      "grad_norm": 3.538180112838745,
      "learning_rate": 2.3037011414735388e-05,
      "loss": 0.2877,
      "step": 2560
    },
    {
      "epoch": 6.22,
      "grad_norm": 2.9912898540496826,
      "learning_rate": 2.234520927014874e-05,
      "loss": 0.4797,
      "step": 2570
    },
    {
      "epoch": 6.25,
      "grad_norm": 2.68037748336792,
      "learning_rate": 2.1653407125562088e-05,
      "loss": 0.5114,
      "step": 2580
    },
    {
      "epoch": 6.27,
      "grad_norm": 5.079796314239502,
      "learning_rate": 2.096160498097544e-05,
      "loss": 0.3604,
      "step": 2590
    },
    {
      "epoch": 6.3,
      "grad_norm": 3.052543878555298,
      "learning_rate": 2.026980283638879e-05,
      "loss": 0.4338,
      "step": 2600
    },
    {
      "epoch": 6.3,
      "eval_accuracy": 0.8730703259005146,
      "eval_loss": 0.5135151743888855,
      "eval_runtime": 5.9846,
      "eval_samples_per_second": 194.834,
      "eval_steps_per_second": 24.396,
      "step": 2600
    },
    {
      "epoch": 6.32,
      "grad_norm": 5.780861854553223,
      "learning_rate": 1.9578000691802147e-05,
      "loss": 0.3725,
      "step": 2610
    },
    {
      "epoch": 6.34,
      "grad_norm": 4.87053108215332,
      "learning_rate": 1.88861985472155e-05,
      "loss": 0.2491,
      "step": 2620
    },
    {
      "epoch": 6.37,
      "grad_norm": 2.2995293140411377,
      "learning_rate": 1.819439640262885e-05,
      "loss": 0.2911,
      "step": 2630
    },
    {
      "epoch": 6.39,
      "grad_norm": 1.6383118629455566,
      "learning_rate": 1.7502594258042202e-05,
      "loss": 0.2562,
      "step": 2640
    },
    {
      "epoch": 6.42,
      "grad_norm": 4.9596991539001465,
      "learning_rate": 1.6810792113455554e-05,
      "loss": 0.5795,
      "step": 2650
    },
    {
      "epoch": 6.44,
      "grad_norm": 2.922712802886963,
      "learning_rate": 1.6118989968868905e-05,
      "loss": 0.421,
      "step": 2660
    },
    {
      "epoch": 6.46,
      "grad_norm": 2.0401623249053955,
      "learning_rate": 1.5427187824282254e-05,
      "loss": 0.4283,
      "step": 2670
    },
    {
      "epoch": 6.49,
      "grad_norm": 0.9165148735046387,
      "learning_rate": 1.4735385679695607e-05,
      "loss": 0.4512,
      "step": 2680
    },
    {
      "epoch": 6.51,
      "grad_norm": 4.587483882904053,
      "learning_rate": 1.4043583535108959e-05,
      "loss": 0.4664,
      "step": 2690
    },
    {
      "epoch": 6.54,
      "grad_norm": 4.216481685638428,
      "learning_rate": 1.335178139052231e-05,
      "loss": 0.3407,
      "step": 2700
    },
    {
      "epoch": 6.54,
      "eval_accuracy": 0.87221269296741,
      "eval_loss": 0.5147121548652649,
      "eval_runtime": 6.1635,
      "eval_samples_per_second": 189.179,
      "eval_steps_per_second": 23.688,
      "step": 2700
    },
    {
      "epoch": 6.56,
      "grad_norm": 1.7551047801971436,
      "learning_rate": 1.2659979245935664e-05,
      "loss": 0.4725,
      "step": 2710
    },
    {
      "epoch": 6.59,
      "grad_norm": 4.851523399353027,
      "learning_rate": 1.1968177101349016e-05,
      "loss": 0.4639,
      "step": 2720
    },
    {
      "epoch": 6.61,
      "grad_norm": 6.040704727172852,
      "learning_rate": 1.1276374956762366e-05,
      "loss": 0.3146,
      "step": 2730
    },
    {
      "epoch": 6.63,
      "grad_norm": 1.6925532817840576,
      "learning_rate": 1.0584572812175717e-05,
      "loss": 0.3665,
      "step": 2740
    },
    {
      "epoch": 6.66,
      "grad_norm": 2.9491493701934814,
      "learning_rate": 9.89277066758907e-06,
      "loss": 0.467,
      "step": 2750
    },
    {
      "epoch": 6.68,
      "grad_norm": 2.1744699478149414,
      "learning_rate": 9.200968523002422e-06,
      "loss": 0.3542,
      "step": 2760
    },
    {
      "epoch": 6.71,
      "grad_norm": 3.170931577682495,
      "learning_rate": 8.509166378415774e-06,
      "loss": 0.5874,
      "step": 2770
    },
    {
      "epoch": 6.73,
      "grad_norm": 3.2446773052215576,
      "learning_rate": 7.817364233829124e-06,
      "loss": 0.3705,
      "step": 2780
    },
    {
      "epoch": 6.76,
      "grad_norm": 3.8055498600006104,
      "learning_rate": 7.125562089242477e-06,
      "loss": 0.3164,
      "step": 2790
    },
    {
      "epoch": 6.78,
      "grad_norm": 2.3979437351226807,
      "learning_rate": 6.4337599446558285e-06,
      "loss": 0.4043,
      "step": 2800
    },
    {
      "epoch": 6.78,
      "eval_accuracy": 0.869639794168096,
      "eval_loss": 0.5081329345703125,
      "eval_runtime": 6.6143,
      "eval_samples_per_second": 176.285,
      "eval_steps_per_second": 22.073,
      "step": 2800
    },
    {
      "epoch": 6.8,
      "grad_norm": 1.7395985126495361,
      "learning_rate": 5.74195780006918e-06,
      "loss": 0.3624,
      "step": 2810
    },
    {
      "epoch": 6.83,
      "grad_norm": 2.924905300140381,
      "learning_rate": 5.050155655482532e-06,
      "loss": 0.4046,
      "step": 2820
    },
    {
      "epoch": 6.85,
      "grad_norm": 11.709400177001953,
      "learning_rate": 4.358353510895884e-06,
      "loss": 0.4807,
      "step": 2830
    },
    {
      "epoch": 6.88,
      "grad_norm": 6.416582107543945,
      "learning_rate": 3.666551366309236e-06,
      "loss": 0.4782,
      "step": 2840
    },
    {
      "epoch": 6.9,
      "grad_norm": 6.1391448974609375,
      "learning_rate": 2.9747492217225875e-06,
      "loss": 0.4852,
      "step": 2850
    },
    {
      "epoch": 6.92,
      "grad_norm": 3.525520086288452,
      "learning_rate": 2.2829470771359392e-06,
      "loss": 0.4282,
      "step": 2860
    },
    {
      "epoch": 6.95,
      "grad_norm": 1.4197200536727905,
      "learning_rate": 1.591144932549291e-06,
      "loss": 0.4337,
      "step": 2870
    },
    {
      "epoch": 6.97,
      "grad_norm": 4.016748905181885,
      "learning_rate": 8.993427879626428e-07,
      "loss": 0.3915,
      "step": 2880
    },
    {
      "epoch": 7.0,
      "grad_norm": 2.1515309810638428,
      "learning_rate": 2.0754064337599448e-07,
      "loss": 0.4095,
      "step": 2890
    },
    {
      "epoch": 7.0,
      "step": 2891,
      "total_flos": 3.5833623598425784e+18,
      "train_loss": 0.7298227465279536,
      "train_runtime": 1041.6701,
      "train_samples_per_second": 44.372,
      "train_steps_per_second": 2.775
    }
  ],
  "logging_steps": 10,
  "max_steps": 2891,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 7,
  "save_steps": 100,
  "total_flos": 3.5833623598425784e+18,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}